Datasets:

shreyaspimpalgaonkar

/

ycombinator_s24

like 0

Ember Robotics

ember-robotics

https://www.emberrobotics.com

San Francisco, CA, USA

Ember Robotics is building camera diagnostics tools for robots and automotive. We worked on optimizing camera performance at Tesla Autopilot and now we're building software that takes mitigating camera failures from hours to minutes.

Camera diagnostics for robots

Industrials

Industrials -> Manufacturing and Robotics

S24

Active

Early

https://www.ycombinator.com/companies/ember-robotics

https://yc-oss.github.io/api/batches/s24/ember-robotics.json

Title: Ember Robotics: Camera diagnostics for robots | Y Combinator URL Source: Markdown Content: Ember Robotics: Camera diagnostics for robots | Y Combinator =============== []( "Y Combinator") [About]( [What Happens at YC?]( Interview Guide]( Blog]( [Companies]( [Startup Directory]( Companies]( Directory]( YC]( [Startup Jobs]( [All Jobs]( Engineering]( Operations]( Marketing]( Sales]( Internship Program 2024]( Job Guide]( Coaching]( Startup Jobs Blog]( [Find a Co-Founder]( [Library]( [SAFE]( [Resources]( [Startup School]( Investors]( News]( Open main menu Apply for **W2025** batch.[Apply]( "Apply for W2025 batch.") [Home]( Robotics Ember Robotics ============== Camera diagnostics for robots [S24]( Active [developer-tools]( Francisco]( * * * [Company]( [Jobs](  [ * * * ### Camera diagnostics for robots Ember Robotics is building camera diagnostics tools for robots and automotive. We worked on optimizing camera performance at Tesla Autopilot and now we're building software that takes mitigating camera failures from hours to minutes. Ember Robotics Founded:2024 Team Size:2 Location:San Francisco Group Partner:[Diana Hu](  []( "LinkedIn profile") ### Active Founders ### Shivani Mouleeswaran, Founder Co-founder and CEO @ Ember Robotics. Previously worked at Tesla Autopilot optimizing camera performance for autonomy. Shivani has spent 5 years working on camera systems for autonomous robots and co-founded Ember to improve to the pains she's seen around dealing with robot hardware and system reliability. Fun fact she once earned the title of Camera Queen! Shivani Mouleeswaran [Ember Robotics](  []( "LinkedIn profile") ### Ritika Shrivastava, Founder Ritika is a roboticist who loves helping robots see 🕶! She is a proud UC Berkeley alumna (B.S. EECS '21 & M.S. EECS '22) 🐻, where she researched and taught about computer vision and robotics. Formerly a computer vision engineer at Tesla Autopilot, Ritika is excited to build tools that help roboticists! In her free time, she enjoys playing tennis 🎾, building gizmos, and reading 📖! Ritika Shrivastava [Ember Robotics](  []( "LinkedIn profile") ### Company Launches [### Ember Robotics: Camera diagnostics for robots]( **tldr; Got camera problems? We’ve got solutions.** =================================================== [Ember]( builds software to monitor and resolve robot camera issues - so you can reduce vision failures from the source. Is robot camera reliability or degraded vision driving you nuts? Book a demo [here]( or reach out to us at [[email protected]](mailto:[email protected])! ### **The problem** We’ve spent years working on vision systems for autonomous robots and found similar issues everywhere: ☐ Robotics companies work on lean teams that have to do lots of things super fast - **hardware reliability is often secondary to core product features** ☐ **Camera failures cause large fires** because they float up the stack to critically damage perception, controls, and safety ☐ **Digging through layers of software / hardware to figure out what went wrong is painful** (ex. sometimes you spend hours goose-chasing a “perception issue” just to find a loose cable) ### **Our solution** **A camera diagnostics framework that ties in real-time data, logs, and reports to contextualize your robot’s vision health.** Support the full life-cycle of your robot with features like: ✅ Out-of-the-box tracking for common issues (disconnection, frame drops, time sync) 💬 Automated suggestions for error resolution ⏱️ Real-time alerts 🖌️ Image quality analysis tools to optimize computer vision 📈 Historical data analysis for predictive maintenance ### **Try it out: Plug-and-play camera testing** Bench-test cameras or diagnose issues in the field with one easy tool. **Fill a config file, launch our service, and get fast metrics on camera health.** [demo]( Currently we support Intel® RealSense™ D400 cameras and ROS - more cameras and pub-sub models coming soon! ### **Meet the team** Hi, we are [Shivani]( & [Ritika]( We built software at Tesla Autopilot to optimize camera performance for autonomy. We’ve covered the camera stack end-to-end (low-latency image capture, real-time camera monitoring, online camera calibration, multi-trip reconstruction, and SLAM) - we’re passionate about helping robots see! ### **Our ask** Is robot camera reliability or degraded vision driving you nuts? Book a demo [here]( or reach out to us at [[email protected]](mailto:[email protected])! If you know anyone working on robots, drones, or other multi-camera systems (surveillance, manufacturing, etc.) we would love a warm intro. And for those who are just generally robot-curious - always happy to chat! #### Company Photo Footer ------ [Y Combinator]( "Y Combinator") ### Programs * [YC Program]( * [Startup School]( * [Work at a Startup]( * [Co-Founder Matching]( ### Company * [YC Blog]( * [Contact]( * [Press]( * [People]( * [Careers]( * [Privacy Policy]( * [Notice at Collection]( * [Security]( * [Terms of Use]( ### Resources * [Startup Directory]( * [Startup Library]( * [Investors]( * [SAFE]( * [Hacker News]( * [Launch YC]( * [YC Deals]( ### Make something people want. [Apply]( [Twitter]( © 2024 Y Combinator

Kura

kura

https://www.usekura.com

San Francisco, CA, USA

Kura is a DevOps copilot that helps software teams manage their cloud infrastructure. Kura integrates directly with services like AWS to help respond to incidents, provision resources, and search over your entire cloud. Developers today spend 30% of their time managing cloud infra, and we cut that by 10x so teams can get back to shipping features quickly. Kura's founding team includes CEO Mark Dawson, an engineer and second time founder, and CTO Trevor Reed, former Google engineer and Mars Rover driver at NASA JPL.

The AI DevOps Copilot for Software Teams

B2B

B2B -> Engineering, Product and Design

S24

Active

Early

https://www.ycombinator.com/companies/kura

https://yc-oss.github.io/api/batches/s24/kura.json

Title: Kura: The AI DevOps Copilot for Software Teams | Y Combinator URL Source: Markdown Content: ### The AI DevOps Copilot for Software Teams Kura is a DevOps copilot that helps software teams manage their cloud infrastructure. Kura integrates directly with services like AWS to help respond to incidents, provision resources, and search over your entire cloud. Developers today spend 30% of their time managing cloud infra, and we cut that by 10x so teams can get back to shipping features quickly. Kura's founding team includes CEO Mark Dawson, an engineer and second time founder, and CTO Trevor Reed, former Google engineer and Mars Rover driver at NASA JPL. Kura Founded:2024 Team Size:2 Location:San Francisco ### Active Founders ### Mark Dawson, Founder Software Engineer. Second time founder. Currently focused on AI + Cloud infrastructure. ### Trevor Reed, Founder Co-founder at Kura, building the AI DevOps Copilot for software teams. Former Mars Rover driver with over 1km of drive distance on Mars, and former software engineer at Google processing worldwide YouTube data. ### Company Launches [### Kura | The AI DevOps Copilot]( **TL;DR:** [Kura]( integrates directly with services like AWS, helping software teams respond to incidents, provision infrastructure, and optimize complex cloud systems. Hello YC! We’re [Trevor]( and [Mark]( founders of Kura. After building distributed systems for a combined ~15 years at places like Google and NASA, we know the challenges and triumphs of cloud-based products. We started Kura with the core belief that recent AI advancements can unlock the potential of software teams building on the cloud. **What’s the problem?** ----------------------- As web apps become increasingly core to the world, maintaining cloud infrastructure requires increasingly more time and resources. * Distributed systems are getting more complex than ever * Managed solutions trade off ease against control and cost * Cloud infra engineers are few and far between, leaving much of the work to teams who may not have extensive cloud experience **What can Kura do?** --------------------- Kura is a chatbot and workspace that integrates directly with a live index of your cloud infra from services like AWS, GCP, and Azure. **Kura can answer questions about your cloud system**, providing links and live information for queries like: * “Is my main API cluster healthy?” * “Which container version is running in prod?” * “Can you help me deploy a new container via CDK?” **Kura can run pre-packaged and user-defined “runbooks”** to proactively surface issues via notifications. For example: * Scheduled checks for over/under-utilized computing resources * Discovering and surfacing cost spikes and trends * Searching for common security vulnerabilities With Kura, software teams can build faster and more confidently on the cloud, all while maintaining full control of their critical infrastructure. **Our ask:** ------------ * If your team is building on the cloud and would like to learn more or sign up as a pilot customer, please reach out to [[email protected]](mailto:[email protected]). * If you know software teams or developers running into roadblocks due to cloud complexity, please send this post their way - we’d love to chat and learn more! * Follow us at [@kura\_labs]( to follow along on the journey! #### Company Photo

Spaceium Inc

spaceium-inc

https://spaceium.com

Long Beach, CA, USA

Spaceium builds fully automated space stations to refuel and repair spacecraft. We have secured $86.1 million in binding commercial contracts and have an additional $230 million in the pipeline. Additionally, we have 1 billion dollars in letters of intent. We have successfully tested our hardware, which will launch to space next year. Our fully automated space stations will efficiently store and transfer both cryogenic (extremely cold) and non-cryogenic fuels with zero loss during storage and transfer. And they can transfer fuel seamlessly to any spacecraft design using our proprietary modular robotic arm. Our customers include launch vehicles, orbital transfer vehicles, moon landers, and spacecraft that benefit from increased payload capacity and extended travel capabilities. Our long-term vision is to build service hubs along the space superhighway to connect Earth to Moon and Mars and help humanity to become multi multi-planetary species

Fully automated space stations for refueling and repair of spacecraft

Industrials

Industrials -> Aviation and Space

S24

Active

Early

https://www.ycombinator.com/companies/spaceium-inc

https://yc-oss.github.io/api/batches/s24/spaceium-inc.json

Title: Spaceium Inc: Fully automated space stations for refueling and repair of spacecraft | Y Combinator URL Source: Markdown Content: **TL;DR:** **Spaceium is building unmanned servicing stations in space to improve mission efficiency. Our stations will offer proprietary cryogenic fuel storage and transfer technology, enabling in-space refueling and providing robotic repair solutions, a capability that sets us apart. We aim to launch servicing stations in Low Earth Orbit in 2025 and expand to Geostationary Orbit, the Moon, and Mars by 2030.** "Helping space explorers go "where no man has gone before" - Captain James T. Kirk :) **\_ \_** Hi, everyone. We are [Ashi Dissanayake]( and [Reza Fetanat]( the team behind [Spaceium]( **Here's a brief introduction:** **Ashi** is an Aerospace Engineer specializing in **Orbital Mechanics and rocket propulsion.** Ashi has worked on real-time simulation models for uncrewed return missions to asteroids and on **NASA-related aerodynamics projects**. **Reza** is also a space enthusiast and a Mechanical Engineer specializing in **Robotics, structures, and Cryogenics**. Reza has worked on revitalizing the **LR-101 engine, which was used on Atlas, Thor, and Delta launch vehicles.** We first met while leading the University of Ottawa's rocketry team, where we built high-powered rockets to send scientific payloads to different altitudes. In our spare time, we enjoyed working together to build rockets in the backyard, sometimes leading to interesting experiences with backyard fires. 🔥 Throughout our careers, we've noticed the constraints on space missions due to the lack of ability to refuel or repair spacecraft in orbit. The current solutions are costly, and a centralized location where spacecraft can refuel and repair is a massive necessity. We are on a mission to build the missing piece of space infrastructure that connects the space superhighway, Just like how service stations were built on Earth back in the day! **The problem** --------------- All spacecraft and space missions face a standard set of challenges. Fuel is a limited and difficult resource to transport. In the event of a significant issue, spacecraft cannot be repaired in space. These challenges have significantly hindered efficiency in current space missions and have been a substantial obstacle in humanity's quest to explore space further. Unlike a road trip where refueling is common and convenient, spacecraft face a unique challenge. Spacecraft cannot refuel along the way. Instead, they must carry all the fuel required for the mission from Earth. **An overwhelming 85% - 95% of the launch weight is fuel on a typical launch vehicle, with 80-90% burned off to reach orbit.** 🚀 As a result, spacecraft that transport loads in space (space tugs) can only perform, **on average, 25% as many missions as they're technically capable of due to the absence of refueling in space.** 🛰️ Furthermore, even if a spacecraft wanted to carry additional fuel, it is not easy. Current spacecraft require cryogenic fuels such as liquid oxygen, methane, and liquid hydrogen, which must be stored and transferred at extremely low temperatures of -285°F. 🥶 Environmental factors in space, such as heat and radiation, increase the temperature of these fuels, causing them to boil off (vaporize). In addition, **there is no current solution for spacecraft repair**. This means that today's spacecraft must operate flawlessly to maximize efficiency, which is only sometimes the case. For instance, Eutelsat Communications' satellites experienced power loss mid-mission with no option for repair, resulting in a **$30 million** revenue loss. **With Spaceium's multifunction robotic arm, the communication satellite could have been fixed.** 🛠️ **Solution** ------------ _Automated servicing stations in space 🌎🚀🛰️_ Our primary solution aims to improve space missions' efficiency by addressing refueling challenges. **The developed cryogenic fuel storage and transfer technology is designed to store fuel without any boil-off for multiple years in space.** As a result, spacecraft can now refuel in space, carry more payload, and travel longer distances. Two fundamental principles guide our proprietary Zero Boil Off (ZBO) technology. * Firstly, we prevent heat from entering the system through a unique tank design, insulation, and coating system. * Secondly, our innovative multistage closed-loop cooling system evenly removes the heat from the cryogenic fuel and transfers it to radiators, ensuring zero boil-off. Our cryogenic storage system is not only efficient but also highly scalable. Our current cooling system is **65% more efficient** than any other cooling system, and our scalable storage system can store up to 30 metric tons of fuel, making it adaptable to a wide range of space missions and requirements. After successfully addressing the challenge of transporting large amounts of space fuel, we turned our attention to developing technology for delivering this fuel to spacecraft in space. This is achieved through the use of a **multifunction robotic arm**. This advanced arm can securely dock with incoming spacecraft and initiate the fuel transfer. **The fluid system built within the arm allows the transfer of different fuel types through the robotic arm**. The built-in thermal system ensures that fuel is transferred to the customer's spacecraft at the correct temperature without any boil-off issues associated with cryogenic fuels in space. Additionally, the same robotic arm is versatile enough to handle non-cryogenic fuels as well, providing a comprehensive solution for refueling spacecraft in space. In addition to safe and efficient fuel transfer, the robotic arm is equipped with an interchangeable end effector designed for in-space repairs. In-space repairs are achieved through a fully automated robotic arm that can retrieve tools from a stored "toolbox" within the system. This innovative feature ensures that even in the event of a significant issue, spacecraft can be repaired in space, minimizing the risk of mission failure and maximizing mission efficiency. Our vision is to transform the space industry. Starting in Low Earth Orbit (LEO), our servicing infrastructure will progressively expand to encompass Geostationary Earth Orbit (GEO), the moon, and eventually Mars. We are on a mission to empower space missions to go further, explore more, and achieve what was once deemed impossible. **Who can benefit from our service stations in space** * **Launch vehicles** * **Orbital transfer vehicles** * **Space tugs** * **Moon landers** * **Satellites** **Ask** ------- * **Feedback, comments, or questions.** Reach out to us to learn how we can help with your missions. * **Warm Intros.** If you know or work with any launch providers, space companies, or satellite providers who could benefit from our service, please introduce us to them. 🛰 * **Follow.** Follow us on [LinkedIn]( [X]( Cool news coming soon * **Share.** Please share this post with your network so we can reach more space companies looking for refueling, repairs, and other services for their future missions. 👾 Email us [HERE](mailto:[email protected]). ---------------------------------------------- Thank you!

Fuse

fuse-2

http://www.fuseinsight.com

London, England, United Kingdom; Remote

AI-powered analysis of healthcare contracts and payments to help providers detect underpayments, compare their negotiated rates, and identify opportunities to boost revenue from insurers. If you know senior finance execs in health systems, partners in healthcare PE, or mid to large sized healthcare practice owners, please do connect us. We won’t waste their time: +1 628 468 3232.

Boosting contracted revenue for US healthcare providers

Healthcare

Healthcare

S24

Active

Early

https://www.ycombinator.com/companies/fuse-2

https://yc-oss.github.io/api/batches/s24/fuse-2.json

Title: Fuse: Boosting contracted revenue for US healthcare providers | Y Combinator URL Source: Markdown Content: ### Boosting contracted revenue for US healthcare providers AI-powered analysis of healthcare contracts and payments to help providers detect underpayments, compare their negotiated rates, and identify opportunities to boost revenue from insurers. If you know senior finance execs in health systems, partners in healthcare PE, or mid to large sized healthcare practice owners, please do connect us. We won’t waste their time: +1 628 468 3232. Fuse Founded:2024 Team Size:2 Location:London, United Kingdom ### Active Founders ### Sho Sugihara, Founder Sho is a repeat founder. He scaled his last business to $6m ARR within twelve months from launch. Before that, he worked on data analytics projects at McKinsey for some of the largest healthcare systems in the world, including the NHS and several of the top ten private healthcare insurers. He studied Social Sciences at Cambridge and Global Health at the Karolinska Institutet. ### Oscar Wilsby, Founder Oscar has built large-scale payments and credit data analytics models that outperformed incumbents like Experian by a significant margin. These models helped power the first startup he co-founded and scaled to a multi-million $ run rate. Before tech, Oscar was a post-doc researcher and completed an MEng and PhD in Machine Learning and Computational Physics at Cambridge. He’s worked in research divisions for companies like McLaren F1 and Dyson, applying machine learning to complex data problems. ### Company Launches [### Fuse - Underpayment detection for healthcare providers]( **TL;DR:** [Fuse]( offers AI-powered analysis of healthcare contracts and payments to help providers detect underpayments and collect the revenue they’re owed from insurers. **❌ Problem** ------------- Our “aha” moment came when healthcare provider execs repeatedly told us they had no idea if their revenue was accurate. One even described their insurance claims process as a “black hole”. Several factors contribute to this issue: 1. Growing usage of AI. Insurers are using sophisticated models to analyze claims and find ways to subtly underpay. 2. Too many contracts: An average hospital has over 50 contracts with insurers with varying reimbursement rates for the same procedure, making payment accuracy verification costly and near impossible. 3. M&A wave: Mergers are forcing providers to consolidate contracts and revenue tracking systems, adding further complexity. As a result, providers spend months analyzing revenue and chasing payments, forfeiting up to 11% of topline revenue if they cannot afford the overhead to do so. With provider operating margins already negative since COVID-19, the situation creates severe margin compression and cash flow risk. **Solution** ------------ We use LLMs to extract data from the stacks of contracts that dictate provider payment terms from insurers. We reconcile the data against remittances to identify underpayments and auto-generate appeals. Our mission is to level the playing field for providers with AI-powered contract analytics. Beyond this, we want to aid the transition towards Value-Based Care by powering sophisticated contract modeling as payment models become even more complex. **Team** -------- Our team specializes in complex financial data analytics. We previously co-founded a business that grew from $0 to $6M ARR in 12 months, developing large-scale data models for payment and credit analytics that outperformed major players like Experian. [Oscar]( (CTO) was an academic who completed a PhD and postdoc in Machine Learning at Cambridge. He’s worked in research divisions for companies like McLaren F1, and brings expertise in building top performing data models in highly regulated industries. [Sho]( (CEO) met Oscar during their first week of undergrad at Cambridge. He’s a repeat entrepreneur and closed a $100k contract within a year for his first company, and has worked on data analytics projects for some of the largest providers and insurers in the world at McKinsey. **🙏 Our Ask** -------------- **If you know senior finance execs in healthcare systems or partners at healthcare PE firms please do connect us!** [**[email protected]**](

Modern Realty

modern-realty

https://modernrealty.io

San Francisco, CA, USA; Remote

Modern Realty is an AI real estate agent that helps you buy homes without relying on a traditional Realtor. Regular buyer's Realtors aren't incentive-aligned with the buyers, and therefore often make the process information-opaque. We offer a self service alternative.

The AI real estate agent for home buying

Real Estate and Construction

Real Estate and Construction -> Housing and Real Estate

S24

Active

Early

https://www.ycombinator.com/companies/modern-realty

https://yc-oss.github.io/api/batches/s24/modern-realty.json

Title: Modern Realty: AI Realtor for Home Buyers | Y Combinator URL Source: Markdown Content: ### AI Realtor for Home Buyers Interpret real estate data and gain an edge on other home buyers. ### Latest News Modern Realty Founded:2024 Team Size:2 Location:San Francisco ### Active Founders ### Raffi Isanians, Founder Modern Realty is the new way of buying a house. Our AI real estate agents make buying a house easier, faster, and far more enjoyable. Have any issues? Don’t know how much to offer? Get connected to one of our experienced agents or attorneys who are on call and ready to help. ### Raymond Xu, Founder CTO at Modern Realty. Ex-Google AI engineer, specializing in RAG (Google Assistant, Google Document AI). ### Company Launches [### Modern Realty - The AI real estate agent]( **What is Modern Realty?** 🏠 [Modern Realty]( is an AI real estate buying experience that helps you buy homes without relying on a traditional realtor. Historically, sellers pay a 3% commission to the buyer's agent. However, due to a recent lawsuit against the National Association of Realtors, buyers will soon have to pay their agents themselves, which can be very expensive (i.e., $30,000 for a $1 million house). With tools like Zillow, finding the right house at the right price is easier than ever. However, buying a house still requires an agent, even though buyers are doing most of the work themselves. Modern Realty changes this by offering an AI self-service solution. Modern Realty guides you through buying a home step-by-step. You can start with or without a specific home in mind. On the website, you can schedule in-person showings and use tools to help you determine how much to offer. When you're ready, you can create and submit an offer through our platform. Using the software is free; we only charge if you successfully buy a house. Plus, if you need more help, our team of lawyers and real estate agents are always available to guide you through the process. **How it works 🧠** At Modern Realty, we provide everything you need to make the best, most informed home-buying decision. Your agent guides you through each step of the home-buying process, including: 🔑 Disclosure summary: Understand your homeowner’s association (pet ownership and rental policy), land zones, tax, liens/debts, inspections, and repairs 📝 Make Offers Online: Generate your offers using our website. An attorney will review all offers, at no additional cost, before they’re sent to the seller. 📅 Schedule Private Showings: Our AI tools automatically schedule showings for you directly with the sellers without having to talk to an agent. 📈 Competitive Market Analysis: Make the best bidding choice by seeing all the available data on the surrounding nearby sales (for example, median % over list price) **About Us** 🌟 * Why we are working on this * We both purchased homes somewhat recently and thought that the process could be dramatically improved by letting people get their hands dirty in the process themselves * Founders * [Raymond Xu]( (left): Ex-Google Machine Learning Software Engineer, previously created a RAG product in 2020 * [Raffi Isanians]( (right): Attorney, Realtor, has been an attorney for $10b+ of asset transactions **Get started 🌱** Visit [modernrealty.io]( or schedule a call with us to get started [here]( #### YC Sign Photo #### Company Photo

Void

void

https://voideditor.com

San Francisco, CA, USA

Void is an open source AI code editor. We let developers use Cursor-style AI features without sacrificing privacy.

The open source Cursor alternative

B2B

B2B -> Engineering, Product and Design

S24

Active

Early

https://www.ycombinator.com/companies/void

https://yc-oss.github.io/api/batches/s24/void.json

Title: Void: The open source Cursor alternative | Y Combinator URL Source: Markdown Content: ### The open source Cursor alternative Void is an open source AI code editor. We let developers use Cursor-style AI features, without having to send their code to Cursor’s API. Void Founded:2024 Team Size:2 Location:San Francisco ### Active Founders ### Andrew Pareles, Founder Andrew is the CEO of Void. Before starting Void, he bootstrapped an edtech company, did quantum computing research @ Johns Hopkins APL, and studied computer science @ Cornell. ### Mathew Pareles, Founder Mathew is the CTO of Void, the open-source AI code editor. He studied physics @ Cornell and has been hacking with machine learning models like GPT-3 before they were popular. Before starting Void, he co-founded DeriveIt, an edtech startup that grew to thousands of users. ### Company Launches [### Void: The open source Cursor alternative]( **_TL;DR:_** [Void]( is the open source Cursor alternative. It offers the AI features of an editor like Cursor, while allowing you to host your own AI model locally and keep your data private. Check us out on [GitHub]( About us -------- Hi everyone, we’re [Mat]( and [Andrew]( and we’re the team behind Void. Mat’s been prompting Transformers since before GPT3 became mainstream, and Andrew did quantum computing research at JHU APL. We’re best friends who have been programming together since we were 8. _left to right: Mat (CTO), Andrew (CEO)_ ❌ The problem: the leading AI IDE is closed source -------------------------------------------------- Cursor is a closed-source AI editor, which means you need to send your private data through Cursor’s backend every time you use it. This leads to obvious privacy concerns. It’s also expensive for developers, and it means one individual has full control over a powerful AI model. We think the future of AI should be open. You should be able to use whatever model you like and not be locked into using a privatized commercial one that’s built to collect your data. ✅ The solution: open source --------------------------- The solution: create an open-source Cursor alternative. This way, you can choose to self-host or communicate directly with any model you like, anywhere you like. There are lots of benefits to open-sourcing besides privacy, too: on Void, you can access community-made AI features beyond the ones that Cursor offers, and gain full access to all your prompts. Key features ------------ We’re building a lot. Here are the features we’re most excited about: * **File system awareness:** Void can answer questions about your entire codebase. It builds an index of all your files and uses it automatically, so you don’t manually select relevant files. * **View and edit underlying prompts:** Void lets you open up the hood and view/edit every single prompt in your chat history. This is really important if you want to understand how the LLM got its answer, and it’s lacking in most closed-source tools. * **Fast edits across 1000s of lines:** Even if your file is thousands of lines long, Void applies changes almost instantly. We prompt the AI to make changes the way a human would instead of rewriting the whole file. * **Full privacy / data control:** All of Void’s prompt-building is done on your own computer. This means you can self-host your own model and have your data never leave your network. Or, you can connect directly to Claude, GPT, or Gemini, and not worry about sending your company data through a middle layer of communication. We think the future of AI is open, and we’re excited about building it. **🙏 Asks** ----------- * Check out the [**Void editor**]( and star us on GitHub! * If you’re interested in helping us build, join our [Discord]( or shoot us an email. We rely on contributions, even small ones, and we’ll help you get up and running as a contributor! * Please share this post! Help spread the word that we’re building an open source Cursor alternative. #### YC Sign Photo

Spherecast

spherecast

https://www.spherecast.ai/

Spherecast is an AI-powered inventory management software for multi-channel ecom brands, supporting supply chain managers with demand planning, replenishment, and inventory optimization. With Spherecast, brands know: 1. When to re-order 2. How much of their products 3. To which warehouse 4. At what cost Spherecast helps avoid stockouts and overstock, improve inventory visibility, and save time.

AI-powered Inventory Management for Ecommerce

B2B

B2B -> Supply Chain and Logistics

S24

Active

Early

https://www.ycombinator.com/companies/spherecast

https://yc-oss.github.io/api/batches/s24/spherecast.json

Title: Spherecast: AI Supply Chain Manager for Ecommerce | Y Combinator URL Source: Markdown Content: ### AI Supply Chain Manager for Ecommerce Spherecast is an AI-powered inventory management software for omni-channel e-commerce brands, supporting supply chain managers with demand planning, inventory optimization, and supplier communication. Spherecast tells brands when to reorder, how much, which warehouse to stock, and at what cost. It then autonomously executes by communicating with suppliers, manufacturers, and warehouses. Spherecast helps avoid stockouts and overstock, improve inventory visibility, and save time. Spherecast Founded:2023 Team Size:5 Location: ### Active Founders ### Leon Hergert, Founder Leon is the Co-founder and CEO of Spherecast (S24). He has worked in supply chain and logistics planning twice (aviation and e-commerce) and is now doubling down on the problem space with Spherecast. He has a background in Information Systems from the Technical University of Munich and Technology Management from CDTM. Besides that, he loves playing basketball. ### Paul Dietrich, Founder Paul always wanted to work on his own projects. After building some small apps as side projects, he co-founded a software agency alongside his studies and scaled it to multiple 100k in total revenue. This helped him sharpen his skill set in full-stack development, design, and entrepreneurship. Now at Spherecast, he's responsible for building a product our customers love. ### Pascal Schindler, Founder Growing up with an affinity for building, he developed a passion for the intersection of operations, machine learning, and optimization. From building robotics automation at Deadalus (YC W20) to completing his academic research with a focus on non-linear optimization at British Columbia, these experiences led him to pursue his passion as the CTO of Spherecast ### Company Launches [### Spherecast – Inventory copilot for multi-channel brands]( [Spherecast]( is an inventory copilot supporting multi-channel brands with their inventory decisions. **❌ The problem** As soon as brands go multi-channel, e.g. enter a marketplace or are placed at offline retailers, they face new complexities when it comes to planning demand. Besides that, scaling into multiple channels often entails adding new warehouse locations through the supply chain, be it by sourcing raw materials by yourself or adding FBA as an outbound fulfiller. This is accompanied by rising logistics costs and inventory constraints (such as the low inventory level fee of Amazon or expiration date requirements for sales channels). The combination of these aspects leaves supply chain managers with a sheer amount of different constraints to handle – often in very large spreadsheets. This leads to inventory distortion costs – the cost of stockouts and overstock – which is a $1.8 trillion USD problem for retailers worldwide. **✅ The solution** Spherecast combines demand planning, replenishment and inventory optimization in one platform. We connect to shop systems and ERPs automatically. With the platform, brands can automatically get: A library of machine learning forecasts per product and channel to optimize forecast accuracy combined with a collaborative environment to plan with multiple stakeholders. Replenishment optimization, considering * Inbound orders and shipments to any warehouse * Outbound (future) send-ins (e.g. to another warehouse) * Linear constraints such as dynamic safety stock, target coverage and lead times * Non-linear constraints such as MOQ are integrated With that, we believe inventory management should transform into inventory optimization for everyone. **🚀 Our mission** We want to create a world where stockouts do not exist anymore. Imagine any product at the right place, at any time — enabled by Spherecast. **🙏 Our ask** If you happen to know supply chain managers at multi-channel e-commerce brands – we’d love to chat! * Book a demo [**here**]( or contact us via [**[email protected]**](mailto:[email protected]) * Tell your network about inventory optimization for e-commerce and share this post * Check out our website [**spherecast.ai**]( _From left to right: Felix Grote (Founding Business Developer), Pascal Schindler (Co-Founder & CTO), Paul Dietrich (Co-Founder & CPO), Leon Voland (Founding Engineer), Leon Hergert (Co-Founder & CEO)_

Callback

callback

https://getcallback.ai

San Francisco, CA, USA

Callback is AI-native business process outsourcing. Companies use us to stay lean and focused on their core business, while we orchestrate and execute their critical tasks.

AI-native business process outsourcing

B2B

B2B -> Operations

S24

Active

Early

https://www.ycombinator.com/companies/callback

https://yc-oss.github.io/api/batches/s24/callback.json

Title: Callback: AI-native business process outsourcing | Y Combinator URL Source: Markdown Content: ### AI-native business process outsourcing Callback is AI-native business process outsourcing. We help our customers stay focused on their core business while we use AI to automate tasks like invoice processing, PDF report parsing, and data labeling. Callback Founded:2024 Team Size:2 Location:San Francisco ### Active Founders ### Alan Xie, Co-Founder & CEO Alan is the co-founder and CEO of Callback. He was previously a product manager at Microsoft and AWS and the co-founder of an entertainment analytics startup. He studied CS at Harvard and Columbia and has an MBA from Harvard Business School. ### Paulo Bautista, Founder (COO) Paulo is the co-founder & COO of Callback. He previously managed thousands of delivery drivers and call center agents in the Philippines holding roles like Director of Operations and Head of BizOps. Paulo holds an MBA from Harvard Business School. ### Company Launches [### Callback - AI business process outsourcing]( **TLDR**: [**Callback**]( **is the first AI-native BPO built using a modern tech stack.** Our software platform helps our customers stay focused on their core business by automating their ops processes with humans in the loop. We’ve handled tasks such as image labeling, invoice extraction, and PDF report parsing — with greater accuracy and speed than traditional BPOs or off-the-shelf software solutions. **Our team** [Alan Xie]( was a senior product manager at AWS and Microsoft, where he built enterprise productivity software used by hundreds of millions of people. He holds computer science degrees from Harvard and Columbia, as well as an MBA from Harvard Business School. [Paulo Bautista]( was the first employee and Head of Business Operations at Ninja Van in the Philippines, where he managed a team of several hundred delivery couriers and grew the business to 8 figures in annual revenue. Before YC, he ran an 800-person call center as the Director of Operations and Strategy for a late-stage restaurant tech startup. Paulo studied Business Economics at the University of the Philippines and holds an MBA from Harvard Business School. **Why are we building Callback?** As companies grow, they naturally accumulate process overhead involving manual, repetitive operations tasks. Outsourcing these processes to BPOs requires months of coordination, training, and QA — friction that only the largest multinational corporations can tolerate. We empower **companies of all sizes** to offload their manual processes onto our platform. Callback is a software layer for business operations that makes your processes more **transparent**, **efficient**, and **cost-effective**. Onboarding onto our platform is straightforward: 1. Customers define a specific process workflow as a **Blueprint** 2. They execute **Jobs** based on their Blueprints 3. Jobs rely on pre-defined **automation** as well as **human QA** Do you have any ops processes you wish you could delegate? Unstructured data that you need to turn into a specific structured format? Email us at [[email protected]](mailto:[email protected]) or book time with us [here](

Random Labs

random-labs

We are building a fully local opensource software engineer. Software thrives on opensource tools and it's only right that this technology be built for all to use.

Random labs is building fully open source software agents

B2B

B2B -> Engineering, Product and Design

S24

Active

Early

https://www.ycombinator.com/companies/random-labs

https://yc-oss.github.io/api/batches/s24/random-labs.json

Title: Random Labs: Random labs is building fully open source software agents | Y Combinator URL Source: Markdown Content: ### Random labs is building fully open source software agents We are building a fully local opensource software engineer. Software thrives on opensource tools and it's only right that this technology be built for all to use. Random Labs Founded:2024 Team Size: Location: ### Active Founders ### Kiran (Random Labs), Founder Building opensource code generation tools ### Mihir (Random Labs), Founder CTO@Random Labs ### Company Launches [### Random Labs: An open source software agent that lives in your codebase]( **_Software is like playdough. It morphs into the creators desires._** **_Or at least it should._** This is what fundamentally drives us as a company. We believe that software development should stay in the hands of people. We believe that there is near infinite software the world needs and that there are not enough developers to build it. This leads us to the problem we are solving. _We are making it possible for the next 20 million engineers to come online._ We believe AI tools can be used to create jobs, not replace them. And we believe that open-source tools will continue to be the path forward for developers. Most software agent companies have deeply optimized for swe-bench. However, this is not a representative task. We are building tooling we actually use. It’s written parts of itself already and we are extremely excited to share it with you. **How are we building this?** Our approach is to start with the IDE. Developers are most comfortable in the ide, so that’s where we are starting. We are integrating our agent directly, and it is able to make changes across massive codebases (below is it writing a feature in the cal\[.\]com codebase which is 1,155,075 lines long at the time of writing). So, we are making an agent that is capable of _living in your codebase_ with almost perfect context. It’s not just autocomplete. It’s not just chat with your codebase. It’s something beyond a software workflow. Our tooling makes the code feel fluid. _And thats special._ Here it is editing the codebase: Here is the resulting component: **Our ask!** 1. If you’re interested in using our tool once it launches, please join the waitlist here: [ 2. If you’re interested in helping us build this, PLEASE reach out at [[email protected]](mailto:[email protected]) ### Hear from the founders #### What is the core problem you are solving? Why is this a big problem? What made you decide to work on it? We are making software engineering agents and code generation tools for every person. As more and more people realize they can code now, we will enable them to create. It’s a huge problem because there is infinite software to build if we want to improve life on earth, yet, theres nowhere near enough talent to build it. This tech will make it possible for anyone to create. #### What is your long-term vision? If you truly succeed, what will be different about the world? If we succeed, everyone will be able to code, and the number of software jobs available with grow by an insanely large number (10-20 million more devs).

Modus

modus

https://himodus.com

Remote

Modus helps companies hire and retain only the people they need instead of becoming bloated messes. Modus connects your HRIS, ATS, and analyzes extensive data across your organization so leaders can detect, inspect, and correct their workforce plans - without relying on spreadsheets or complex systems. Lin experienced these challenges firsthand as a VP BizOps at HashiCorp, scaling from 100 employees to IPO, while Amar faced similar issues scaling 40+ person teams as a fractional CTO. They’re motivated by the belief that companies should have the least number of employees possible.

Headcount management to save costs as you scale

B2B

B2B

S24

Active

Early

https://www.ycombinator.com/companies/modus

https://yc-oss.github.io/api/batches/s24/modus.json

Title: Modus: Unlock revenue per employee with AI-powered headcount management | Y Combinator URL Source: Markdown Content: ### Unlock revenue per employee with AI-powered headcount management Modus helps companies hire and retain only the people they need instead of becoming bloated messes. Modus connects your HRIS, ATS, and analyzes extensive data across your organization so leaders can detect, inspect, and correct their workforce plans - without relying on spreadsheets or complex systems. Lin experienced these challenges firsthand as a VP BizOps at HashiCorp, scaling from 100 employees to IPO, while Chris and Amar faced similar issues scaling 40+ person teams as a fractional CTO. They’re motivated by the belief that companies should have the least number of employees possible. Modus Founded:2024 Team Size:3 Location: ### Active Founders ### Lin Ling, Founder Lin is Co-founder and CEO of Modus. Before Modus, she was a VP BizOps at HashiCorp and helped scale from 100 to 2500 employees to IPO (rang the bell in NY). Prior to that, she implemented ERP systems at five Fortune 500 companies as a Deloitte Consultant. The clunky nature of these tools led her to appreciate the flexibility of spreadsheets but also the craft in modern software. Lin received her Bachelor's in Econ, Comp Sci, and Math at NYU Stern. ### Amar Kota, Founder Amar is the co-founder of Modus. Before Modus, he co-founded a SaaS and developed their AI to identify key remarks in large financial disclosure statements. After his child was diagnosed with ASD, Amar focused on data analysis that led him to Modus where he is building a solution that plagued him while managing his workforce. Amar received his MS from Carnegie Mellon University and is a tenured teacher for the University of California. ### Company Launches [### Modus: Headcount management to save costs as you scale]( [Modus]( is headcount management that helps companies never over-hire as they scale. Modus brings Finance, HR, and Hiring Managers to detect, inspect, and correct their workforce plan in one unified workflow. Today, much time is wasted reconciling headcount across departments and exporting across disparate systems. Incomplete and outdated data adds to the pain. Companies overhire, then fire. Cycle perpetuates. Bloat and inefficiencies persist. [Lin]( experienced these pains directly as VP of BizOps at HashiCorp, scaling from 100 to 2500 through IPO. Modus was founded on the belief that companies should have the fewest employees possible. 🤯 **The Problem: Headcount planning and reporting remains manual and painful** ------------------------------------------------------------------------------- Overhiring in tech has led to over 200,000 layoffs, costing companies billions. Workforce costs account for 70-80% of a company’s expenses, yet companies still make workforce decisions in error-prone, manual spreadsheets with incomplete data. Re-orgs, department changes add to the pain. Reconciliation is messy and time-consuming. Accuracy suffers. The cycle perpetuates. As Taylor Swift might say, _“I’ve seen this film before, and I didn’t like the ending.”_ 🪄 **Our Solution: AI-powered headcount management** ---------------------------------------------------- Modus integrates with your existing HRIS and ATS systems, using AI to help you detect anomalies, inspect, and correct your workforce plan. With Modus: ❌ Overhiring ❌ Layoffs ❌ Incomplete data ❌ Useless meetings ✅ Hire and retain only the people you need **✅** Increase visibility **✅** Collaborate off a single source of truth **✅** Save costs as you scale The more Modus learns, the more it streamlines and automates your headcount management processes at scale. 🙋🏻‍♀️**The Ask** ------------------ **Share this post!** Know anyone in your network managing headcount, bogged down by spreadsheets? A VP in Finance, HR, or Operations? Please share this post (or the blurb below). * _Modus is headcount management to save costs as you scale. With the help of AI, our software analyzes extensive data across your organization, helping you remove the bloat and correct your workforce plan – all in one unified workflow. Founders are Lin (ex-HashiCorp), Amar (ex-Finfare), and Chris (ex-OpenGov, 2x founder) with combined experience scaling companies from 0 to 1, to $8B+, and through IPO. Contact _[[email protected]_](mailto:[email protected])_ for a demo and trial._ ### Hear from the founders #### What is your long-term vision? If you truly succeed, what will be different about the world? Our ‘level 5’ vision is to automate workforce management at scale and to build efficient companies with the least number of employees possible.

pap!

pap

https://joinpap.com/

Berkeley, CA, USA

Consumers miss out on $50B+ every year because they're busy and unaware, and that money is sitting in their email inbox. So, we automated reading email receipts and claiming the money they're entitled to, offering a service to which you can sign up once, do (literally) nothing, and earn $. Where is this money coming from? - Retailers: When the price of clothes or furniture you've already bought drops, you're entitled to a refund for the difference - Airlines: When a flight you've booked gets delayed or canceled, you're entitled to a cash compensation. - Enterprises: When a class-action lawsuit is filed against a company you engage with, you're entitled to money. - And many more, that even the government holds some of your unclaimed funds. Today, the dream of one service that automatically claims the money you deserve when you shop, fly,... and live is (finally) possible. We started with automating refunds in the shopping industry and have secured $5,000 in refunds to our users, growing 136% MoM. With AI, we've been able to repurpose that same automation to tackle refunds in the flight industry and the class-action lawsuit landscape. Why us? (1) We're obsessed with reclaiming what's rightfully yours: Kamal and Anthony grew up in Lebanon, where people's money was unfairly taken away. (2) Weijie was the lead infra engineer for Wish's recommendation system, the feature that fueled their IPO. Our combined experience includes Apple, Addepar, Bolt, Wish, and StackAdapt. And we're all working towards one goal: move those billions of dollars where they belong.

The world owes you money. We get it for you.

Consumer

Consumer

S24

Active

Early

https://www.ycombinator.com/companies/pap

https://yc-oss.github.io/api/batches/s24/pap.json

Title: pap!: The world owes you money. We get it for you. | Y Combinator URL Source: Markdown Content: ### The world owes you money. We get it for you. Consumers miss out on $70B+ every year because they're busy and unaware, and that money is sitting in their email inbox. So, we automated reading email receipts and claiming the money they're entitled to, offering a service to which you can sign up once, do (literally) nothing, and earn $. Where is this money coming from? - Retailers: When the price of clothes or furniture you've already bought drops, you're entitled to a refund for the difference - Airlines: When a flight you've booked gets delayed or canceled, you're entitled to a cash compensation. - Enterprises: When a class-action lawsuit is filed against a company you engage with, you're entitled to money. - And many more, that even the government holds some of your unclaimed funds. Today, the dream of one service that automatically claims the money you deserve when you shop, fly,... and live is (finally) possible. We started with automating refunds in the shopping industry and have secured $8,000 in refunds to our users, growing 233% MoM. With AI, we've been able to repurpose that same automation to tackle refunds in the flight industry and the class-action lawsuit landscape. Why us? (1) We're obsessed with reclaiming what's rightfully yours: Kamal and Anthony grew up in Lebanon, where people's money was unfairly taken away. (2) Weijie was the lead infra engineer for Wish's recommendation system, the feature that fueled their IPO. Our combined experience includes Apple, Addepar, Bolt, Wish, and StackAdapt. And we're all working towards one goal: move those billions of dollars where they belong. pap! Founded:2024 Team Size:3 Location:Berkeley, CA ### Active Founders ### Kamal Nahas, Co-Founder, CEO Kamal earned a B.S. in Electrical Engineering and Computer Sciences from UC Berkeley, worked at Apple as a Security/ML software engineer intern, and met hundreds of engineers while working at Human Capital. He grew up in Lebanon, where he led operations resulting in 230+ families fed for an entire month and 25+ civilians housed. ### Anthony Rassi, Co-Founder, COO Anthony graduated from USC in Computer Engineering and Computer Sciences and worked at Addepar as a product engineer intern. He grew up in Lebanon, where he won a national hackathon, built a 5-figure affiliate marketing business, and led an engineering team at the country’s largest retailer. ### Weijie Wang, Founder Weijie is the Co-founder & CTO of pap!. Before pap!, Weijie spent 6 years in the e-commerce domain and was the lead infra engineer for Wish’s recommendation system. Waterloo & UIUC grad. ### Company Launches [### pap! — Automatic refunds for consumers]( **TLDR:** Every year, consumers miss out on more than **$50B** because they are not claiming the money they are entitled to. We built AI agents that claim it for them. The Problem 🚩 -------------- When stuff you’ve already bought goes on sale or when flights you’ve booked are delayed, policies entitle you to money. But spotting these opportunities and claiming the cash you’re owed is time-consuming and confusing. Industries subject to such policies include retail, airlines, hotels, technology, and even the government. Our Solution 💸 --------------- We're building AI that connects to your email inbox to: scan your e-receipts, identify opportunities to get you dollars, and automatically claim them by contacting customer service via email, phone, forms, or chat. We started by automating refunds in the shopping industry and have secured $6,650, growing **233% MoM**. With AI, we've been able to generalize that same automation to tackle refunds in the flight industry and the class-action lawsuit landscape. That’s when we understood why this had never been done before… Why now? 🚨 ----------- AI enables us, for the first time, to automatically: 1. **Reason about unstructured data at scale** 2. **Handle complex, nuanced customer service interactions** Generalizing securing refunds from one industry to all industries turns each step of the process into an unstructured data processing problem. For example, having a parser that can only understand Cotsco receipts and a scraper that can only read the price from the Costco website is not enough anymore: you need a parser that can understand changing receipts from all enterprises and a scraper that can read information from any website. The ability to reason about unstructured data at scale replaces the rigid, static software, solving the issue. ### In short, PAP! represents an opportunity to finally realize the dream customer journey — sign up once, forget we exist, and earn money effortlessly. Our Team 👨‍👨‍👦 ----------------- * [Weijie]( (CTO) spent 6 years in e-commerce and was the lead infra engineer for Wish’s recommendation system, the product that fueled their IPO. He was a Senior SWE at Bolt and StackAdapt. He graduated from Waterloo & UIUC. * [Kamal]( (CEO) earned a B.S. in Electrical Engineering and Computer Sciences from UC Berkeley, worked at Apple as a Security/ML software engineer intern, and met hundreds of engineers while working at Human Capital. He grew up in Lebanon, where he led operations resulting in 230+ families fed for an entire month and 25+ civilians housed. * [Anthony]( (COO) graduated from USC in Computer Engineering and Computer Sciences and worked at Addepar as a software engineer intern. He also grew up in Lebanon, where he won a national hackathon, built a 5-figure affiliate marketing business, and led an engineering team at the country’s largest retailer. Our Ask 🙏 ---------- 1. Sign up at [joinpap.com]( (_bonus_: share your feedback with us at [[email protected]]( 2. **Share this post and tell your friends about it!** Everyone deserves to earn the money they are entitled to.

Capitol AI

capitol-ai

https://www.capitol.ai/

Washington, DC, USA; Remote

Capitol is an infinitely smart AI research assistant with access to your important documents, enabling you to create high-quality content instantly. You can generate informed articles, white papers, marketing content, and tweet storms. Capitol has a D2C product, and a B2B product offering for companies that want to embed real-time content creation and editing into their applications. Politico has exclusively partnered with Capitol to power all Politico Pro users with AI content creation for policy and regulatory analysis. In the future, all publishers of content and data will have an AI research layer. Capitol will be the category-defining product to unlock this transformation.

Infinitely smart research assistant that generates content from data

B2B

B2B -> Productivity

S24

Active

Early

https://www.ycombinator.com/companies/capitol-ai

https://yc-oss.github.io/api/batches/s24/capitol-ai.json

Title: Capitol AI: Capitol is a custom perplexity for every website in the world. | Y Combinator URL Source: Markdown Content: ### Capitol is a custom perplexity for every website in the world. As Perplexity or Gemini replaces the Google Search we all use today, it will remove the need to ever visit a website. This is an existential risk to all media brands, which is why Capitol went from zero to $1.68M in revenue during YC: all media companies need to offer THEIR OWN Perplexity-like experience to maintain their direct relationship with customers. The most trusted brand in political news and analysis, Politico, chose us to power their Politico Pro Subscription product. Consumers expect instant personalized answers. Publishers won’t build this on their own. Capitol’s API lights up publishers' websites and apps so their end users can generate personalized search results, cited research reports, and even tweet storms using the publisher’s proprietary data—all within their paywall. Capitol is the new last mile of content delivery for publishing. This market is $16b in 2023, and will grow to $85.6b in 2030. Gemini and Perplexity are the imperial forces. Capitol is arming the rebels. ### Jobs at Capitol AI US / CA / Remote (US; CA) $100K - $140K 0.10% - 1.00% 3+ years Capitol AI Founded:2021 Team Size:11 Location:Washington, DC ### Active Founders ### Shaun Modi, Founder & CEO Shaun Modi is the founder and CEO of Capitol AI, an infinitely smart research assistant that instantly generates high-quality content from your documents. One of the founding designers of Airbnb, Shaun has been named one of Business Insider’s Top 75 Designers in Technology. Shaun has designed software products for technology giants, the government, and its agencies, including Google, the White House, DoD, NASA, CDC, and Motorola. At Airbnb, Shaun worked side by side with CEO Brian Chesky. ### thomas hallaran, Founder Software engineer and technical manager with a product mindset. Two decades of experience, including IC and Director positions in academia, venture backed startups, and the US Federal Government. ### Company Launches [### Capitol AI - Create powerful content from data]( Hey! We're Shaun and Tom from [Capitol AI]( and we’re thrilled to announce the launch of Capitol 2.0. **TL;DR** --------- After we got into YC, we asked ourselves: “What would Microsoft Word look like if it were built in the age of AI?” Our answer: A product that seamlessly combines deep research, generative images + text, and AI-assisted editing. [**This is Capitol 2.0**]( **The Business Problem** Content creation for business users is a slow, disjointed process involving multiple tools like search, proprietary databases, spreadsheets, slides, PDFs, and word processors. This complexity limits businesses with data-driven content creation workflows from scaling their content production and selling more products. **Our Solution** ---------------- Capitol simplifies content creation by integrating research and editing into one seamless experience—automating tasks to supercharge productivity. Capitol AI empowers enterprise customers with an API/SDK that automates the research and content creation process. It gives them the tools to personalize and scale their product offerings with less headcount. [**Watch the video**]( **B2B Customers** ----------------- **Enterprise** [Politico]( the global authority on the intersection of politics, policy, and power has exclusively partnered with Capitol to empower all [Politico Pro]( users with AI content creation for policy and regulatory analysis. **Startups & SMB** For startups and small businesses with repeatable data-driven processes like Cloud GTM provider [Invisory]( and the innovative AI-powered grant platform [AI Grants]( Capitol is automating the creation of **revenue-generating** business documents with our API-accessible artifact generation pipeline. **Why Now?** ------------ In a time when publishers are increasingly being courted by AI model builders for licensing deals, Capitol AI is arming the rebels—helping publishers compete for attention and grow their customer base by automating content generation from their high-value data. Capitol will be the category-defining product to unlock this transformation. **Tech** -------- Capitol includes a proprietary model orchestration, eval framework, chain of thought reasoning, and RAG pipelines that work together to enable our customers to generate and iterate on content in our product or via API. **How Capitol Works** --------------------- Rewrite, reformat, & transform your content Type ‘/’ anywhere to generate new content in-line Generate new content with a single click ### **🙏🏽 Our asks** **1\. Try Capitol** ------------------- Head to [ and try Capitol today. Get 3 months free with the special YC discount code **FOUNDERMODE** **2\. Integrate Capitol Into Your Product!** -------------------------------------------- Capitol’s embedded content generation will transform your application’s user experience with real-time RAG and document creation from your data. By leveraging Capitol’s API/SDK, you can drive more engagement with your product and have more time with your users. Common use cases include: * Give your users the ability to create bespoke reports and insights, pulling forward hard-to-find data * Turn enterprise document search into a content generation flow * Transforming your knowledge base or help center into a more helpful experience for your customers * Let your users create social media content directly on your site Contact us at [[email protected]](mailto:[email protected]) **Founders** ------------ * [Shaun Modi]( CEO. One of the first product designers hired at Airbnb. Worked side by side with CEO Brian Chesky crafting the user experience of Airbnb. Fmr Google, Fmr Nasa, named one of Business Insider's Top 75 Designers in Technology. BFA from RISD. * [Tom Hallaran]( CTO. Fmr Director US AI Center of Excellence, Fmr VP of engineering at acquired adtech startup Selectable Media, Initial technical lead and architect of the largest political payment app in US. Undergrad from Columbia. [X]( [Linkedin]( [Instagram]( #### YC Sign Photo #### Company Photo ### Hear from the founders #### How did you decide to apply to Y Combinator? What was your experience applying, going through the batch, and fundraising at demo day? Brian Chesky encouraged us to apply. He hired Shaun at Airbnb. We wish we had done it earlier, but in some ways, the timing of this couldn’t be better. Brian believes the position we are in is very similar to Airbnb's right before they joined YC. And even similar dynamics to when Justin Kan made the last-minute referral for Airbnb’s founders (Brian, Joe, and Nate) to join YC. Going through the batch has been an incredible experience. Garry, Ooshma, and Aaron are phenomenal partners to work with and boil down opportunities and challenges to overcome to their essence. We’ve made huge strides in the batch regarding product velocity and customer acquisition. We’ve signed a $1.3m customer term sheet with one of the most influential media brands during YC! #### What's the history of your company from getting started until the present day? What were the big inflection points? Tom and I were collaborating on Covid-19 logistics applications leveraging AI models for HHS and the US DoD and the interfaces that prime USG integrators like IBM, SAP, and Deloitte produced were unusable. We saw an opportunity to develop integration and design standards for creating intelligent content leveraging models, and Capitol AI was born. We started the company with a contract from the 2nd largest health insurer in the US to automate data storytelling for their self-insured employer business line. We approached this problem with a mix of classical and bayesian statistical analysis and automating insight generation leveraging DBT. After GPT 3.5 was released we saw an opportunity to center our product stack around LLMs. We made a pivot away from healthcare and so that we could collect user data and feedback to move the product forward as quickly as possible. We now have built a product that is in high demand we just closed a $1.3m enterprise contract, with more to come. #### What is your long-term vision? If you truly succeed, what will be different about the world? In the future, all content and data publishers will have an AI research layer. Capitol will be the category-defining product that unlocks this transformation, creating a more useful and personalized internet. ### Selected answers from Capitol AI's original YC application for the S24 Batch #### Describe what your company does in 50 characters or less. Create documents that convince, persuade & educate #### How long have each of you been working on this? How much of that has been full-time? Please explain. Just under 3 years, full time. #### What is your company going to make? Please describe your product and what it does or will do. Capitol is building a product that makes it easy to create content to convince, persuade and educate an audience. The product is a fast way to create text, image and data rich documents. Capitol is like a ghostwriter, researcher, and intuitive document editor all in one. ### YC S24 Application Video

Moment Robotics

moment-robotics

http://www.momentrobotics.com/

Remote

Moment Robotics makes robots to automate the movement of goods, starting inside hospitals. We transport, place, and organize supplies, samples, and equipment so hospital staff can focus on their patients.

Robots to automate the movement of goods within hospitals

Industrials

Industrials -> Manufacturing and Robotics

S24

Inactive

Early

https://www.ycombinator.com/companies/moment-robotics

https://yc-oss.github.io/api/batches/s24/moment-robotics.json

Title: Y Combinator | File Not Found URL Source: Warning: Target URL returned error 404: Not Found Markdown Content: 404 --- [Back to the homepage]( support please contact [[email protected]](mailto:[email protected])

Blaze

blaze-2

https://blaze.money

Mexico City, CDMX, Mexico; Remote

Blaze is Global Venmo for cross border payments. We combine the UX of Venmo with the speed, global reach, and low cost of USDC. If you're a Californian nomad living in Mexico City, you can use Blaze to pay rent to your Mexican landlord. And if you're an expat living in Colombia, you can use Blaze to pay your Spanish teacher for classes. In addition to P2P, we're also enabling a network of travel-related businesses to accept payments on Blaze and reach our customer base of global citizens. For example, a Travel Concierge business in Jamaica can accept payments on Blaze from her US customers coming to visit for the summer. During a 6 month stint traveling around Asia, Faiyam experienced first hand the frustrations of paying people from different countries. Luc was experiencing the same in his travels around Latin America. Fees were exorbitant, apps that worked for one person didn't work for the other, and the user experiences were cumbersome. Enter Blaze. Cross-border P2P payments is projected to be a $9T market by 2030. Our long term vision is that you should be able to drop in to any country on the planet and pay anyone you meet simply, quickly, and affordably. It's easy to send an sms to anyone anywhere in the world without thinking about where they're from. We want to do the same with payments. The advent of consumer friendly blockchain tech, especially stablecoins, presents a unique opportunity to do this. The legacy financial system stitches together otherwise siloed country-specific systems to handle cross-border payments. Blockchain has no such notion of a border in the first place. As such the present moment is a unique, generational opportunity to rewrite the rules of global finance. Faiyam studied Math at Duke and engineering at Cornell before building a $1M ARR company with Luc. Luc’s a former technical lead at Spotify and Artsy, who studied Engineering at NYU and has cofounded 2 VC-backed startups, one that hit #1 in app store and got him sued by Drake. Igor led design at Bitpanda, Austria’s first unicorn.

Global Venmo for cross border payments

Fintech

Fintech -> Payments

S24

Active

Early

https://www.ycombinator.com/companies/blaze-2

https://yc-oss.github.io/api/batches/s24/blaze-2.json

Title: Blaze: Global Venmo for cross border payments | Y Combinator URL Source: Markdown Content: ### Global Venmo for cross border payments Blaze (YC S24) is global Venmo for cross border payments. It's a peer to peer payments app that uses USDC to make payments fast and cheap between any two people anywhere in the world. The founders scaled their last startup to $1M in ARR, studied mathematics and engineering at Duke, NYU, and Cornell, and have led teams at Spotify, Artsy, and Bitpanda. They're live in the US and Mexico, available for download on iOS and Android: Blaze Founded:2023 Team Size:5 Location:Mexico City, Mexico ### Active Founders ### Faiyam Rahman, Founder Faiyam is a full-stack engineer with a bachelors in mathematics from Duke and a masters of engineering from Cornell Tech. He is now the cofounder and CEO of Blaze Payments. He is an avid traveler, most recently having completed a 6 month tour of India, Bangladesh, Nepal, Bali and Portugal. He is also an avid practitioner of yoga and meditation, and he enjoys long walks, the spoken word, and celebrating the joy of being alive. ### Luc Succès, Founder Luc is a 2 time VC backed founder. His first startup, Often, launched a keyboard app that enabled you to text your friends with Drake lyrics. Upon launch, Often blew up to #1 on the App Store and got over 400k downloads. He scaled his second startup Seasons, a rental service for high end menswear to $1mm in ARR. Luc is a seasoned tech leader, having led teams at Artsy & Spotify. He is also an avid traveler, having lived and worked in over 50 countries worldwide. ### Igor Zagnienski, Founder Igor holds a bachelor's degree in marketing from the School of Economics and Law in Kielce. Prior to Blaze, he co-founded Space Individuals, a top-tier job site dedicated to the space industry. Adding to his diverse experience, he was a Senior Designer at Bitpanda, Austria's first billion dollar tech success. ### Company Launches [### Blaze - Global Venmo for cross border payments]( **TL;DR**: Blaze is building a global Venmo for cross-border payments. For example, if you're a Californian nomad living in Mexico City, you can use Blaze to pay rent to your local landlord. Or if you're an expat living in Colombia, you can use Blaze to pay your teacher for Spanish classes. Check [**us out**]( on iOS and Android! A [quick intro]( for you: A little over a year ago, we found ourselves traveling all over the world. Luc was spending time between Mexico, Colombia, and other parts of South America, while Faiyam was nomad-ing around Asia. Everywhere we went, it was a pain to transact with our new friends from other countries. We found ourselves wanting a Venmo like experience that worked for our international lives. So we decided to build it. Blaze is the payments app we wanted for our own lives, and if you have a lot of international contacts, perhaps it is for you too. _from left to right:_ 💻 [Luc]( studied engineering at NYU before going on to start 2 VC-backed startups, one of which went to #1 on the app store and got him sued by Drake (true story). He’s also been a tech lead at Spotify and Artsy, and has lived and worked in over 50 countries worldwide. 🧘‍♂️ [Faiyam]( studied mathematics at Duke, then engineering at Cornell before going on to build a $1M ARR business with Luc at his 2nd startup. He’s also an avid practitioner of yoga and meditation. 🧑‍🎨 [Igor]( is a former lead designer at BitPanda, Austria’s first unicorn. Prior to that, he spent his time jumping out of planes and running Space Individuals, one of Europe’s leading platforms to find jobs in Space. 📱 **App Preview** ------------------ **🙏 Asks** ----------- * Help us beta test! [**Download**]( on iOS and Android. You can share feedback at [[email protected]](mailto:[email protected]). * **Share this post!** If you know someone with an international lifestyle who would want to use Blaze, please share this post with them. * **Connect us with** **businesses** in travel, hospitality, or other spaces that could use help with cross-border payments. #### YC Sign Photo #### Company Photo

reworks

reworks

https://www.reworks.ai

San Francisco, CA, USA

reworks is the fastest way for agentic AI companies to integrate with external software. Instead of spending weeks integrating with traditional API:s or writing custom scraping scripts, teams use reworks to build browser-based integrations that don't break.

Browser-based integrations that don't break

B2B

B2B -> Productivity

S24

Active

Early

https://www.ycombinator.com/companies/reworks

https://yc-oss.github.io/api/batches/s24/reworks.json

Title: reworks: Turn external software into APIs that don’t break | Y Combinator URL Source: Markdown Content: ### Turn external software into APIs that don’t break reworks is the fastest way for agentic AI companies to integrate with external software. Instead of spending weeks integrating with traditional API:s or writing custom scraping scripts, teams use reworks to build browser-based integrations that don't break. reworks Founded:2024 Team Size:2 Location:San Francisco ### Active Founders ### Melvin Hagberg, Founder Melvin is the CEO and co-founder of Reworks. At a young age, Melvin fell in love with computers and ended up getting his first job as a software engineer at an European unicorn at the age of 14. While in high school, Melvin built and sold software integrations for $150K. After graduation, he served as the Head of Tech at a Stockholm-based e-commerce startup for 2 years, and studied at the top finance school in Northern Europe (SSE) before ultimately dropping out 3 months in to start Reworks. ### Erik Heimdal, Founder Erik is the co-founder of Reworks. A self-taught programmer who began his journey with a passion for cyber security. Prior to Reworks, he worked part time for a startup while studying CS at one of the top engineering schools in Europe, did bug hunting for Fortune 500 companies, and competed for the #1 ranked cyber security team in Sweden. ### Company Launches [### reworks - the fastest way to integrate with external software]( Hey everyone! We’re Melvin and Erik from reworks. **TL;DR:** reworks is the fastest way for agentic AI companies to build integrations with external software. Instead of spending weeks integrating with traditional APIs or writing custom scraping scripts, teams use reworks to build browser-based integrations that don't require any maintenance. For example, we help customers in: * 🧑‍⚕️ **healthcare** integrate with legacy EHR and PMS systems. * 💰 **finance** to perform actions or scrape data from legacy ERPs, bookkeeping software, and e-commerce platforms. * 📦 **order management** to integrate with ERPs, WMS, and CRMs. * ⚖️ **legal** integrate with DMS and CLM systems. * 🤝 **HR & Payroll** build connections with HRIS, ATS, and payroll software. * … and much more! **❌ The Problem** ----------------- Teams building agentic AI today spend up to half of their time on integrations with external software. For example, a team building AI scheduling software for healthcare might have to integrate with 10s or even 100s of EHR systems - wasted time that could instead be spent on improving the core product. Additionally, many legacy systems lack official APIs, leading to developers building custom, browser-based integrations. Writing the scraping scripts is time-consuming and often results in highly unreliable and heavy maintenance. Underlying UI changes, such as an innocent popup, can completely break integrations. **🏆 The Solution** ------------------- With reworks, teams can record workflows in any browser-based software and turn them into reliable API endpoints in minutes. Integrations are built by simply recording yourself navigating the software UI and explaining in natural language what the return JSON data should look like. With reworks, you get: * **Days or weeks are saved** per integration using our AI-powered workflow recorder. * **Worry-free maintenance** - If workflows break due to changes in the underlying UI, we use AI and computer vision to repair them, keeping your endpoints reliable at all times. * **Fully managed authentication** for your end users, including storing their passwords and securely using them to log in to software on their behalf. To illustrate, here’s a demo of Melvin building an integration to Amazon in <6 minutes: [ **🧚 Why we started working on this** ------------------------------------- reworks started off as an AI assistant for e-commerce customer support. It was all smooth sailing until we started to onboard customers and realized that for each new customer we wanted to onboard, there were several systems that we had to integrate our AI agent with (e.g., ERPs, WMS, and CRMs), each taking weeks to build via traditional APIs. We spoke to over 40 other YC companies building in agentic AI and realized that external integrations are fundamental for the new wave of AI companies, but building them is way more painful than it should be. We’re on a mission to change that. **🇸🇪 The Team** ----------------- We’re two passionate AI builders originally from Sweden. [**Melvin Hagberg**]( (left) started his career as a software engineer at a unicorn at the age of 14. While in high school, he built a business selling integrations for $150k and has spent the last two years as the Head of Tech at a Stockholm-based e-commerce startup. He spent two months at the top finance school in Northern Europe (SSE) before deciding to drop out to build reworks. [**Erik Heimdal**]( (right) is a self-taught programmer who began his journey in cyber security and reverse engineering. Prior to reworks, he studied CS at one the top engineering schools in Europe, competed for the #1 ranked cyber security team in Sweden and worked as a software engineer at several startups. **🙋Our Ask** ------------- **Have a backlog of integrations to do?** We are offering concierge experience to early customers. That means we’ll use our own product to deliver integrations to you in the next 24 hours. Reach out to learn more at [[email protected]](mailto:[email protected]) or [book a time with Melvin]( **Please give us feedback!** Drop us a line at [[email protected]](mailto:[email protected]). **Share this post!** Intros are highly appreciated. Quick blurb to copy and paste:  reworks makes it easy to build browsed-based integrations with any external software (e.g. EHR, PMS, ERPs, WMS, and CRMs). They are currently offering to use their own tool to build integrations for you in the next 24 hours. Learn more at [reworks.ai]( or reach out to [[email protected]](mailto:[email protected]).

FirstWork

firstwork

https://firstwork.com

San Francisco, CA, USA; Remote

Firstwork is a next-generation HR platform designed specifically for companies with dynamic workforce needs. Our mission is to streamline and automate key HR processes to help businesses operate more efficiently and effectively. What We Do (for now): (1) Smart Onboarding 📋: Our platform simplifies the onboarding process by automating the collection and verification of credentials, such as right-to-work documents, insurance forms and driver’s licenses. This reduces administrative burden and accelerates the time-to-productivity for new hires. (2) AI-Assisted Activation 🤖: Leveraging AI, we enhance communication and messaging for new hires, ensuring they receive timely and relevant information. This helps in better engagement and faster integration into their roles. (3) Perpetual Compliance ✅: Our system uses AI and OCR technology to continuously monitor and ensure compliance with various regulatory requirements. This includes tracking expirations of critical documents, updating records, and generating compliance reports, thus reducing the risk of penalties and legal issues.

Rippling for frontline workers

B2B

B2B -> Human Resources

S24

Active

Early

https://www.ycombinator.com/companies/firstwork

https://yc-oss.github.io/api/batches/s24/firstwork.json

Title: FirstWork: Rippling for seasonal workers | Y Combinator URL Source: Markdown Content: ### Rippling for seasonal workers Firstwork is a next-generation HR platform designed specifically for companies with dynamic workforce needs. Our mission is to streamline and automate key HR processes to help businesses operate more efficiently and effectively. What We Do (for now): (1) Smart Onboarding 📋: Our platform simplifies the onboarding process by automating the collection and verification of credentials, such as right-to-work documents, insurance forms and driver’s licenses. This reduces administrative burden and accelerates the time-to-productivity for new hires. (2) AI-Assisted Activation 🤖: Leveraging AI, we enhance communication and messaging for new hires, ensuring they receive timely and relevant information. This helps in better engagement and faster integration into their roles. (3) Perpetual Compliance ✅: Our system uses AI and OCR technology to continuously monitor and ensure compliance with various regulatory requirements. This includes tracking expirations of critical documents, updating records, and generating compliance reports, thus reducing the risk of penalties and legal issues. FirstWork Founded:2024 Team Size:9 Location:San Francisco ### Active Founders ### Vardhan Kapoor, Founder Co-founder and CEO @ Firstwork, Rippling for frontline workers. Previously: Head of Region, APAC at Remitly (NASDAQ: RELY) where I launched and scaled Remitly’s expansion into the APAC region, building the P&L from ground up to high-eight figures USD ARR across 5+ geographies. Strategy Head at Deliveroo (LSE:ROO) where I led regional growth and strategy. ### Shubham Choudhary, Founder Co-founder and CTO @Firstwork, Rippling for frontline workers. Previously: Director of Engineering @Deel, led the development and management of PEO, Immigration/Mobility, Time-Off & Time-Tracking products at Deel. Built out the core infrastructure for the Employer Of Record (EOR) systems at @Rippling, which enabled Rippling’s market entry into end-to-end employee lifecycle in 50+ countries. ### Company Launches [### 🚀 Firstwork - Rippling for frontline workers]( **tl;dr**: [**Firstwork**]( **is building Rippling for frontline workers.** Hi everyone! We are Vardhan and Shubham, and we’re on a mission to build an all-in-one HR and compliance platform for frontline workers. 🧨 **The Problem:** Managing workforces in sectors like transport, delivery, and logistics is complex and inefficient. Many sectors have seasonal onboarding/offboarding and onerous credential validation that bottleneck HR Ops teams. Traditional HRIS solutions are scattered across multiple systems, which leads to delays, compliance issues, and high operational costs. Working at Deliveroo and Remitly, where we spent a lot of time partnering with frontline workers across the gig economy, trucking, and logistics industries. Managing HR and compliance was a constant struggle. 🎉 **The Solution:** At **Firstwork**, we realize that worker onboarding, compliance, and management are not standalone problems but interconnected opportunities that require a unified approach. It’s with these insights that we’re building a compound product, with an initial focus on: * **Smart Onboarding 📋:** Automates credential verification with Intelligent Process Automation and OCR, reducing admin burdens and speeding up time-to-productivity. * **AI-Assisted Communication 🤖:** Enhances new hire activation for better engagement and faster integration. * **Perpetual Compliance ✅:** Uses AI and OCR to ensure continuous compliance, tracking document expirations and updating records. **Using Firstwork, Employees:** ✔️ Get onboarded faster and with less hassle ✔️ Receive timely and relevant communication for better engagement ✔️ Stay compliant with regulations effortlessly **and Companies:** ✔️ Avoid compliance issues and potential penalties ✔️ Improve overall worker efficiency per hour  ✔️ Save time, costs, and reduce administrative burdens 💲 **Opportunity:** The scope and impact of our market are significant. Companies in logistics, warehousing, delivery, hospitality, and BPO centers can benefit from our platform, improving their workforce management processes and saving costs. **Our Team:** [Vardhan Kapoor]( our CEO, has firsthand experience with blue-collar worker engagement challenges at Deliveroo and Remitly, where he scaled various business units to high eight-figure USD ARR. [Shubham Choudhary]( our CTO, built the core HR infrastructure at Rippling and Deel and holds multiple US patents. 👋 **Ask:** How you can help * We’d love to connect with companies in the logistics, ride-hailing, warehousing, trucking, delivery, hospitality, and BPO centers. * If you know potential customers or partners, please reach out at [[email protected].](mailto:[email protected]) * We’d also love any feedback or advice from this incredible community. #### YC Sign Photo

Thunder Compute

thunder-compute

https://thundercompute.com

Atlanta, GA, USA

Thunder Compute has created virtual GPU-over-TCP technology (think PCI-express over a network). This technology creates a new cloud instance with virtual, on-demand GPUs. What this means for developers is one cloud instance now accommodates any development workflow. A developer can build and debug in a CPU-only dev machine, then scale to a massive cluster with one command. This flexibility saves time spent configuring instances and cost of paying for idle compute during development. In the background, our technology dramatically cuts CapEx for cloud providers. It does this by: 1) Placing multiple workloads on a single GPU with dynamic memory sharing, similar to vCPUs on AWS, improving utilization 2) Allowing CPU-only instances to share pooled GPU resources rather than dedicated cards that can't be accessed by other machines, further improving utilization GPU virtualization also saves OpEx by reducing IT and procurement processes required to get developers access to GPUs. Notably, this appears to be the most efficient technology for serverless GPU computing, achieving instantaneous cold starts by decoupling the GPU environment from the physical GPU.

Cloud Computing Technology to virtualize GPUs over TCP.

B2B

B2B -> Infrastructure

S24

Active

Early

https://www.ycombinator.com/companies/thunder-compute

https://yc-oss.github.io/api/batches/s24/thunder-compute.json

Title: Thunder Compute: Software to virtualize GPUs over TCP. | Y Combinator URL Source: Markdown Content: ### Software to virtualize GPUs over TCP. Thunder Compute creates software to virtualize GPUs over a network. With a single command we trick your computer into thinking it’s directly attached to any number of cloud GPUs, saving developer hours of setup while allowing us to serve 5x more users with the same hardware. Developers save time spent configuring instances and cost of paying for idle compute during development. In the background, our technology dramatically cuts CapEx of GPUs in any data center or cloud. It does this by: 1) Placing multiple workloads on a single GPU 2) Allowing CPU-only instances to share pooled GPU resources GPU virtualization also saves OpEx by reducing IT and procurement processes required to get developers access to GPUs. Thunder Compute Founded:2024 Team Size:2 Location:Atlanta, GA ### Active Founders ### Carl Peterson, Founder CEO of Thunder Compute, formerly Management Consultant at Bain ### Brian Model, Founder CTO of Thunder Compute, formerly Quantitative Developer at Citadel Securities ### Company Launches [### Thunder Compute: Access cheap on-demand GPUs with our virtual GPU-over-TCP technology]( With [Thunder Compute]( run your current workload on cloud GPUs with one line of code. You can develop without paying for a GPU and then instantly connect your environment to as many GPUs as you need to move to production. ### **Watch it in action** [ Hi everyone, we’re Carl and Brian, aka the team behind [Thunder Compute.]( Before Thunder Compute, [Carl]( was a Management Consultant at Bain & Company, and [Brian]( was a Quantitative Developer at Citadel Securities. Carl and Brian met at Georgia Tech and have been close friends for over 6 years. ### **Problem** Developers spend hours configuring cloud platforms to meet the ever-changing needs of their projects. Today’s cloud platforms trade-off between: * Scalability * Cost * Simplicity ### **Solution** Thunder Compute is a flexible cloud environment that adapts to the needs of any project with a single command, for example: **_Using your local filesystem?_** ``` $ tnr start --sync ``` **_Scaling to 50 A100 GPUs?_** ``` $ tnr device A100 -n 50 ``` **_Running a Jupyter Notebook?_** ``` $ notebook ``` **_Developing without a GPU?_** ``` $ tnr device cpu ``` All of this is available at industry-leading prices and can be set up in minutes - just **pip install tnr.** ### **Our innovation** Think VMware for GPUs: Thunder Compute tricks your computer into thinking it’s attached to a GPU, which actually sits across a network. This means that CPU-only machines can behave exactly as if they have dedicated GPUs, while the physical GPUs are actually shared among several machines. ### **Our offer** To help us test our systems at scale, we are giving **free open beta access to NVIDIA T4 instances.** Try Thunder Compute today at [**thundercompute.com**]( all we ask in return is that you please report any bugs 🙂 If you have questions, need technical support, or want to rant about the current state of GPUs, join our discord at [** or contact us at [**[email protected].**]( #### YC Sign Photo ### Hear from the founders #### How did your company get started? (i.e., How did the founders meet? How did you come up with the idea? How did you decide to be a founder?) Carl Peterson and Brian Model met as freshmen at Georgia Tech and became close friends. The idea for Thunder Compute sparked from Brian's experience in his Systems for AI lab, where researchers made GPU reservations weeks in advance via Google Sheets, severely hindering research progress. This inefficiency inspired us to explore ways to improve GPU utilization and the developer experience of using GPUs. \\Our thesis was (and still is) that virtualization over a network is the optimal way to manage GPUs within a cloud platform and founded Thunder Compute to bring our technology to market. ### YC S24 Application Video

Central

central

https://centralhq.com

San Francisco, CA, USA

- [Our fever dream Bookface launch](https://bookface.ycombinator.com/posts/81742) - [$10K+ high value YC deal + reviews](https://bookface.ycombinator.com/deals/2463)

We eliminate founder bullshit: payroll, benefits, govt. filings etc.

B2B

B2B -> Human Resources

S24

Active

Early

https://www.ycombinator.com/companies/central

https://yc-oss.github.io/api/batches/s24/central.json

Title: Central: Eliminating business bullshit: payroll, benefits, govt. filings, etc | Y Combinator URL Source: Markdown Content: ### Eliminating business bullshit: payroll, benefits, govt. filings, etc \- \[Our fever dream Bookface launch\]( - \[$10K+ high value YC deal + reviews\]( ### Jobs at Central Washington, DC, US / San Francisco, CA, US / Remote (US) $200K - $250K 0.25% - 1.00% 6+ years Central Founded:2023 Team Size:13 Location:San Francisco ### Active Founders ### Josh Wymer, Founder Founder / CEO @ Central, ex-Mixpanel, ex-Meta ### Pranav Kashyap, Founder Pranav is the COO at Central, helping other companies automate back-office work. He previously built payment and tax products for millions of users at Stripe, rebuilt Mixpanel's self-serve and messaging products and spent a number of years working with technology companies at McKinsey and Sequoia Capital. He holds a bachelor's degree in Computer Science from IIT Bombay and an MBA from Wharton. ### Nilay Modi, Founder Founder at Central ### Company Launches [### Central: Payroll + benefits for startups. Save weeks of work & fines.]( **📝 TL;DR:** ------------- * [Central]( is the first payroll & benefits platform for startups that automates all the painful compliance work – saving you weeks of work and fines. * **Start for free -** [** * _10 mins to setup + we’ll do all the migration work._ —— -- 🤫 **CONFESSION:** **I’ve paid over $20,000 in government fines.** ------------------------------------------------------------------ At my last YC W20 company, Clayboard, all I wanted to do was design and build. Instead, I was drowning in back-office work. Even with the current recommended platforms, payroll, benefits, compliance was a never-ending pain in the… Why do founders still need to understand all this? Why do we still have to register in multiple states to run payroll, set up worker’s compensation, pay random payroll related taxes, deal with scary government mail, update addresses across state filings, troubleshoot with insurance companies? A tragedy… —— -- **👋 HELLO CENTRAL: Save weeks of work & avoid fines.** ------------------------------------------------------- Central is the first payroll, benefits & ops platform designed for busy founders, not HR/finance professionals. Unlike other platforms, Central handles each compliance task from start to finish. You don’t need to learn or do anything. You just keep building because that’s what truly matters. Think of Central as abstracting away the government and also declarative i.e. you set your desired end state, and Central makes it happen. So, here’s everything you get: **Payroll & compliance** 1. **Hire and pay anyone globally:** Central supports founders & employees in all 50 U.S. states and contractors in 200+ countries. We’ll handle reimbursements, offer letters, NDAs, I-9 verification, and more. 2. **Automated payroll state registrations:** Existing platforms make founders interface with state agencies, leading to errors & fines. Central automates this entire process resulting in accurate tax IDs and withholding rates. 3. **Automated payroll tax filings:** Existing platforms cover some taxes, and leave founders to handle the rest, leading to mistakes and fines. Central pays all taxes and verifies payments with federal, state, and local agencies. 4. **Automated worker’s comp:** When you hire your first employee in a state, you need state-specific worker’s compensation insurance. Existing platforms make you go to other platforms to get a policy, leading to missed coverage and fines. Instead, Central sets it up and ensures payment automatically. 5. **Government mail monitoring**: Founders often miss important government mail due to address changes. To combat this, Central adds a virtual mailing address to all of your state filings. This virtual address stays the same when you move, so Central can notify you about government mail and take the appropriate action. **Business compliance & insurance** 1. **Foreign qualifications & city registrations:** Central handles the necessary business filings & the registered agent in your HQ city/state and other states when required. 2. **Continuous BOI reporting to FinCEN:** The Corporate Transparency Act requires companies to submit beneficial ownership information (BOI) reports to FinCEN. Central files this for you and keeps it updated as officers and their information change. 3. **Business insurance:** Central can set up insurance for general liability, directors and officers, cyber, errors and omissions, and more. **Benefits** 1. **Medical, dental & vision insurance:** Central offers plans from top carriers for small groups and individuals. We automate setup and payments, making it easy for you. Pranav (co-founder at Central) is a licensed broker and expert and can guide you to avoid common mistakes. 2. **Life insurance:** Integrated with payroll for pre-tax deductions. 3. **401k:** Central partners with Guideline to provide easy payroll integration and pre-tax deductions. And, support via a shared Slack channel, phone & email. —— -- **🏎️ GET STUFF DONE INSTANTLY VIA SLACK** ------------------------------------------ Existing platforms are a maze. You need to dig through 7 layers of UI with a shovel to get anything done. With Central, you simply ask our bot, and it handles it for you (with your confirmation). It’s like your company’s first head of HR/finance. Our customers love the ease of: * **Retrieving company info & docs:** Get company EIN, certificate of incorporation, W-9, etc. * **Updating employee info:** Adjust salaries, start dates, titles, etc. * **Accessing personal docs:** Find offer letters, paystubs, health insurance cards, W-2s etc. * **Tracking offers & payroll:** See when employees sign, onboard, and get paid. * **Understanding their paystubs:** Break down taxes, pay dates, and more. ([Link to GIF]( if not loading) —— -- **🧑‍💻 ABOUT US** ------------------ * **Mission:** Liberate founders to create value. * **Started:** 1.5 years ago * **Customers:** 50+ YC companies * **Team:** 3 co-founders + 7 full-time employees + 3 contractors * **Founders:** * [Josh Wymer]( Previously led engineering teams at Meta and Mixpanel (second product engineer). * [Pranav Kashyap]( Previously led product teams at Stripe and Mixpanel * [Nilay Modi]( Previously a co-founder and fine payer at Clayboard (W20) —— -- **🎁 START FOR FREE -** [ ------------------------------------------------------------------------------------------- _10 mins to setup + we’ll do all the migration work_

Anglera

anglera

https://www.anglera.com/

San Francisco, CA, USA

At Anglera, we're building an AI-native operating system for e-commerce. We previously enriched millions of products at Uber Eats, and we're leveraging our Stanford AI expertise to reimagine how businesses manage product data at scale. This is an existing $10B software category (Product Information Management).

AI-native operating system for e-commerce

B2B

B2B -> Retail

S24

Active

Early

https://www.ycombinator.com/companies/anglera

https://yc-oss.github.io/api/batches/s24/anglera.json

Title: Anglera: AI-native operating system for e-commerce | Y Combinator URL Source: Markdown Content: ### AI-native operating system for e-commerce At Anglera, we're building an AI-native operating system for e-commerce. We previously enriched millions of products at Uber Eats, and we're leveraging our Stanford AI expertise to reimagine how businesses manage product data at scale. Anglera Founded:2024 Team Size:2 Location:San Francisco ### Active Founders ### Amay Aggarwal, Founder Amay is the Co-Founder of Anglera. He graduated from Stanford with Bachelor's & Master's degrees, specializing in AI/ML. Previously, he led Catalog AI efforts to onboard, enrich, & standardize millions of products sold on Uber Eats. ### Ray Iyer, Founder Ray is the co-founder of Anglera. Before Anglera, Ray spent 2 years as a Product Manager at Uber, where he led a XFN team to design and launch the end-to-end consumer and driver experiences for UberTasks (a 0-\>1 growth bet). Ray also launched the first foundational CPG Ads products (brand storefronts, sponsored items) on the grocery delivery marketplace. Ray has a MS and BS in Computer Science from Stanford, with an AI/ML specialization. ### Company Launches [### 🛍️ Anglera: The smartest way to manage e-commerce products]( **TLDR**: [Anglera]( is an AI-powered platform designed to help e-commerce companies manage their product data. We empower brands, retailers, and marketplaces to effortlessly onboard, enrich, and list hundreds of product SKUs. This saves time and improves sales. Contact us to learn more: [[email protected]](mailto:[email protected]) ### **👋 Who we are** Hello! We’re [Amay]( & [Ray]( and we’re on a mission to redefine how e-commerce companies manage their product data. Whether you’re a brand, retailer, or marketplace, we know that dealing with e-commerce products can be challenging. We faced this problem first-hand at Uber Eats, where we built tech to enhance millions of listings uploaded by our restaurant and retail partners. ### **🔍 Problem** Most rapidly growing e-commerce companies spend several hours each week managing their products across sales channels. Teams drown in spreadsheets, manually adding descriptions, images, and attributes for hundreds of items. This time-consuming process diverts focus away from actually growing their core business. ### **💡 Solution** We’re tackling the $20B Product Information Management (PIM) industry by building the smartest platform for managing e-commerce product data. [ Anglera empowers every e-commerce business with their own AI catalog ‘expert’ trained on thousands of top-converting product listings. We not only help our customers onboard new products faster, but also improve all of their listings to drive better sales. **Key Features:** * **✅ Faster Time-To-Market**: Instantly onboard products from any source (Website URL, Spreadsheet, PDF, etc.) * **✅ Better Sales**: Automatically enrich descriptions, images, & attributes to drive conversion * **✅ Wider Reach**: Start selling on new channels with just one click (Shopify, Amazon, etc.) We measure success for our customers with two simple metrics: 1) $ increase in sales, and 2) # of hours saved. ### **🙏 Asks** 1. **Let’s chat** if you're a brand/retailer/marketplace looking to improve how you manage your product listings. Happy to share data-driven learnings from years of experience working on this at Uber Eats. 2. **Introduce us** to any e-commerce companies in your network. Whether it's a friend running their own brand or someone who works at a larger retailer or marketplace, we’d love to connect. Contact: [[email protected]](mailto:[email protected]) | Website: [ | Calendar: [Book Time](

Zimi

zimi

https://www.withzimi.com

Zimi is a tech-enabled fulfillment provider that is bringing Amazon Prime-like delivery to exporters in emerging markets. We help merchants deliver to their international buyers in 2 days instead of 2 weeks, and at a fraction of the typical shipping cost (80% cheaper on average). Cross-border e-commerce is a $2 trillion opportunity by 2030, with 64% of that coming from emerging markets across Africa, Asia, and Latin America. However, merchants in these regions face major challenges like high shipping costs, lengthy delivery times, and complex logistics. Zimi solves these problems by localizing inventory, handling customs clearance and last-mile delivery, and streamlining the returns process. Our aim is to make selling internationally as easy as selling locally.

Amazon-like fulfillment for exporters in emerging markets

B2B

B2B -> Supply Chain and Logistics

S24

Active

Early

https://www.ycombinator.com/companies/zimi

https://yc-oss.github.io/api/batches/s24/zimi.json

Title: Zimi: Fulfillment by Amazon for Exporters in Emerging Markets | Y Combinator URL Source: Markdown Content: ### Fulfillment by Amazon for Exporters in Emerging Markets Zimi is a tech-enabled fulfillment provider that is bringing Amazon Prime-like delivery to exporters in emerging markets. We help merchants deliver to their international buyers in 2 days instead of 2 weeks, and at a fraction of the typical shipping cost (80% cheaper on average). Cross-border e-commerce is a $2 trillion opportunity by 2030, with 64% of that coming from emerging markets across Africa, Asia, and Latin America. However, merchants in these regions face major challenges like high shipping costs, lengthy delivery times, and complex logistics. Zimi solves these problems by localizing inventory, handling customs clearance and last-mile delivery, and streamlining the returns process. Our aim is to make selling internationally as easy as selling locally. Zimi Founded:2023 Team Size:2 Location: ### Active Founders ### Audrey Djiya, Founder/CEO Audrey Djiya is the Co-Founder & CEO of Zimi. Before Zimi, Audrey led international expansion and unlocked $M in revenue for e-commerce businesses like Peloton and launched fintech products to enable commerce at start-ups like Duplo (YC W23). She launched Nigeria’s first Buy-Now-Pay-Later solution. Audrey earned her MBA from Stanford University and BA from Dartmouth College. ### Peter Nsaka, Founder/CTO Software Engineer with 8+ years of experience. Previously a Senior Software Engineer and Project Tech Lead @ Shopify. Currently working on Zimi to bring Amazon-prime like fulfillment to African exporters - making it easy, fast, and cheap for these businesses to sell internationally. ### Company Launches [### Zimi: Fulfillment by Amazon for exporters in emerging markets]( **🛎TLDR:** We’re bringing Amazon-prime-like fulfillment to exporters in emerging markets - making it easy, fast, and cheap for these businesses to sell internationally. **For example: Businesses selling with Zimi can deliver parcels to their international buyers for as little as ~$10 (instead of $80) and in just 2 days time.** **❌ Problem** **Cross-border e-commerce is projected to reach $2 trillion** in merchandise value by 2030, with 64% of this opportunity coming from emerging markets. However, merchants in these regions face significant obstacles: * **High shopping costs** often exceed 100% of order value * **Lengthy delivery times** of 2 weeks or longer * **Costly cross-border return processes** are a major financial drain * **Complex and time-consuming logistics management** (customs/duties, paperwork, tracking) These challenges result in an estimated $50 billion annual loss for African businesses alone, with similar figures in Latin America and South Asia. **📦 Solution** [Zimi]( makes selling globally as easy, affordable, and simple as selling locally by optimizing cross-border logistics: * **Cheap and Faster Shipping:** We enable 2-day delivery at up to 80% less than current shipping costs by localizing inventory * **Simplified Returns:** We take the headache out of international returns, managing the entire process to reduce costs and improve customer satisfaction, saving businesses thousands of dollars. * **Streamlined Cross-Border Operations:** Zimi handles the complexities of international shipping, customs clearance, warehousing, and last-mile delivery, freeing merchants to focus on growing their business instead of logistics Zimi’s first customer sold $25K in merchandise within 48 hours of announcing 65% cheaper shipping costs and fast delivery timelines with Zimi. We’re on track to unlock millions in revenue for businesses across India and Nigeria - our first two key export markets. **🧑🏾‍💼👩🏾‍💼About Us** We’re Peter and Audrey. We met while completing our respective graduate school programs in CS and an MBA at Stanford. * [Peter]( is a second-time founder with experience in cross-border freight, SMB payments, and building commerce products as a project tech lead at Shopify * [Audrey]( is a product leader, having scaled e-comm businesses like Peloton internationally and developed fintech solutions to enable commerce in emerging markets. Zimi is named after Audrey’s great-grandmother, a quick-witted entrepreneur who successfully kept food on the table as a single mom through her export business. With Zimi, we’re creating opportunities for hustlers like Audrey’s great-grandmother by making global commerce more accessible to everyone. **🙏Our Ask** If you know any businesses in South Asia, Latin America, or Africa that are selling to North America and/or Europe, please let us know ([[email protected]](mailto:[email protected])) or schedule some time with us [here]( #### YC Sign Photo

Hestus, Inc.

hestus-inc

https://www.hestus.co

San Francisco, CA, USA

Hestus is AI-powered CAD software that speeds up hardware development. While engineers are typically quick at coming up with initial design concepts, they are bogged down getting from idea to the final design due to countless tedious and repetitive tasks. We are building CAD software that automates these mundane tasks and speeds up the design execution process, while letting engineers spend more time on the creative part of the design process.

An AI-powered CAD software

B2B

B2B -> Engineering, Product and Design

S24

Active

Early

https://www.ycombinator.com/companies/hestus-inc

https://yc-oss.github.io/api/batches/s24/hestus-inc.json

Title: Hestus, Inc.: An AI-powered CAD software | Y Combinator URL Source: Markdown Content: ### An AI-powered CAD software Hestus is AI-powered CAD software that speeds up hardware development. While engineers are typically quick at coming up with initial design concepts, they are bogged down getting from idea to the final design due to countless tedious and repetitive tasks. We are building CAD software that automates these mundane tasks and speeds up the design execution process, while letting engineers spend more time on the creative part of the design process. Hestus, Inc. Founded:2024 Team Size:2 Location:San Francisco ### Active Founders ### Kevin Chu, Founder Kevin is leading the engineering and product as the CTO at Hestus (S24). Previously, he led Autonomous Vehicle engineering at Cruise, growing the team from 20 to 300 over seven years and holds a Robotics PhD from UCLA. He also worked in multiple robotics startups including energy and medical spaces ### Sohrab Haghighat, Founder Sohrab is a 2x YC founder, currently running Hestus, Inc (S24). Prior to that, Sohrab was founder and CEO of SpaceRyde (W19), a space transportation company, and ran the company for 6 years before selling the IP. Before starting SpaceRyde, Sohrab was an early employee of Cruise and ran Planning and Controls Teams. Sohrab holds a PhD in Aerospace Engineering, has published 13 patents and 23 technical papers.

DeepSim, Inc.

deepsim-inc

https://www.deepsim.io

Mountain View, CA, USA

DeepSim is building an AI physics simulator. We are currently developing the only thermal simulator to meet AI chip design needs and are validating our tool with Intel. We are a team of three electrical engineering PhDs from Stanford with backgrounds in semiconductor fabrication and design.

An AI physics simulator for AI chip design

B2B

B2B -> Engineering, Product and Design

S24

Active

Early

https://www.ycombinator.com/companies/deepsim-inc

https://yc-oss.github.io/api/batches/s24/deepsim-inc.json

Title: DeepSim, Inc.: An AI physics simulator for AI chip design | Y Combinator URL Source: Markdown Content: ### An AI physics simulator for AI chip design DeepSim is building an AI physics simulator. We are currently developing the only thermal simulator to meet AI chip design needs and are validating our tool with Intel. We are a team of three electrical engineering PhDs from Stanford with backgrounds in semiconductor fabrication and design. DeepSim, Inc. Founded:2020 Team Size:3 Location:Mountain View ### Active Founders ### Connor McClellan, Founder Stanford EE PhD grad now building an AI physics simulation platform at DeepSim to solve the frustrating parts of my PhD. I'm also really into espresso making and will be happy to make you your best cup of coffee. ### Alexander Gabourie, Founder Alex is a co-founder of DeepSim, Inc., where he develops AI physics simulators. He earned his PhD in Electrical Engineering from Stanford, specializing in nanoscale thermal simulations, and brings over a decade of experience in the semiconductor industry. ### Chuck Koroglu, Founder Stanford EE PhD with background in semiconductor engineering, and physics modeling of integrated circuits. Excited about transforming the landscape of engineering simulations with new technology that combines physics, experimental data and AI. ### Company Launches [### DeepSim – AI physics simulations]( **TLDR:** AI-accelerated 3D physics simulator for engineers [DeepSim]( makes it easy for engineers to run 3D physics simulations fast. This means rapid design iterations for better products and real-world monitoring for better decisions. We built a new physics simulation platform that leverages AI physics models for very fast, ultra-scalable physics simulations. — Hi everyone! We are [Connor,]( [Alex,]( and [Chuck,]( and we love physics simulations. While working on our PhDs at Stanford, we saw how physics simulators needed a refresh to solve modern problems. That’s why we started [DeepSim.]( The Problem ----------- Running 3D physics simulations is **difficult**, **expensive**, and **slow**. 1. **Setup:** Experienced engineers still spend countless hours meshing their designs before even running a single simulation. 2. **Run:** Many simulations take hours to even weeks on expensive compute systems. Some are even intractable and require highly trained engineers to simplify the problems at a significant cost of accuracy. 3. **Attention:** Each step requires constant babysitting as simulators are fragile and can fail at any time. 🧑‍🔧️ Engineers hate the tedious simulation setup and wasted time due to failed simulations. 👔 Managers hate that engineers can’t go faster and use tons of cloud credits. 💰 CEOs/VPs hate that engineering delays stall a product launch. The Solution ------------ Leverage AI to **simplify**, **speed up**, and **scale up** 3D physics simulations. We are building a new AI physics simulation platform, automating away the physics setup and simulating designs 1000X faster with no drop in accuracy. Our platform uses lightweight, easy-to-train AI models combined with a powerful GPU-accelerated solver to deliver solutions quickly and efficiently. This approach eliminates the bulkiness of traditional finite element method (FEM) tools and overcomes the inflexibility of other AI physics simulators. At [DeepSim]( we have already worked with semiconductor companies to build the most powerful thermal simulator for circuit design, running thermal simulations with billions of nodes in minutes on a single GPU, which is 1000X more detail than a commercial tool can deliver using 50+ CPU cores in the same amount of time. With DeepSim's simulator, we are able to model extremely complex geometries with length scales spanning 6 orders of magnitude, intractable to solve with FEM tools. Below is a thermal simulation of an integrated circuit, complete with a heat sink and airflow, that only took 10 minutes to run. We can resolve hot spots in individual transistors (~10 nm in size) within a whole chip (1 cm). Use Cases --------- This is a breakthrough in 3D physics simulations that empowers engineers in the same way FEM did 50 years ago. We’re excited to see where engineers take our tool! Some initial use cases are: * Speeding up the design of complex systems with faster simulations * Utilizing design automation to speed up development * Developing digital twins for system operation The Team -------- Chuck, Alex, and Connor founded DeepSim to solve the frustrating parts of their PhDs. The Ask ------- We would love to talk with engineers that… * design and operate battery packs for EVs or grid-scale energy storage * design and deploy satellites * work on thermal management of semiconductors, from the chip scale to data centers * have any complaints about slow or frustrating simulations Contact us at [[email protected]](mailto:[email protected])! #### YC Sign Photo

Helium

helium

https://tryhelium.com

San Francisco, CA, USA

Helium is building self-improving software — AI that continuously experiments on key moments in the user funnel to boost revenue. Our first customer, a social app with 20M MAU, uses Helium to test new in-app paywalls to get more users to subscribe. Zach previously made $100M+ in growth experiments at Uber, Shishir led the first LLM features for Alexa, and Anish worked at Meta building ML to automatically experiment on ads, increasing Instagram ads conversion by 10%.

Self Improving Software

B2B

B2B -> Engineering, Product and Design

S24

Active

Early

https://www.ycombinator.com/companies/helium

https://yc-oss.github.io/api/batches/s24/helium.json

Title: Helium: Self Improving Software | Y Combinator URL Source: Markdown Content: ### Self Improving Software Helium is building self-improving software — AI that continuously experiments on key moments in the user funnel to boost revenue. Our first customer, a social app with 20M MAU, uses Helium to test new in-app paywalls to get more users to subscribe. Zach previously made $100M+ in growth experiments at Uber, Shishir led the first LLM features for Alexa, and Anish worked at Meta building ML to automatically experiment on ads, increasing Instagram ads conversion by 10%. Helium Founded:2024 Team Size:3 Location:San Francisco ### Active Founders ### Zach Witzel, Founder Made \>$100M in growth experiments at Uber. Now building tryhelium.com to personalize upsell flows. We help consumer subscription companies double their conversion. ### Anish Doshi, Founder Cofounder @ Helium. Previously ML @ Meta, Apple, Alteryx, Berkeley. Into 🥊 ,♟️, and 🎹 ### Shishir Jakati, Founder Cofounder and CTO of Helium. Before Helium I worked at Amazon Alexa building predictive models to determine the quality of content and utterances spoken by Alexa, reducing cloud costs 10x. I love ML, LLMs, and MLOps - definitely reach out if I can help! ### Company Launches [### 🎈 Helium: Improve your paywall at the speed of thought]( . **TLDR** -------- [Helium]( lets mobile companies launch and personalize upsell flows without code. We previously made \>$100M in conversion experiments at Uber and Meta and are building the tools we wish we had. With Helium, growth teams can add new paywalls or experiments in minutes! Helium also helps drive conversion with SmartSelect, our AI that finds the best paywall experience for each user. If you know anyone working on consumer apps, [give us a shout]( If we can’t help them earn 5x our price in new revenue, we’ll give their money back. . **👋 We’re Helium!** -------------------- We’re building tools to experiment on in-app paywalls without code. * [Zach]( (CEO) worked on in-app upsell at Uber. He ran dozens of conversion experiments that made \>$100M. * [Anish]( (CPO) worked in ad optimization at Meta. He drove 10% in Instagram ad conversion through AI to personalize ads to users. * [Shishir]( (CTO) worked on Alexa AI, using ML / LLMs to make Alexa smarter for millions of users. Early on we realized how dumb it was that Zach spent years doing crude in-product personalization at Uber, while Anish built tools to automatically personalize tens of thousands of ad variants. While ads have come a long way, software upsell is stuck in the past — it’s too slow, too manual, and too one-size-fits-all. . **👴📟📼 Mobile Paywalls are stuck in 2008** -------------------------------------------- In 2023, mobile in-app purchases crossed $150 billion, with $45B from subscriptions. Mobile companies we talked to, from Headspace to Hinge, spend ~1/3 of eng bandwidth on upsell, but optimizing in-app conversion is still highly manual. Teams spend months experimenting with paywall entry points, messaging, UX, targeting, premium features, and discounting. The problem is that… * **⏰ Each change takes weeks**: Because paywalls live in native mobile code, making a change takes at least 6 weeks to code, test, launch to the app store, get it approved, and then get installs of the new version. * **🫨 Upsell becomes a maze**: There are typically dozens of entry points (onboarding, profile, feed, modals, etc) leading to different experiences. One prospect had 87 separate paywalls, so rolling out even small changes turns into weeks of manual work. * **🆘 Personalization is hard:** Personalization drives conversion (1 customer saw +40% conversion personalizing copy) but takes a ton of work. Most companies struggled with identifying key segments, designing custom experiences, implementing new logic in code, and dealing with many-variant experiments. . **🎈 Introducing, Helium** -------------------------- Helium helps consumer app companies launch, manage, and improve paywalls without code. It allows growth or product teams to run tests on their own (and stop bothering engineers) as they work on upsell conversion. * **🖥️ Visual editor**: Add new paywalls or multi-step flows all from the editor. All you need is one line of code to wire up your upsell flow with our SDK (React Native, Swift, & Android) * **🔬 Experimentation**: Add test copy, imagery, UI, or more, and run an experiment in 3 clicks. Results populate in your analytics tool * **👥 Personalization made simple**: Add as many variants as you want, and our AI algorithm (Smart Select) will find the right one for each user to drive conversion. Solve problems like… * What CTA works best in each language? * What feature do we emphasize for each customer segment? * Which paywall do we show on onboarding vs app open? . **😍 Ask: Intros to consumer mobile folks 😍** ---------------------------------------------- If you know anyone who leads growth or monetization at a consumer mobile company, we’d love to talk to them to see if we can help! We’re doing a design partner deal for the next few months… * White glove support for your implementation (or we can integrate Helium into your codebase directly for you) * We’ll run 3-5 experiments of your choice on your behalf * If we don’t make you 5x times our price in additional revenue, we’ll offer you a refund **If you know anyone who may be a fit, you can email us at** [**[email protected],**](mailto:[email protected]) or just [**click here.**](

Entangl

entangl

https://www.entangl.ai/

San Francisco, CA, USA

Entangl is automating engineering design. Starting with datacenter design, our AI agent automatically detects errors while engineering teams work. For each error detected, Entangl suggests a detailed step-by-step solution. 20,000 datacenters are being built every year, making this a massive market. We are also actively training our next AI agent that can autonomously implement solutions (think code completion but for hardware engineering).

Automating hardware design end-to-end without errors

B2B

B2B -> Engineering, Product and Design

S24

Active

Early

https://www.ycombinator.com/companies/entangl

https://yc-oss.github.io/api/batches/s24/entangl.json

Title: Entangl: Automating hardware design end-to-end without errors | Y Combinator URL Source: Markdown Content: ### Automating hardware design end-to-end without errors Entangl is automating engineering design. Starting with datacenter design, our AI agent automatically detects errors while engineering teams work. For each error detected, Entangl suggests a detailed step-by-step solution. 20,000 datacenters are being built every year, making this a massive market. We are also actively training our next AI agent that can autonomously implement solutions (think code completion but for hardware engineering). Entangl Founded:2024 Team Size:2 Location:San Francisco ### Active Founders ### Shapol M, Founder Co-founder & CEO at Entangl. Previously co-founded and led the Karman Space Programme, launching one of the most powerful British reusable rockets and growing it into Europe's best-sponsored program. MEng in Aeronautical Engineering from Imperial College London. ### Antanas Zilinskas, Founder Co-founder and CTO at Entangl. Previously led the engineering of a 25km reusable student rocket. Also co-founded an AI research lab at university focusing on humanoids, as well as trajectory predictions and flight readiness monitoring for rockets. Research experience at Tokyo Tech, simulating aerotrains using ANNs. ### Company Launches [### Entangl - Automating engineering design 🚀]( **tl;dr:** We automate the detection of design errors in engineering projects and suggest solutions autonomously—saving time, money, and, potentially, lives. — Hey folks, [Shapol]( and [Antanas]( here. We used to lead a reusable rocket program and oversaw the launch of four missions. We hated engineering design software so much that we created our own — which is why we’re here today :) **🧨 The Problem: Design errors get caught too late, resulting in delays or dangerous outcomes.** Organizations carry out design reviews every few weeks/months to ensure they catch errors early. As aerospace engineers, we've experienced firsthand the painful reality of cross-team design reviews. Teams grind for weeks to prepare for these meetings, yet costly errors still slip through to manufacturing or even operation (think doors blowing off planes). The issue? Engineers work in silos, unaware of how their decisions impact the broader system. For example, changing the O-ring material might unexpectedly impact the seal integrity at low temperatures when building a space shuttle booster rocket (see NASA Space Shuttle Disaster below ⬇️). 🎉 **The Solution:** **Autonomously detecting errors across an organization.** ------------------------------------------------------------------------------ Humans are terrible at dealing with large amounts of data and understanding how everything interacts with one another. Machines, on the other hand, are great! They can crawl across all the knowledge bases (GitHub, Google Drive, OneDrive, etc.) and detect changes as they are made. Our platform identifies design problems autonomously across engineering projects. It then suggests targeted solutions to the right engineers and delivers daily insights and reviews, preventing expensive problems early on (poor engineering cost Boeing $100B+). This saves each engineer two months of work every year and makes engineering safer. 💲**Opportunity: Safer and quicker engineering** ------------------------------------------------ Engineers don’t need to spend two months a year grinding for design reviews. Your engineering projects don’t need to be delayed by an average of seven months. Prevent errors early to move faster and save your company’s runway. **👋** **Ask: How you can help** * Connect us to engineering leaders in your network. [Click here](mailto:[email protected]). * Give us product feedback. Book a [15-minute demo here]( * Sign up today! [Click here]( #### Company Photo ### Hear from the founders #### How did your company get started? (i.e., How did the founders meet? How did you come up with the idea? How did you decide to be a founder?) We met whilst designing & launching rockets. Whilst doing that, we realised that we could automate a lot of our engineering design work with graph neural networks and transformer models. \\Then we wondered if we could develop an AI system that can design any mechanical system (i.e. a car, rocket, etc). So, we started Entangl. We are creating an the smartest engineer ever! #### What is your long-term vision? If you truly succeed, what will be different about the world? We are creating an AI system that can design anything. Ambitious engineering projects—like space colonies and humanoid robots—haven't materialized yet. It's not due to a lack of money but because it's extremely difficult to assemble enough talented people to work together on them. Our aim is to transform this from being limited by human resources to something that is limited by computing power.

Mineflow

mineflow

http://mineflow.ai/

Mineflow is an AI platform for mineral exploration. We generate predictions for mineral deposit shapes and locations, empowering mining companies from initial exploration to advanced feasibility studies. Geologists need to know precisely where their deposits are because drilling is expensive. In early trials with a lithium mining company, we found that Mineflow predicts the shape of hard rock lithium deposits more than an order of magnitude more accurately than our competitors. Ryan received a BS in Artificial Intelligence from Carnegie Mellon's School of Computer Science where he helped teach both the grad-level Deep Learning course and the grad-level Search Engines course. He built ML models at Google in the Display Ads optimization org for 2 years and launched products that generated ~125m USD in ARR.

Mineflow predicts the shape and location of mineral deposits

Industrials

Industrials

S24

Active

Early

https://www.ycombinator.com/companies/mineflow

https://yc-oss.github.io/api/batches/s24/mineflow.json

Title: Mineflow: Mineflow predicts the shape and location of mineral deposits | Y Combinator URL Source: Markdown Content: ### Mineflow predicts the shape and location of mineral deposits Mineflow is an AI platform for mineral exploration. We generate predictions for mineral deposit shapes and locations, empowering mining companies from initial exploration to advanced feasibility studies. Geologists need to know precisely where their deposits are because drilling is expensive. In early trials with a lithium mining company, we found that Mineflow predicts the shape of hard rock lithium deposits more than an order of magnitude more accurately than our competitors. Ryan received a BS in Artificial Intelligence from Carnegie Mellon's School of Computer Science where he helped teach both the grad-level Deep Learning course and the grad-level Search Engines course. He built ML models at Google in the Display Ads optimization org for 2 years and launched products that generated ~125m USD in ARR. ### Jobs at Mineflow San Mateo, CA, US / Remote (US) $80K - $150K Any (new grads ok) Mineflow Founded:2024 Team Size:1 Location: ### Active Founders ### Ryan Goggins, Founder studied AI at CMU in SCS, built ML models for Google Ads, now trying to predict the shape of mineral deposits underground ### Company Launches [### ⛏️ Mineflow - AI mineral exploration]( * * * Hey YC! I'm Ryan from [Mineflow]( Quick backstory: I've always been fascinated by how mining companies discover deposits. That passion led me to leave Google after two years of working on deep learning for ads to start Mineflow. I’m convinced that mineral exploration is as much a statistics problem as it is a geology problem, and early feedback from lithium mining companies suggests I might be on the right track! ### The problem Accurately predicting deposit shapes is critical for geologists because each drill hole can cost tens of thousands of dollars. However, **traditional computational models fall short** because they: 1. Can't incorporate highly multimodal data 2. Don’t learn from data across multiple exploration sites 3. Rely on naive algorithms for their inferences As a result, mining companies often **drill thousands of holes** before an exploration site becomes operational, driving up costs. ### The solution Mineflow lets geologists **upload any dataset from their exploration sites**. Then, Mineflow **builds a custom AI model to predict the shape and location of mineral deposits**. By training a deep learning model on the whole dataset, Mineflow is able to make predictions with **an order of magnitude of accuracy more than our competitors**. As a result, Mineflow can significantly reduce the number of drill holes needed, saving mining companies millions in exploration costs. We can produce predictions in 2 dimensions (think overhead, bird' s-eye view) and 3 dimensions (think 3D grid of cubes). 3D gold deposit we are modeling: 2D prospectivity map we produced for an exploration project: ### Team [Ryan]( earned a BS in Artificial Intelligence from Carnegie Mellon SCS and built ML models for the Google Display Ads optimization team for two years. While at Google, he launched products and deep learning models that generated ~$125m USD of ARR. ### Ask If you know anyone in mineral exploration or oil and gas exploration, I'd love to [connect]( with them to see how Mineflow can help. Grab space on my calendar [here]( #### YC Sign Photo #### Company Photo ### Hear from the founders #### What is your long-term vision? If you truly succeed, what will be different about the world? The discovery rate of economically viable mineral deposits near the earth's surface is declining each year. As the "low-hanging fruit" is claimed and mined, we’ll increasingly need to drill deeper to retrieve critical minerals like gold, copper, and lithium. While mining companies today can afford to drill hundreds of holes to precisely model a deposit at a depth of 200 meters, the cost of drilling will rise sharply as we go deeper. This means fewer holes can be drilled, necessitating more powerful deep learning models to accurately predict the shape and location of these deeper mineral deposits. Mineflow will lead the way in developing a family of foundation models for predicting the earth's subsurface composition and structure. Ultimately, I hope every mining company will use Mineflow's models to guide their exploration with precision. ### Selected answers from Mineflow's original YC application for the S24 Batch #### Describe what your company does in 50 characters or less. AI-driven geological modeling.

Kart AI

kart-ai

https://kart.ai

San Francisco, CA, USA

Kart transforms the product discovery experience for the most innovative e-commerce companies. At a fraction of the cost and time, we build customized AI shopping assistants like Amazon Rufus, trained to answer any question on your catalog.

The AI shopping assistant for all e-commerce brands

B2B

B2B -> Engineering, Product and Design

S24

Active

Early

https://www.ycombinator.com/companies/kart-ai

https://yc-oss.github.io/api/batches/s24/kart-ai.json

Title: Kart AI: The AI shopping assistant for all e-commerce brands | Y Combinator URL Source: Markdown Content: ### The AI shopping assistant for all e-commerce brands You might have seen that Amazon recently launched Rufus, an AI conversational shopping assistant that’s everywhere on their online store. Every retailer now wants the same technology on their site - and that’s what we enable. Kart AI Founded:2024 Team Size:2 Location:San Francisco ### Active Founders ### Rez Havaei, Founder Rez is the Co-Founder and CEO @ Kart AI. Rez dropped out of his Master's program at CMU to move to SF and build Kart. Previously he worked at Jane Street as a trader, as well as NVIDIA and Cohere as an AI researcher. He also helped build an LLM evaluation platform and benchmark at vals.ai as the founding engineer. ### Gordon Chi, Founder Gordon is the co-founder & CTO @ Kart AI. Previously graduated with a MS in Computer Science at Stanford, where he was the first author of two AI papers and part of the winning 2024 Stanford Alexa Prize Team. Previously worked at SIG and TikTok, contributing to the insides of an ecommerce operation that makes over $7B / year. ### Company Launches [### Kart AI: The AI shopping assistant for all e-commerce brands]( **TL;DR** Kart transforms the product discovery experience for the most innovative e-commerce companies. At a fraction of the cost and time, we build customized AI shopping assistants like Amazon Rufus, trained to answer any question on your catalog. **Ask** We would love: * any leads to brands or e-commerce shops that would benefit from a better assistant * stores with multiple brands are awesome, i.e., Best Buy * we are mainly in the electronics/furniture/fashion space, but happy to help any store * follow us on [Linkedin]( **Problem: E-Commerce Search is changing. Amazon Rufus is coming.** Top brands, like Amazon, are all investing in sales assistants on their websites. This is no coincidence: ChatGPT and Perplexity have shown how valuable chatbots can be for search. AI shopping assistants have been shown to boost conversion rates by 15%. Product search is just one sector, and a very large one at that. A poll conducted by Anderson Consulting showed that 62% of online customers would buy more products if live chat and support were available. A survey by Invesp shows that chatters tend to spend 60% more per purchase. (Source: The Harris Poll, 2023. You can read more [here]( and [here]( **Solution: We build your product search assistant!** Until now, human sales assistants have always been coveted but too expensive to maintain for non-luxury brands. Now, we build agents for your online store for a fraction of the cost. Our team was formerly employed at Cohere and Tiktok Shop, competed in International Linguistics Olympiads, and won the recent Amazon Alexa Prize Social Chatbot challenge. We held a demo on [ for electronics and helped 10,000+ users with their latest electronic purchases. Here’s what our assistant can do for you: 1. **Give product recommendations:** Trained on your catalog, we know what’s popular and other trending products across the industry. We help customers find exactly what they need, keeping our recommendations as objective as possible while maximizing sales conversion rate. 2. **Answer non-product questions:** The current search is bad with nuanced questions. We handle all these cases: your customers can now ask questions ranging from _best father’s day gifts_ to a _summary of reviews for the latest samsung oled monitor._ Whenever you are ready to transform your online store, please contact us at [[email protected].](mailto:[email protected]) We can’t wait to transform the future of shopping with you. ### Other Company Launches ### Kart AI – Your new electronics shopping assistant We help you find and compare electronic products. [Read Launch ›](

TradeFlow

tradeflow

https://www.tradeflowai.com/

New York, NY, USA

TradeFlow is AI-powered settlement for financial securities. During her time in financial services, Taryn saw firsthand how inefficient and costly it is to settle trades. Historically, it’s been difficult to improve this process because it’s heavily reliant on manual emails, disparate systems, and inconsistently structured data. However, the advent of powerful AI technology allows us to unlock greater automation than ever before. TradeFlow empowers back offices to do more with less by offloading these time-intensive tasks to AI.

AI-powered settlement for financial securities

Fintech

Fintech

S24

Active

Early

https://www.ycombinator.com/companies/tradeflow

https://yc-oss.github.io/api/batches/s24/tradeflow.json

Title: TradeFlow: AI-powered settlement for financial securities | Y Combinator URL Source: Markdown Content: ### AI-powered settlement for financial securities TradeFlow is AI-powered settlement for financial securities. During her time in financial services, Taryn saw firsthand how inefficient and costly it is to settle trades. Historically, it’s been difficult to improve this process because it’s heavily reliant on manual emails, disparate systems, and inconsistently structured data. However, the advent of powerful AI technology allows us to unlock greater automation than ever before. TradeFlow empowers back offices to do more with less by offloading these time-intensive tasks to AI. TradeFlow Founded:2024 Team Size:2 Location:New York ### Active Founders ### Taryn Reardon, Founder Taryn is the Co-Founder and CEO of TradeFlow. She has expertise in financial services, with experience in electronic trading, middle office, and digital transformation roles. Prior to TradeFlow, Taryn worked at Goldman Sachs, where she managed institutional client relationships to drive bottom-line results. Taryn graduated from Cornell University with a B.S. in Policy Analysis & Management. ### Liam Breen, Founder Liam is the Co-Founder and CTO of TradeFlow. Previously, he was a software engineer at Meta, where he drove critical privacy initiatives for Facebook and built a data mining tool to identify malicious entities on the platform. Liam graduated from Cornell University with a B.S. in Computer Science. He is also a graduate of Phillips Exeter Academy. ### Company Launches [### TradeFlow: AI software to automate stock trade settlement]( [**TradeFlow**]( uses AI to automate stock trade settlement for financial firms. **About Us** ------------ [Taryn]( and [Liam]( met at Cornell University 7 years ago. After graduating, Taryn worked at Goldman Sachs in global markets, and Liam worked at Meta as a software engineer. **The Problem** --------------- During her time in financial services, Taryn saw firsthand how inefficient and costly it is to settle trades. Trade settlement requires extensive coordination across firms, yet the current financial infrastructure is extremely fragmented. The industry acknowledges the inefficiency, but it’s been difficult to improve because workflows are heavily reliant on manual emails, disparate systems, and inconsistently structured data. To make matters worse, the US just shortened the settlement timeline from 2 days to only 1, forcing operations professionals to perform the same workload in half the time. **The Solution** ---------------- The advent of powerful AI technology can unlock greater efficiencies than ever before. Leveraging LLMs, TradeFlow digests and centralizes data from all key sources and automates action items, like booking trades and writing emails. Instead of completing time-intensive tasks by hand, operations professionals can offload them to TradeFlow and simply review the work at the end. We empower back offices to do more with less, so trades can settle faster with a fraction of the human effort. **Our Ask** ----------- * Let us know at [[email protected]](mailto:[email protected]) if you have any financial services leaders in your network (brokers, banks, institutional investment firms)! Here is a blurb that you can share with others: _“An ex-Goldman sales trader and an ex-Meta engineer created _[**_TradeFlow_**]( an AI-powered software to automate post-trade operations. Reach out to the team at _[[email protected]_](mailto:[email protected])_ to learn more & schedule a demo.”_ * Follow us on [**LinkedIn**]( #### YC Sign Photo

MagiCode

magicode

https://www.magicode.ai/

Stanford, CA, USA

MagiCode is a realtime code review system for AI-generated code. MagiCode acts as the communication channel between any AI code-writing tool and the human developer by offering realtime feedback and suggestions, helping to ensure the generated code is bug-free, context-aware, and production-ready.

Realtime code review for AI generated code

B2B

B2B -> Engineering, Product and Design

S24

Active

Early

https://www.ycombinator.com/companies/magicode

https://yc-oss.github.io/api/batches/s24/magicode.json

Title: MagiCode: Realtime code review for AI generated code | Y Combinator URL Source: Markdown Content: ### Realtime code review for AI generated code MagiCode is a realtime code review system for AI-generated code. MagiCode acts as the communication channel between any AI code-writing tool and the human developer by offering realtime feedback and suggestions, helping to ensure the generated code is bug-free, context-aware, and production-ready. MagiCode Founded:2024 Team Size:2 Location:Stanford, CA ### Active Founders ### Jeff He, Founder Before founding MagiCode, Jeff earned a Master’s degree from Stanford, where he contributed to the Stanford AI Lab, and an undergraduate degree from UC Berkeley, where he was part of the Berkeley AI Research Group (BAIR). He has published four papers on self-driving cars and reinforcement learning in top conferences like NeurIPS and ICRA. Previously, Jeff also worked as an SDE & a quant, where he developed his expertise in AI and machine learning. ### Anirudh Sriram, Founder Before founding MagiCode, I earned a Master’s degree from Stanford University, where I contributed to the Stanford AI Lab and Stanford Vision and Learning Lab. I also hold an undergraduate degree from IIT Madras. I have published six papers on large language models (and NLP in general) in prestigious conferences such as NeurIPS, ACM, and ACL. Prior to MagiCode, I gained extensive experience through my work at Google Research, Microsoft, Walmart Labs, etc. ### Company Launches [### MagiCode - Proactive AI Pair Programmer]( **TLDR:** [MagiCode]( is a **proactive** AI pair programmer that intuitively understands developers' actions in real-time, actively anticipates their next steps, and offers assistance even before they ask. **Who are we? 👥** We are [Anirudh]( and [Jeff]( co-founders of [MagiCode]( We met at Stanford when we were looking for teammates to take the Lean Launchpad course together. Between us, we have had the opportunity to work at Google Research, Microsoft, Amazon, Stanford AI Lab (SAIL), and Berkeley AI Research (BAIR). Both of us have worked a lot with self-driving cars and LLMs during our university days and are stoked to be working at the intersection of both fields. **Why are we building this? ⚙️** We were active users of various AI coding tools even before they were named “pair programmers”. However, the experience was underwhelming because: 1. they are **not really “intelligent”**: we always had to type out multiple shortcuts for our queries to be routed to the right AI agent to handle. 2. when we were deeply immersed in a task/implementation, we often didn’t know **how, or what** to ask the AI Coding tool. For example, when implementing a button that triggers a POST/ request to the backend API, we have to constantly jump around multiple files, making modifications across different functions. In this scenario, before asking the AI to help us with implementation, we actually have to spend time thinking about “how” to frame our questions: * How do we explain what the functionality of this button is? * Which file names should we mention to the AI as being relevant? * How can we succinctly phrase all relevant context, including current progress or past edits/changes that we have made, so that the AI will give us the best output? As a result, we often resorted to doing everything ourselves to avoid the tedious back-and-forth. In those moments, we realized what we wanted was a tool that could **fetch global context** without relying on shortcut commands, anticipate our needs, and even understand what we were thinking before typing anything— something that was always **“on the same page”** with us. That’s exactly what led us to create MagiCode— a tool that can answer all the developer’s questions intelligently, and also **proactively assist** without needing to be prompted. **How are we different from existing solutions? 🤔** * **Proactive Assistance:** We bring the answers to you without you even asking. MagiCode is designed to constantly analyze the project's state and make suggestions for what you can do next. As a user, all you have to do is click on a suggestion and watch as MagiCode works on files asynchronously and brings you the finalized results with easy-to-view diff files. * **Personalized Memory:**  We have personalized memory retainment of your editing history, editor state, and user activity. We designed an intelligent system that processes all this information in a low-latency manner and uses it naturally when answering the user’s queries. ### **Watch MagiCode in action! 🤯** [ ### **Our ask 🙏** We are currently onboarding users for MagiCode. * If you’d love to try MagiCode, we are active on **Discord** and would be happy to help you onboard and answer any questions. You can join our discord here: [ * If you know an engineer who is interested in trying out AI Coding tools, tell them about MagiCode! * You can also reach out to us directly at [[email protected]](mailto:[email protected]) to talk more about the future of AI-powered coding. Feel free to book a time with us here: [ 😊 Let’s build with Magic!

Storia AI

storia-ai

https://storia.ai

Millbrae, CA, USA

With AI increasingly automating away code generation, software engineers will spend more time reading, judging, and architecting code rather than writing it. Storia is building an open-source copilot that knows a company's codebase and its context. We are starting with Sage, a Perplexity-like agent for helping developers understand, judge, and generate software. Given an existing codebase, developers can ask Sage questions such as: 1) Given my project’s SLA and latency constraints, what is the appropriate underlying vector database to use? How would I incorporate it into my existing codebase? 2) Why should I pick Redis over Milvus as my underlying vector store? 3) Does this codebase in our organization still work and what steps are required for a complex integration with another library? Sage’s answers are directly supported by documentation and external references like GitHub, Stack Overflow, technical design documents, and project management software, preventing hallucinations. Today, Sage has up-to-date knowledge about open-source repositories (indexed daily). Tomorrow it will have a deep understanding of every line of code on the Internet. For teams, Sage will know about your private codebase too. No group has yet solved how to build an AI system that comprehends a codebase and its context and can empower every developer to architect better code, faster. This requires new research advances because vanilla RAG and out-of-the-box LLMs aren’t going to cut it. We have 20+ years of software engineering and AI research experience. Julia worked on precursors of Gemini using contextual neural techniques before they were called “RAG” (and applied it to products like Google Keyboard and Pixel phones). Mihail built the earliest LLMs at Amazon Alexa and launched the first contextual deep learning conversational AI system in production at Alexa.

Open source AI copilot that knows your company's code and its context

B2B

B2B -> Engineering, Product and Design

S24

Active

Early

https://www.ycombinator.com/companies/storia-ai

https://yc-oss.github.io/api/batches/s24/storia-ai.json

Title: Storia AI: Open source AI copilot that knows your company's code and its context | Y Combinator URL Source: Markdown Content: ### Open source AI copilot that knows your company's code and its context With AI increasingly automating away code generation, software engineers will spend more time reading, judging, and architecting code rather than writing it. Storia is building an open-source copilot that knows a company's codebase and its context. We are starting with Sage, a Perplexity-like agent for helping developers understand, judge, and generate software. Given an existing codebase, developers can ask Sage questions such as: 1) Given my project’s SLA and latency constraints, what is the appropriate underlying vector database to use? How would I incorporate it into my existing codebase? 2) Why should I pick Redis over Milvus as my underlying vector store? 3) Does this codebase in our organization still work and what steps are required for a complex integration with another library? Sage’s answers are directly supported by documentation and external references like GitHub, Stack Overflow, technical design documents, and project management software, preventing hallucinations. Today, Sage has up-to-date knowledge about open-source repositories (indexed daily). Tomorrow it will have a deep understanding of every line of code on the Internet. For teams, Sage will know about your private codebase too. No group has yet solved how to build an AI system that comprehends a codebase and its context and can empower every developer to architect better code, faster. This requires new research advances because vanilla RAG and out-of-the-box LLMs aren’t going to cut it. We have 20+ years of software engineering and AI research experience. Julia worked on precursors of Gemini using contextual neural techniques before they were called “RAG” (and applied it to products like Google Keyboard and Pixel phones). Mihail built the earliest LLMs at Amazon Alexa and launched the first contextual deep learning conversational AI system in production at Alexa. Storia AI Founded:2023 Team Size:2 Location:Millbrae, CA ### Active Founders ### Mihail Eric, Founder I'm a cofounder of Storia. I've been an AI researcher/engineer for 12+ years. Some of my accomplishments: 1) built the first deep learning dialogue models at the Stanford NLP Group, 2) published AI research papers that have 2000+ citations, 3) built the first LLMs at Amazon Alexa AI, 4) launched the first deep-learning-based dialogue system in production at Alexa, and 5) founded and sold a machine learning education startup used globally in industry and academia. ### Julia Turc, Founder Cofounder @ Storia & previously at Google Research. 10+ years in NLP. Built the first bidirectional RNN for machine translation at Oxford. At Google, I helped build pivotal technologies like federated learning, the AI layer on top of Android, and RAG for precursors of Gemini. Excited about how NLP is now redefining software engineering. ### Company Launches [### Storia: A contextual AI pair programmer for developer teams]( Hi everyone. We’re Julia and Mihail, the team behind [Storia]( We have 20+ years of software engineering and AI research experience. [Julia]( worked on precursors of Gemini using contextual neural techniques before they were called “RAG” (and applied it to products like Google Keyboard and Pixel phones). [Mihail]( built the earliest LLMs at Amazon Alexa and launched the first contextual deep-learning conversational AI system in production at Alexa. **👎 The Problem: <5% of software teams use AI code understanding and generation systems** Despite all the hype around AI coding following the release of GitHub Copilot, a disappointingly small percentage of engineering teams have actually adopted AI assistants into their developer workflows. Why? What developers don’t like about systems today: 1. **Code search is not accurate enough**, slowing dev velocity. 2. **Code assistants are not grounded in their team’s codebase**, so they lack the necessary context to produce useful and high-quality answers. 3. **Code outputs can be incorrect** due to hallucination which strains dev patience and trust. **🚀 The Solution: A contextual AI pair programmer built for any team’s codebase** We’re building [Sage]( a Perplexity-like agent for helping developers understand, analyze, and generate software. Given an existing codebase, developers can ask Sage questions such as: * _For my project’s SLA and latency constraints, what is the appropriate underlying vector database to use? How would I incorporate it into my existing codebase?_ * _Why should I pick Redis over Milvus as my underlying vector store?_ * _Does this codebase in our organization still work, and what steps are required for an integration with another library?_ A few of the features Sage supports: * **We index repos daily**, providing users with the most up-to-date info about how to use a library (no stale, deprecated API definitions), * We have a Perplexity-like focus on **citing the source of every generated line of code**, enhancing developer trust. * We've developed algorithms grounded in research for **syntactic and semantic understanding of a codebase** so that we can provide more relevant context and use it effectively given a user query. We are now actively working with seed to series C partner companies to help them integrate Sage into their developer workflows. **🌅 The Opportunity: Making every engineer on a software team a 10x developer** There are ~28M developers worldwide. Github Copilot is the frontrunner in AI assistant tooling and reached $100M ARR in 2 years, which is only a small percentage of the $913B spent on software development in 2023. **🙏 Our Ask** We’re still onboarding a closed group of beta partners to integrate with Sage. We would appreciate connecting with engineering leaders (VP of engineering, Head of Engineering, CTO)  at Seed-Series C companies. Additionally, if you want your open-source repository indexed reach out to us at [[email protected].](mailto:[email protected]) #### YC Sign Photo #### Company Photo

CardLift

cardlift

https://withcardlift.com

Mountain View, CA, USA

💸 CardLift builds co-branded browser extensions to help credit card issuers maximize their share of wallet (and in turn, per user revenue). 💸 💼 Built by the serial extension founder behind TravelArrow (the world's largest travel browser extension, 300k users, 1.5M+ ARR) and OctoShop (acquired by NYSE:IBTA), CardLift is backed by YCombinator as well as prominent fintech founders (CTO @ Ramp, Founder @ Imprint, Founder @ Mulberry). How it works - We build banks and credit card issuers browser extensions to help them capture a larger share of wallet among their cardholders by - 🏃 Migrating Existing Spend - When a credit card program onboards a new user, we make it dead simple for these users to switch their card-on-file at all of their top online merchants to their new card with one click. - Our AI-powered browser extension not only automatically detects where our partners' users spend the most (Netflix, Amazon, T-mobile, Progressive), but also lets them switch out their primary payment instantly. Users press a button in their browser and their card-on-file everywhere instantly switches to their shiny new card (our partner's card!). 😌 Making Checkout Seamless - Our fast checkout technology embeds “One Tap Checkout ” on the checkout page at over 50,000 e-commerce sites - making our partners' cards the easiest way for users to complete any transaction on their browser. - They can even customize the copy to tell the user why to use your card for that specific purchase. For example Chase could prompt their users to use their Sapphire Preferred card on Avis BECAUSE it offers car rental insurance. 🤑 Making Our Partners' Cards More Rewarding - We partner with over 1000 of the country’s top merchants to offer additional affiliate-based cash back that our partners can tack on to their existing cash back. User can now earn double or even triple the cash back on Sephora when they user our partners' “One Tap Checkout” button instead of their other cards.

We help card issuers maximize their share of wallet.

Fintech

Fintech

S24

Active

Early

https://www.ycombinator.com/companies/cardlift

https://yc-oss.github.io/api/batches/s24/cardlift.json

Title: CardLift: We help card issuers maximize their share of wallet. | Y Combinator URL Source: Markdown Content: ### We help card issuers maximize their share of wallet. 💸 CardLift builds co-branded browser extensions to help credit card issuers maximize their share of wallet (and in turn, per user revenue). 💸 💼 Built by the serial extension founder behind TravelArrow (the world's largest travel browser extension, 300k users, 1.5M+ ARR) and OctoShop (acquired by NYSE:IBTA), CardLift is backed by YCombinator. How it works - We build banks and credit card issuers browser extensions to help them capture a larger share of wallet among their cardholders by - 🏃 Migrating Existing Spend - When a credit card program onboards a new user, we make it dead simple for these users to switch their card-on-file at all of their top online merchants to their new card with one click. - Our AI-powered browser extension not only automatically detects where our partners' users spend the most (Netflix, Amazon, T-mobile, Progressive), but also lets them switch out their primary payment instantly. Users press a button in their browser and their card-on-file everywhere instantly switches to their shiny new card (our partner's card!). 😌 Making Checkout Seamless - Our fast checkout technology embeds “One Tap Checkout ” on the checkout page at over 50,000 e-commerce sites - making our partners' cards the easiest way for users to complete any transaction on their browser. - They can even customize the copy to tell the user why to use your card for that specific purchase. For example Chase could prompt their users to use their Sapphire Preferred card on Avis BECAUSE it offers car rental insurance. 🤑 Making Our Partners' Cards More Rewarding - We partner with over 1000 of the country’s top merchants to offer additional affiliate-based cash back that our partners can tack on to their existing cash back. User can now earn double or even triple the cash back on Sephora when they user our partners' “One Tap Checkout” button instead of their other cards. CardLift Founded:2024 Team Size:2 Location:Mountain View ### Active Founders ### Rithwik Pattikonda, Founder Founder at CardLift (YC S24, ZFellows) | Prev Founded TravelArrow and OctoShop (acq by Ibotta) ### Company Launches [### CardLift - Achieve top-of-wallet status for your credit card program]( ### **TL;DR** [CardLift]( builds co-branded browser extensions to help credit card issuers maximize their share of wallet (and, in turn, per-user revenue). ### How It Works We build your credit card program a co-branded browser extension. When a user signs up for your card, you can incentivize them to add your extension to their browser. Your users are now 2 - 4x more valuable. **Migrate Existing Spend** When your credit card program onboards a new user, we make it dead simple for them to switch their card-on-file at all of their top online merchants to your card with one click. Our AI-powered browser extension not only automatically detects where your users spend the most (Netflix, Amazon, T-mobile, Progressive), but also lets them switch out their primary payment method to your card with one click. Users press a button in their browser, and their card-on-file everywhere instantly switches to their shiny new card (your card!). Offer your users an incentive to make this switch, and watch your per-user GMV transaction volume skyrocket even more. **Ensure You Continue To Capture Future Spend** We don’t just help when a user signs up to your card program. Our fast checkout technology embeds “One Tap Checkout ” on the checkout page at over 50,000 e-commerce sites - making your card the easiest way for users to complete any transaction on their browser. You can even customize the copy to tell the user why to use your card for that specific purchase. For example, if you were Chase, you could prompt your users to use their Sapphire Preferred card on Avis BECAUSE it offers car rental insurance. **Make Your Card Program More Rewarding** For the final cherry on top, we partner with over 1,000 of the country’s top merchants to offer additional affiliate-based cash back that you can add to your existing cashback. Your users can now earn double or even triple the cash back on Sephora when they use your “One Tap Checkout” button instead of their other cards. **Full Demo** - [** ### About Us CardLift is founded by Rithwik Pattikonda - serial browser extension entrepreneur who has built some of the world’s largest browser extensions - OctoShop (acquired by NYSE:IBTA) and TravelArrow (300,000+ users). At CardLift, his goal is to bring some of the learnings and insights from his previous consumer browser extensions to financial institutions.

ideate.xyz

ideate-xyz

https://ideate.xyz

San Francisco, CA, USA

ideate.xyz is the first AI graphics design co-pilot for content marketers. It’s like an easier-to-use Canva with a built-in AI designer. Users create designs in our editor and chat with our virtual design expert so they can iterate on the work without having to hire an agency.

AI graphics design co-pilot for content marketers

B2B

B2B -> Engineering, Product and Design

S24

Active

Early

https://www.ycombinator.com/companies/ideate-xyz

https://yc-oss.github.io/api/batches/s24/ideate-xyz.json

Title: ideate.xyz: AI graphics design co-pilot for content marketers | Y Combinator URL Source: Markdown Content: ### **Tl;dr:** **ideate.xyz is an AI design co-pilot for content teams to create engaging thumbnail designs for their social media channels. We are starting with thumbnails, the most critical window to connect with audiences - ideate is here to help content teams stand out.** **Looking ahead, ideate will become THE AI design partner for the next generation of content creators, crafting unique and tailored designs for their bespoke brands.** ### **The team:** Hi, [Jasper]( and [Elliott]( here. We are building [ideate.xyz]( - the first AI design co-pilot for content teams to create thumbnail designs for social media channels. **Jasper Jia (Left)** Cofounder & CEO at ideate.xyz. Designer specialising in creative tools and no-code. He believes designers should embrace learning and getting their hands dirty in the technical end of the software building process to craft the best software experience and cultivate fast iteration cycles with the engineering team. He was an ex-Snap designer and MSc Design Engineering graduate from Imperial College London. **Elliott Thompson (Right)** Cofounder & CTO at ideate.xyz. A software engineer with a penchant for visual computing. He studied computer graphics at University College London before taking on a role as a graphics generalist at Snap Inc. He has previously worked on 3D engines, game developer tooling, graphics editors, and e-commerce platforms. ### **The problem:** Design tools are stuck at two extremes: powerful but complex, or simple yet limiting. This seemingly inevitable tradeoff affects millions of content creators trying to stand out with their thumbnail designs, the critical window to connect with their audience. If you’ve had to design thumbnails for social channels, you've likely been here: You have a great idea, so you install the latest Adobe product. Suddenly, you're faced with a blank canvas and a dizzying array of options. Hours of YouTube tutorials later, you're ready to throw in the towel. Time for Plan B. You load up Canva, pick a template that's "close enough," and tweak some colors. It doesn't really feel like yours, and it is so hard to stand out with a “Canva look” when there are over 3.7 million videos uploaded to YouTube alone every day! ### **Our solution:** [ideate.xyz]( - the first AI design co-pilot orchestrating the entire thumbnail design process from initial ideas to final designs that stand out! \* Design with plain English: No prior experience with design or creative software needed. Produce designs via simple and engaging discussions with AI design co-pilot. \* Think of our copilot as your team’s AI design partner - guiding the design process with proactive questions to understand design requirements and implementing expert design decisions after your sign off. \* Collaborative environment: Teams can work with a co-pilot on the same project for a quick feedback cycle and sign-off before exporting the final designs. \* Maintain brand consistency as our AI learns your preferences over time and always generates content with this in mind. \* We fine-tune our prompts for each generation, considering layout composition, image assets, and copywriting in lockstep with each other to produce creative but also consistent results. A typical workflow with ideate.xyz: 1\. Describe your content ideas 2\. First drafts designed by AI co-pilot 3\. Iterate and create more variations by chatting with AI copilot 4\. Export designs in channel-specific dimensions [Quick demo here]( ### **Our story:** As a design and engineer duo, we have spent years crafting creative tools in the WebGL and AR space. We've also built and mentored in the no-code space to help non-techies build out their online presence. Regardless of the domain, one challenge persists for creators — finding a design tool that’s both simple and powerful enough to bring a brand's vision to life. The reason why this problem persists in the market is simple - the inevitable tradeoff that every design software needs to make between power and ease of use. Typically, the more powerful a tool gets, the more complex it becomes. On the flip side, easy-to-use tools often limit creativity with rigid templates. But with AI, we have found the conviction to break this tradeoff. We are on a mission to create a design tool that actually puts your ideas in the driving seat, making the design process less about esoteric design language and more about the content team’s own creative impulses. Our goal is to empower the next generation of creators to build stunning brands and assets. ### **Our ask** Are you currently on a team that regularly has to come up with thumbnail designs for your social media content?  Either wrestling with complex design tools or unhappy with generic templates? Talk to us, and we would love to learn more about your requirements! If you know any professional content creators, we’d really appreciate the connection! Our email is [[email protected]](mailto:[email protected])

Apten

apten

https://www.apten.ai/

San Francisco, CA, USA

Apten is an AI SMS agent that automates sales and marketing for consumer services businesses. We help businesses build customized agentic conversation flows to drive higher lead conversion and personalized messaging.

Customizable AI SMS agents for businesses

B2B

B2B -> Marketing

S24

Active

Early

https://www.ycombinator.com/companies/apten

https://yc-oss.github.io/api/batches/s24/apten.json

Title: Apten: Customizable AI SMS agents for businesses | Y Combinator URL Source: Markdown Content: ### Customizable AI SMS agents for businesses Apten is an AI SMS agent that automates sales and marketing for consumer services businesses. We help businesses build customized agentic conversation flows to drive higher lead conversion and personalized messaging. Apten Founded:2023 Team Size:2 Location:San Francisco ### Active Founders ### Daniel Ho, Founder Daniel is the Co-founder & CEO @ Apten. Previously a PM at Tesla building systems to automate the detection and chargeback of supplier issues ($XXM in cost recovery / year). Built supply chain analytics at Amazon, Lucid Motors, and Walmart. Studied Economics + Data Science at UC Berkeley. ### Roshan Kumaraswamy, Founder Roshan is the Co-founder & CTO at Apten. Before Apten, Roshan spent three years at AWS building sales planning applications & data platforms to empower the AWS Sales org. Before that, he ideated & built multi-data source capabilities for Flux (language) at InfluxData (YC W13) as one of their first interns, and studied EECS at UC Berkeley for 3 years. ### Company Launches [### Apten 💬📲 - AI SMS agents]( **TLDR:** We help B2C businesses automate sales, marketing, and customer service through SMS text. Customize Apten, and it’ll **proactively reach out, respond at any time, and never forget to follow up.** Hey everyone - we’re [Daniel]( & [Roshan]( the founders of [Apten]( 👋 **The Problem** --------------- Businesses currently rely on humans or traditional SMS automation (drip texting) to qualify and nurture leads. This applies to both inbound leads (e.g., ad -\> website -\> fills out contact/quote form and leaves phone number) and the re-engagement of existing leads in a CRM. Example industries where this happens: real estate, insurance, home services **The Solution** ---------------- An AI SMS assistant that qualifies and nurtures leads in a **personable way** while responding 24/7 and automatically following up. [** Customize: * Apten’s knowledge (FAQs, info about your business, specific instructions/rules) * Specific conversation flows and follow-up schedules for **different types of leads** Apten won’t just chat - it’ll **take actions** in a way a human would to help your business by proactively: * Categorizing/tagging leads based on the conversation * Creating notifications based on conditions you define * Determining the appropriate time to follow-up based on a lead's response We’ll provide white-glove onboarding and support the whole way! **Our Ask** ----------- If you’re a business that: * Collects phone numbers from leads * Employs virtual assistants or a contact center * Leverages drip texting campaigns or other SMS automation We’d love to [talk to you!]( You can also email us at [[email protected]](mailto:[email protected])

Bayesline

bayesline

https://bayesline.com

New York, NY, USA

Bayesline is building a GPU-powered financial analytics suite for institutional investors, like hedge funds. Clients use our cloud deployed solution to build custom analytics in seconds, having previously been frustrated with off the shelf solutions that don't generate any alpha. What used to take weeks or months now happens near instantly. Sebastian, CFA (ex Bloomberg Quant & AI Research) and Misha, PhD (ex BlackRock Managing Director) met while working at BlackRock in 2016. With a combined 20 years of experience in the space they set out to change the way the industry uses analytics - from one size doesn't fit all to truly tailored to clients' needs.

Highly customizable and blazingly fast analytics for asset managers.

Fintech

Fintech -> Asset Management

S24

Active

Early

https://www.ycombinator.com/companies/bayesline

https://yc-oss.github.io/api/batches/s24/bayesline.json

Title: Bayesline: Highly customizable and blazingly fast analytics for asset managers. | Y Combinator URL Source: Markdown Content: ### Highly customizable and blazingly fast analytics for asset managers. Bayesline is building a GPU-powered financial analytics suite for institutional investors, like hedge funds. Clients use our cloud deployed solution to build custom analytics in seconds, having previously been frustrated with off the shelf solutions that don't generate any alpha. What used to take weeks or months now happens near instantly. Sebastian, CFA (ex Bloomberg Quant & AI Research) and Misha, PhD (ex BlackRock Managing Director) met while working at BlackRock in 2016. With a combined 20 years of experience in the space they set out to change the way the industry uses analytics - from one size doesn't fit all to truly tailored to clients' needs. Bayesline Founded:2024 Team Size:2 Location:New York ### Active Founders ### Sebastian Janisch, Founder Building blazingly fast financial analytics at Bayesline. Previously quant at Bloomberg and BlackRock building investment models running 4T in assets. ### Misha van Beek, Founder Misha van Beek is a co-founder at Bayesline. Before Bayesline, Misha was a Managing Director at BlackRock where he led innovation on Aladdin’s portfolio risk models spanning tens of trillions in assets. He also led the research on Aladdin’s economic scenario engine, as well as investment models that run roughly $400 billion in strategic asset allocations across all major asset classes. He holds a PhD in Financial Mathematics from the University of Amsterdam. ### Company Launches [### Bayesline - Blazingly fast analytics for Investment Managers]( **TL:DR;** We figured out how to fit custom equity factor risk models in seconds instead of weeks. And we can do this across the entire financial analytics landscape. Ad hoc thematic factor construction (think COVID), risk decomposition through time, optimized factor selection, all on the fly and in seconds. Try out our demo [here]( Why does it matter? ------------------- We each spent nearly a decade at BlackRock and Bloomberg, building products, leading research teams, and empathizing with frustrated clients. They sit at the cutting edge of investment research, coming up with market-beating strategies. **Yet they have to make do with a 20-year-old analytics toolkit** that current vendors offer. We are changing that. If the AI community can fit a 300 billion parameter model, there’s no reason why a factor risk model should take a week to compute. In fact, it doesn’t. **We figured out how to do it live and in seconds, and we can do it across the entire analytics landscape**. Who is behind this? ------------------- Hi, we’re Sebastian and Misha. [**Sebastian**]( CFA** built his expertise in quant research during his time as a Director in BlackRock’s Financial Modeling Group where he implemented and researched equity risk models that analyze trillions in assets. Prior to Bayesline, he was at Bloomberg, where he incubated the next generation of customizable and actionable quant products as part of the Quant & AI Research group. A computer scientist by training with [M.Sc]( in Finance and a passion for quant research, Sebastian spent the last 10 years leveraging the power of machine learning to challenge, innovate, and reshape how institutions think about financial modeling. [**Misha**]( PhD** was among the youngest Managing Directors at BlackRock. He headed the portfolio risk research team that evolved Aladdin’s portfolio risk models across all asset classes. He also headed the team that developed Aladdin’s economic scenario engine and investment models that manage roughly $400 billion in strategic asset allocations. Misha has spent the past 10 years coupling his professional quant training with his personal interest in all things AI and hands-on engineering. Our Ask ------- * **Try our our demo** [here]( (only email and name required) * **Share this post!** Please help spread the word, as you never know who it may help * **Connect us** with quants, portfolio, and risk managers. \* blurb to copy & paste: _The guys at Bayesline are ex-MD level BlackRock and Bloomberg. They are building a new class of analytics for asset managers, starting with equity factor risk and thematic factor construction. They can build fully customized and properly aligned models (custom universe, factors, exposure overrides, etc.) in seconds, far superior to what’s currently available. They deploy on the client cloud, ingest existing vendor data, and provide their engine to compute fully-fledged new risk models and analytics – all on the fly._ #### YC Sign Photo #### Company Photo ### Hear from the founders #### How did your company get started? (i.e., How did the founders meet? How did you come up with the idea? How did you decide to be a founder?) We met in 2016 at BlackRock, and lived together during Covid. We both have a very strong mission to revolutionize financial analytics. We thought that the world’s most renowned institutions would be the best place to do that. But it turns out that they move too slowly and are not ready to truly innovate, especially in the age of AI. Leaving our lucrative careers was the only way to pursue our dreams. #### What is your long-term vision? If you truly succeed, what will be different about the world? All of financial analytics was built in the 90s on fundamentally an outdated tech stack (C++ on CPUs). The AI community has developed a new stack (Python ML packages on GPUs) that is orders of magnitude faster, both in time-to-market and run-time. Our aim is to rebuild all of financial analytics on this stack; from economic scenario simulation engines for banks and insurers to portfolio risk models for hedge funds. In the future, Bayesline will be integrated in all financial institutions, eliminating the days of waiting for calculation that tens of thousands of industry professionals have to deal with.

Bucket Robotics

bucket-robotics

https://bucket.bot

Pittsburgh, PA, USA

250M lbs of plastic is wasted annually in US manufacturing. Bucket Robotics is making defect detection faster and easier to deploy to prevent that. Our products are designed to be easy to integrate into existing automation and technology stacks, flexible enough to bring retooling costs down, and reliable enough to make you forget that it's a robot. Matt and Steph are leveraging their experience from the self-driving industry, where we've integrated high-performance computing and quality sensing into complex, noise-filled environments. We focus on robust hardware and user-friendly software for a new generation of end-users, ensuring ease of use and rapid deployment. Secure and efficient handling of user data can make customers in the manufacturing space nervous to invest in sensing — our team's extensive experience handling self-driving car data in these regulated environments positions us uniquely to tackle these challenges. Companies like Keyence, National Instruments, and FLIR offer either outdated hardware or prohibitively expensive systems, along with software that hasn’t evolved to meet modern user needs. US manufacturing is undergoing a significant transformation, with construction spending hitting a record annualized pace of over $200 billion. The need for a more reliable supply chain has seen over $700 billion in manufacturing megaprojects across North America since 2021, and this growth requires heavy automation. Bucket Robotics is at the forefront of this shift, offering a scalable, cost-effective solution that enhances automation and positions manufacturers closer to their end customers.

Defect detection for manufacturing built from CAD and synthetic data.

Industrials

Industrials -> Manufacturing and Robotics

S24

Active

Early

https://www.ycombinator.com/companies/bucket-robotics

https://yc-oss.github.io/api/batches/s24/bucket-robotics.json

Title: Bucket Robotics: Defect detection for manufacturing built from CAD and synthetic data. | Y Combinator URL Source: Markdown Content: ### Defect detection for manufacturing built from CAD and synthetic data. 250M lbs of plastic is wasted annually in US manufacturing. Bucket Robotics is making defect detection faster and easier to deploy to prevent that. Our products are designed to be easy to integrate into existing automation and technology stacks, flexible enough to bring retooling costs down, and reliable enough to make you forget that it's a robot. Matt and Steph are leveraging their experience from the self-driving industry, where we've integrated high-performance computing and quality sensing into complex, noise-filled environments. We focus on robust hardware and user-friendly software for a new generation of end-users, ensuring ease of use and rapid deployment. Secure and efficient handling of user data can make customers in the manufacturing space nervous to invest in sensing — our team's extensive experience handling self-driving car data in these regulated environments positions us uniquely to tackle these challenges. Companies like Keyence, National Instruments, and FLIR offer either outdated hardware or prohibitively expensive systems, along with software that hasn’t evolved to meet modern user needs. US manufacturing is undergoing a significant transformation, with construction spending hitting a record annualized pace of over $200 billion. The need for a more reliable supply chain has seen over $700 billion in manufacturing megaprojects across North America since 2021, and this growth requires heavy automation. Bucket Robotics is at the forefront of this shift, offering a scalable, cost-effective solution that enhances automation and positions manufacturers closer to their end customers. Bucket Robotics Founded:2024 Team Size:2 Location:Pittsburgh, PA ### Active Founders ### Matt Puchalski, Founder/CEO Matt is co-founder and CEO of Bucket Robotics. Before co-founding Bucket Robotics, Matt distinguished himself in the autonomous vehicle industry through pivotal roles at Stack AV, Latitude AI, and Argo AI. His work at these companies brought the deployment of autonomous vehicles to public roads across the United States and Germany. With deep expertise in integrating and operationalizing AVs, Matt brings a unique vision to revolutionize the robotics industry. ### Steph Wolski, Founder/CTO Steph is co-founder and CTO of Bucket Robotics. Steph began college at the age of 13 and started working in robotics at 18. He was the Remote Guidance Tech Lead at Argo AI where he met Matt. Later, Steph was an early engineer at Stack AV, working on the Autonomy Integration team where he stood up core infrastructure for their self-driving trucks. ### Company Launches [### Bucket Robotics 🤖 Vision for Manufacturing]( ☕ **_TL;DR_** ------------- [_Bucket Robotics_]( makes defect detection models way faster and easier to deploy. CAD goes in — detection model comes out. **👥** **The Team** ------------------- ### Hi everyone, we’re [Matt]( and [Steph]( and we’re Bucket Robotics! **📉 The Problem** ------------------ Defects are a universal challenge in manufacturing, particularly with methods like injection molding, die casting, and 3D printing. Factories using these methods frequently retool their lines for different products making it difficult and costly to gather enough data to train defect detection models. Manufacturers strive to ensure that only the highest quality products reach their customers. Avoiding the shipment of defective products not only protects their reputation but also safeguards expensive machinery from damage, maximizes yield by leveraging advanced analytics, and prevents the progression of faulty parts through the production line, which can lead to compounded waste and increased costs. A 2017 study estimates that [yield losses account for 44% of sheet metal used]( in the production of passenger vehicles — and we want to bring that number down, saving money and the environment. Traditionally, defect detection comes in two flavors: 1. Pay a person to watch your line. 2. Train a vision model the old-fashioned way — wait for defective parts to come out and add your sample to the dataset. Paying people to check parts is inefficient and a miserable job. Collecting enough labeled data samples to train a model is impractical for rapidly retooled outputs, and the labeling is expensive. This often leaves manufacturers with a critical gap in quality assurance, leading to increased scrap, downtime, and potentially unhappy customers with defective parts. **📈 The Solution** ------------------- We help eliminate the need for inefficient human monitoring and outdated defect detection by providing a system where we generate defect detection models using CAD files—something every part designed for modern manufacturing has. Our models can run on an edge computing device as small as a Jetson Orin Nano and be continually refined with real-world data. ### **_Here’s why this is a game-changer:_** * **Early Deployment:** Your defect detector can arrive at your manufacturing line before you’ve manufactured any parts. This means you can catch defects right from the get-go. * **Continuous Improvement:** As defects start to appear, we can automatically refine and improve the model, ensuring your system remains top-notch. * **Classification**: The system tells you how the part is defective, allowing you to dial in your tooling configuration and maintenance cycle—e.g. when it's time to create a new mold. Check out a video on our process here: [Bucket Robotics | Defect Detection]( **🙏How You Can Help** ---------------------- We are seeking connections and collaborations to drive our mission forward. If you or anyone in your network is interested in learning more or helping out, we want to hear from you! We’re specifically looking to connect with: * Quality Assurance and Factory Automation leaders * Companies and brands in high-volume and precision manufacturing * Experts in the manufacturing and defect detection industries If that’s you or someone you know, [please get in touch!]( #### YC Sign Photo #### Company Photo

ProhostAI

prohostai

https://www.prohost.ai/

San Francisco, CA, USA

ProhostAI is building an AI-powered operating system to automate guest communications and hosting operations for vacation rental hosts. We are on a mission to make every host a 5-star host. Our founding team consists of data scientists and engineers from Airbnb and Dropbox.

AI-powered operating system for vacation rental hosts

B2B

B2B

S24

Active

Early

https://www.ycombinator.com/companies/prohostai

https://yc-oss.github.io/api/batches/s24/prohostai.json

Title: ProhostAI: AI Property Manager for Airbnb & Short-Term Rental Hosts | Y Combinator URL Source: Markdown Content: ### AI Property Manager for Airbnb & Short-Term Rental Hosts ProhostAI is building an AI Property Manager for short-term rental hosts on Airbnb and other platforms. Our team consists of an ex-Airbnb Staff Data Scientist and Dropbox Sr Software Engineer who are also Superhosts with 9 properties worth $15M making $1.5M/yr. ### Jobs at ProhostAI San Francisco, CA, US / Remote (US) $100K - $200K 0.10% - 1.00% 3+ years San Francisco, CA, US / Remote $100K - $200K 0.10% - 1.50% 3+ years San Francisco, CA, US / Remote (US) $100K - $200K 0.10% - 1.50% Any (new grads ok) ProhostAI Founded:2024 Team Size:2 Location:San Francisco ### Active Founders ### Bill Ulammandakh, Founder Co-founder and CEO at ProhostAI. Previously Staff Data Scientist at Airbnb, Etsy, Thumbtack and Wayfair. Harvard undergrad. ### Andrew Lawson, Founder Co-founder and CTO at ProhostAI. Previously Software Engineer at Dropbox, JWPlayer, and others. ### Company Launches [### ProhostAI - AI property manager for Airbnb hosts]( [**ProhostAI**]( **is an AI property manager assistant for hosts on Airbnb and other booking platforms.**  We run 9 Airbnb properties as Superhosts (4.99+ rating) worth $15M, and our AI automations have already cut our own costs by 30% and automated 70% of guest communications. **Our key features:** * **AI for Guest Communications:** AI agent drafts responses to every guest message * **AI for Maintenance Ticketing & Cleaning Management** * **AI for Revenue Generation** with automated guest upsells All of this with a first class mobile app, because most hosts manage their properties from their phones! ### Our Founding Team [**Bill**]( * **Bill was a Staff Data Scientist, Algorithms at Airbnb for 6+ years, building AI & ML products** * 3 AI patents for products implemented at Airbnb that drove hundreds of millions in revenue * 9+ years total in data science (Etsy, Wayfair, Thumbtack) * Undergrad from Harvard [**Andrew**]( * **Andrew was a Sr Software Engineer at Dropbox for 6+ years,** working on their API Platform handling millions of requests per second * 9+ years as a software engineer We are Airbnb hosts with 9 properties, 4.99+ ratings across 1000+ guest reviews, and $1.5M/yr revenue. We've been hosts for 5+ years. ### Why we are building ProhostAI * **70% of things a host has to do can be automated with no intervention today** * Guest communications, cleaning coordination & communication, and maintenance ticket management are repetitive and simple tasks ripe for automation with LLMs and AI agents with a high-quality mobile-first experience. * **80%+ of hosts use no SaaS tools yet** but are ready to professionalize their operations with better tools. * **Airbnb hosts are digitally native, working from their mobile phones all the time**, and highly likely to adopt new technologies to improve their operations. * **SaaS for hosts is a $5B market** with 10 million short-term rental listings globally! **🚀Asks** ---------- 1. **If you are a host, join our waitlist at** [**prohost.ai**]( or shoot me an email at [[email protected]](mailto:[email protected])! 2. **Share our website with others hosts** and have them join our waitlist! 3. **Follow us on **[**LinkedIn**]( and **[**X**]( 4. **If you are using Hostaway or Guesty for Pros to manage your properties,** we can get you set up right away! Sign up for a free trial [**here**]( or [**book a demo**](

Miru

miru-ml

https://www.miruml.com

San Francisco, CA, USA

Miru makes it easy for engineers to deploy and monitor software on robots and IoT devices. We provide the infrastructure to deploy updates over-the-air to a fleet of devices by simply pushing new changes to GitHub.

Deployment infrastructure for edge computing

B2B

B2B -> Infrastructure

S24

Active

Early

https://www.ycombinator.com/companies/miru-ml

https://yc-oss.github.io/api/batches/s24/miru-ml.json

Title: Miru: Deployment infrastructure for robotics and IoT | Y Combinator URL Source: Markdown Content: ### Deployment infrastructure for robotics and IoT Miru makes it easy for engineers to deploy and monitor software on robots and IoT devices. We provide the infrastructure to deploy updates over-the-air to a fleet of devices by simply pushing new changes to GitHub. Miru Founded:2024 Team Size:3 Location:San Francisco ### Active Founders ### Armel Talla, Founder/CEO Armel graduated from UT Austin with a B.S. in Computer Science and Mathematics. He previously founded Clips AI, an AI video dev tool that automatically repurposed long videos into clips. During college, Armel interned at Google and YouTube, focusing on building internal tools and infrastructure. ### Vedant Nair, Founder Vedant is the Co-Founder and COO of Miru. He graduated from UT Austin, where he studied Business and Computer Science. He previously founded Clips AI, an AI video developer tool that automatically repurposed long videos into clips. During college, he interned at Bain & Co. and EY-Parthenon, doing GTM consulting for software companies. ### Benjamin Smidt, Founder Ben is the Co-Founder and CTO of Miru. He holds a B.S. in Electrical and Computer Engineering from UT Austin, where he initially spent two years studying Mechanical Engineering, igniting his passion for hardware and robotics. Prior to Miru, Ben founded Clips AI, where he developed production-ready AI models and video processing servers using Kubernetes. ### Company Launches [### Miru - Deployment infrastructure for Robotics and IoT 🦾]( **tl;dr:** Miru helps robotics and IoT teams painlessly deploy software over the air, provision devices, and manage versioning for their fleets. For engineers using Miru, _updating software on thousands of devices is as easy as pushing a change to GitHub._ **Our Ask:** Engineers in robotics and IoT, [get started with Miru today!]( **❌ The Problem** Robotics and IoT companies don’t have robust, off-the-shelf DevOps tools. As a result, to deploy their software, they’re forced to manually SSH/SCP into their devices or build a hacky solution internally. **But building this infrastructure is costly.** These companies often need to dedicate more than half of their software team to build and maintain their custom internal tools. This balloons engineering costs and decreases product velocity. It’s one of the reasons **why many of these companies die.** **🦾 Miru** Miru helps engineers build and maintain their deployment pipelines, saving them months of work. With Miru, you can: 1. Push changes to GitHub, and let Miru handle building, staging, and deploying your software to your devices across the world 2. Manage devices and track software versioning on a centralized dashboard 3. Enjoy obsessive, fast, 24/7 customer support 🫡 **Installation is easy.** If your device already uses Ubuntu, the Miru installation is just one command **🪄** 🧍🏼🧍🏿🧍🏾 **About Us** Hi 👋🏾 \- we’re [Vedant]( [Armel]( and [Ben]( After meeting at our freshman dorm at UT Austin, we instantly clicked and became best friends. Whether we were playing intramural basketball or scaling our previous startup, [Clips AI]( to 7000 MAUs in just three months, we've proven we make an exceptional team. Outside of Clips AI, Armel worked as a software engineer at Google and YouTube, focusing on building internal tools and infrastructure. Vedant brings experience in software GTM consulting from Bain & Co. and EY-Parthenon. Ben, our hardware fanatic, has a background in mechanical and electrical engineering, and he’s built production-ready AI and video processing infrastructure using Kubernetes. We’re passionate about a future where robots can free humans from doing the dull, dirty, and dangerous tasks that dominate billions of people’s lives. We founded Miru to provide the infrastructure to accelerate this transition. We’re obsessed with solving this problem. If you’re a software engineer in robotics or IoT who is struggling with deployment, we’d LOVE to make your life easier. Just reach out at \[vedant\] @ miruml \[dot\] com, or better yet, [start using Miru today](

Redouble AI

redouble-ai

https://redouble.ai/

San Francisco, CA, USA

Companies are increasingly adopting LLMs to automate their workflows, but sufficiently accurate and consistent outputs are hard to obtain, especially in regulated industries where mistakes are costly. As a result, AI companies use human reviewers in production to verify and correct the output of their AI pipelines - which quickly becomes an unsustainable bottleneck. Our solution dynamically learns from our clients' unique domain specific human feedback data; provides recommendations whether to send the output of a LLM pipeline to a human for review; monitors the insights obtained from human reviewers at scale, while also flagging suspicious reviews to ensure consistent final outputs; and integrates easily with our clients' existing pipelines via a small number of simple API calls.

Scale human-in-the-loop for AI workflows in regulated industries

B2B

B2B

S24

Active

Early

https://www.ycombinator.com/companies/redouble-ai

https://yc-oss.github.io/api/batches/s24/redouble-ai.json

Title: Redouble AI: Scale human-in-the-loop for AI workflows in regulated industries | Y Combinator URL Source: Markdown Content: ### Scale human-in-the-loop for AI workflows in regulated industries Redouble AI is the solution to scale human-in-the-loop for AI workflows in regulated industries. Instead of having to rely on slow and expensive human reviewers, Redouble’s clients save 80% of staff costs and significantly improve the accuracy and quality of their outputs by integrating our tool into their workflows. Addressing an $11BN market, the company is already generating revenue with its first set of clients. The founders are two serial entrepreneurs (who have built successful companies in AI and health-tech respectively) and a data engineering veteran who has written more than 120 enterprise-grade software applications. Redouble AI Founded:2024 Team Size:3 Location:San Francisco ### Active Founders ### Martin Bittner, Founder/CEO Martin Bittner is the Co-Founder & CEO of Redouble AI. Before that, he was the CEO of Arctoris, a pharma workflow automation company which he co-founded and led from inception to profitability. In terms of background, Martin is a physician-scientist, having obtained his MD from the University of Freiburg in Germany, and his DPhil in Oncology as a Rhodes scholar at the University of Oxford. ### Haotian Li, Founder Haotian received his training in physics and computational chemistry from Zhejiang University and New York University. In 2017 Haotian cofounded the AI drug discovery start-up Redesign Science as CTO, designed and built the internal computational pipeline that increased the high through put screening efficiency by 100X. ### Andrey Santrosyan, Founder Specializing in devising and executing holistic data and software strategies for drug discovery. Acumen in capturing, storing, and securing research data, constructing analysis platforms to facilitate swift decision-making and rapid optimization cycles. Led the development of over 120 scientific applications, contributing to the successful market introduction of seven pharmaceutical products. ### Company Launches [### Redouble AI: Quality control for AI in regulated industries]( **TL;DR:** [Redouble]( is _the_ solution to scale human-in-the-loop for AI workflows in regulated industries. **The Problem** --------------- Companies are increasingly adopting LLMs to automate their workflows, but sufficiently accurate and consistent outputs are hard to obtain, especially in regulated industries where mistakes are costly. As a result, AI companies are using human reviewers in production to verify and correct the output of their AI pipelines. This human review step quickly becomes a critical bottleneck for exponential growth: 1. Even if the output is accurate 95% of the time, there's no way to tell which 5% is wrong, so the human has to review **_all_** outputs. 2. The cost of human review increases linearly with the pipeline throughput 3. Even the best reviewers make mistakes, especially due to fatigue, when having to identify similar mistakes repeatedly. 4. Domain-specific workflows require domain expert reviewers who are hard to find and expensive to hire. Production-level quality control in regulated industries is an inherently different problem from model performance evaluation, since one cannot optimize the prompts to cover all the edge cases nor prevent all the unacceptable variations in LLM output. **Our Solution** ---------------- Redouble AI is purpose-built to operate within these _“high stakes, few data points”_ areas. To address the problems above, our solution: 1. Dynamically learns from your unique domain-specific human feedback data. 2. Provides recommendations on whether to send the output of your LLM pipeline to a human for review. 3. Monitor the insights from your human reviewers at scale while also flagging suspicious reviews to ensure consistent final outputs. 4. Integrate easily with your existing pipeline with just a couple of simple API calls. In addition, we also surface succinct, actionable, and granular insights that can be used to further optimize your AI workflow. So you can continually improve your pipeline’s performance as you scale. We understand that human-in-the-loop quality control is critical if you are working with AI in the legal, healthcare, insurance, or finance sectors. We are on a mission to ensure that AI can be safely used in these mission-critical spaces without breaking the bank. **Our Ask** ----------- We are looking for companies that provide LLM-based services in regulated industries and use human reviewers. If you use human reviewers in your AI workflow, or if you know people who do, we would greatly appreciate any introductions and will offer any help that we can in return. Schedule a demo on [Redouble.ai]( or reach us through [[email protected]](mailto:[email protected]) **The Team** ------------ [Martin]( is a serial entrepreneur, Rhodes scholar, and MD-PhD by training. In his last venture as co-founder and CEO, he built a profitable tech-enabled life sciences company and raised over $15M in venture funding. [Haotian]( is a serial entrepreneur, engineer, and physicist by training. In his last venture as co-founder and CTO, he built an AI-driven biotech company and raised over $17M in venture funding. [Andrey]( is a software and data engineer and mathematical statistician by training. During his 25 years at Novartis and multiple startups, he has built more than 100 applications across software and data infrastructure.

Formula Insight

formula-insight

https://www.formulainsight.io/

New York, NY, USA

Formula Insight gives hedge funds a central repository of their financial models and lets them track changes in their forecasts so they can make better investment decisions.

Enable Hedge Funds to easily track, query & analyze financial models

B2B

B2B

S24

Active

Early

https://www.ycombinator.com/companies/formula-insight

https://yc-oss.github.io/api/batches/s24/formula-insight.json

Title: Formula Insight: Help Hedge Funds manage, track & query their Excel models | Y Combinator URL Source: Markdown Content: ### Help Hedge Funds manage, track & query their Excel models Formula Insight gives hedge funds a central repository of their financial models and lets them track changes in their forecasts so they can make better investment decisions. Formula Insight Founded:2024 Team Size:2 Location:New York ### Active Founders ### Will Tong, Founder Co-Founder and CEO of Formula Insight. Will brings a decade of financial expertise to the role. He spent the past 4 years in public equities investing, most recently managing a healthcare equities portfolio at Citadel. In 2022, Will distinguished himself as a top 5 equity analyst within Citadel's Global Equities division. ### Stefan Raghavan, Founder Co-Founder and CTO at Formula Insight. Background in aerospace engineering working on rockets and drones, most recently at SpaceX and Zero Aviation. Led design, testing and certification for engine components on Falcon 9 and Starship. Most recently founded Zero Aviation, a defense startup focused on developing long range fixed wing drones. ### Company Launches [### Formula Insight: Help Hedge Funds manage, track & query their Excel models]( **tl;dr** Hedge Funds use spreadsheets to manage billions of dollars for investments. However, it’s inefficient to extract and compare details between various complex models with millions of data points. **The Problem** The current method of Excel record keeping precludes key details and is not intuitive to search and extract. Hedge funds are only able to keep track of a few metrics manually. **The Solution** So we're building an application that can label and store every single cell in a central repository. Our solution monitors forecast changes and quantifies prediction accuracy, enabling more informed investment decisions. **Our backstory** [Will]( managed an Equities Portfolio at Citadel prior to this and [Stefan]( used to be a propulsion engineer at SpaceX. **Our ask** * Please make an introduction on [Linkedin]( or email us at [[email protected]](mailto:[email protected])

Abel Police

abel-police

http://abelpolice.com

Abel takes police body camera footage and turns it into completed police reports. Currently American patrol officers tend to spend one third of their shift writing reports - turning that to zero would increase the effective police force by 50%, radically changing US cities.

Automating police paperwork

Government

Government

S24

Active

Early

https://www.ycombinator.com/companies/abel-police

https://yc-oss.github.io/api/batches/s24/abel-police.json

Title: Abel Police: Automating police paperwork | Y Combinator URL Source: Markdown Content: ### Automating police paperwork Abel takes police body camera footage and turns it into completed police reports. Currently American patrol officers spend one third of their shift writing reports - turning that to zero would increase the effective police force by 50%, radically changing US cities. ### Jobs at Abel Police San Francisco, CA, US $100K - $150K 0.40% - 0.70% 3+ years Abel Police Founded:2024 Team Size:5 Location: ### Active Founders ### Daniel Francis, Founder Software engineer who built and sold a fitness app, now interested in police technology

Taxo

taxo

https://taxo.ai

San Francisco, CA, USA

Taxo integrates with electronic health records (EHRs) to automate medical billing and coding. Its AI-powered solution reduces the time and cost of claims processing by 90%, enabling providers to focus on patient care rather than administrative tasks.

Automated medical billing and coding

Healthcare

Healthcare

S24

Active

Early

https://www.ycombinator.com/companies/taxo

https://yc-oss.github.io/api/batches/s24/taxo.json

Title: Taxo: Automated medical billing and coding | Y Combinator URL Source: Markdown Content: ### Automated medical billing and coding Taxo integrates with electronic health records (EHRs) to automate medical billing and coding. Its AI-powered solution reduces the time and cost of claims processing by 90%, enabling providers to focus on patient care rather than administrative tasks. Taxo Founded:2024 Team Size:8 Location:San Francisco ### Active Founders ### Ahmed Kerwan, Founder Physician-scientist with a research portfolio in surgery, AI, and hospital workflows (\>24,000 citations). Previously a Fulbright Scholar and Knox Fellow. Interested in autonomous healthcare. ### Hassan Tahir, Founder Currently focused on implementing AI in healthcare. Background in building NLP powered products from research to production using state of the art ML architectures including LLMs from 2017 onwards. Author of published ML research in scientific journals. ### Company Launches [### Taxo - Automated medical billing and coding]( **TL;DR** --------- [Taxo]( integrates with electronic health records (EHRs) to automate medical billing and coding. Its AI-powered solution reduces the time and cost of claims processing by over 90%, enabling providers to focus on patient care rather than administrative tasks. Watch our demo [here]( **Why is this important?** -------------------------- In short, medical billing and coding allow healthcare providers to get paid for their services. However, the way it’s currently done is still largely manual and can take several hours per patient. In 2023, the U.S. healthcare system spent $4.8 trillion, with over 35% of costs attributed to administration, much of which stems from the need to manage vast amounts of unstructured and fragmented data. This situation results in healthcare professionals spending more time on paperwork than on patient care, which reduces service capacity, delays treatment, and drives up costs. **Our solution** ---------------- We are starting by focusing on prior authorization, which requires the provider to get an agreement from the payer to cover specific treatments or medications. This process currently requires manually going through hundreds—sometimes thousands—of records to extract evidence and prove ‘medical necessity’ to the payer. By leveraging AI, we’ve been able to reduce the amount of time it takes by 90%. Providers across the US invariably say this has been life-changing for them! **Moving forward** ------------------ Prior authorizations are just the beginning. We’ve already started working with our providers to build for other aspects of the medical billing and coding process: claim denials, insurance verification, patient letters etc. We believe humans in healthcare should focus on what they do best, caring for patients, and let AI handle the rest. **Our ask** ----------- Medical billing and coding are relevant in all sorts of places. If you know decision-makers in any of these organisations, we’d be super grateful for an introduction! Email us at [[email protected]](mailto:[email protected]) 1. Hospitals 2. Physician offices 3. Nursing homes 4. Outpatient facilities 5. Billing companies 6. Home healthcare services 7. Durable medical goods providers 8. Practice management companies 9. Federal and state government agencies 10. Commercial payers And tell them not to worry about hallucinations. All our output is referenced and directly links back to the original patient records! **Team** -------- [Ahmed Kerwan]( a practicing physician, experienced firsthand the immense burden of paperwork that detracted from his ability to provide care. Determined to find a solution, he pursued a fellowship at Harvard to explore the intersection of AI and healthcare. He then partnered with [Hassan Tahir]( a leading deep-learning engineer specializing in state-of-the-art transformer models. Together, they created Taxo, a platform that has already enabled dozens of providers to process thousands of patient cases more quickly and cost-effectively.

Superunit

superunit

http://superunit.ai

Los Angeles, CA, USA

Superunit is an AI-powered accounting agent for companies selling physical goods. We have 6 customers doing $8.3k in booked revenue. For these companies we close the books 5x faster than human teams and with fewer errors. We offer automated bookkeeping, supply chain management, inventory costs, and contribution margin analysis. Our system seamlessly integrates with Quickbooks, eliminating manual data entry and minimizing errors. Users can interact with Superunit via Slack or email, avoiding the need to learn new software. (https://www.loom.com/share/1425f6802d774e1aa00a8ac4b64782ff)

AI accounting agent for companies selling physical goods

B2B

B2B -> Operations

S24

Active

Early

https://www.ycombinator.com/companies/superunit

https://yc-oss.github.io/api/batches/s24/superunit.json

Title: Superunit: AI accounting agent for companies selling physical goods | Y Combinator URL Source: Markdown Content: ### AI accounting agent for companies selling physical goods Superunit is an AI-powered accounting agent for companies selling physical goods. We have 6 customers doing $8.3k in booked revenue. For these companies we close the books 5x faster than human teams and with fewer errors. We offer automated bookkeeping, supply chain management, inventory costs, and contribution margin analysis. Our system seamlessly integrates with Quickbooks, eliminating manual data entry and minimizing errors. Users can interact with Superunit via Slack or email, avoiding the need to learn new software. ( Superunit Founded:2024 Team Size:2 Location:Los Angeles, CA ### Active Founders ### Peter Marler, Founder Peter, CEO of Superunit, leads his second venture after founding the VC-backed Robin Golf. His career spans roles at Facebook, Quantifind, and NBCUniversal. He holds an MBA from Stanford. ### Zach Yellin-Flaherty, Founder Zach is a software engineer who spent years building payments infrastructure at Stripe and Index. He hold two Computer Science degrees from Stanford. ### Company Launches [### Superunit 💆🏼‍♂️ AI accounting agents for companies selling physical products]( **🙈 TL:DR:** Companies selling physical products spend a fortune on accounting and are still left with incorrect financials because of inventory related complications. We fix that. 🥵 **PROBLEM**: Companies selling physical products struggle with accounting and inventory operations. Current bookkeeping methods rely heavily on manual input and are often disconnected from accurate inventory data. This leads to mistakes in important financial figures like cost of goods sold (COGS), inventory value, margins, and cash flow. As a result, businesses spend a lot of time and money only to gain an inaccurate picture of their financial health. These inaccuracies lead to poor budgeting decisions, loan rejections, failed fundraising, non-compliance, and frequent cash flow crises. 😍 **OUR SOLUTION:** A powerful AI accounting agent built specifically for companies selling physical products. Our platform automates bookkeeping and completes a variety of crucial product operations tasks such as calculating COGS, optimizing POs, and conducting contribution margin analysis with only a few clicks. All financial and inventory data is connected and written back to QBO or your ERP, eliminating financial statement errors, saving dozens of hours per month, and thousands on accounting costs. 💡 **THE INSPIRATION:** As the founder of Robin Golf, [Peter]( encountered this problem frequently. Doing the financials required an expensive team and the books were always inexplicably wrong. As as a result, the company was underwritten for debt and equity incorrectly, wasn’t able to make proper budgeting decisions, couldn’t forecast inventory needs, couldn’t optimize channel mix, and was blindsided by cash shortfalls. 👬 **THE TEAM**: [**Peter Marler**]( (CEO) was the co-founder/CEO of Robin Golf, a venture backed, multichannel golf equipment company. Prior to that he held roles at Facebook, Quantifind, and NBCUniversal. He is a graduate of Stanford GSB. [**Zach Yellin-Flaherty**]( (CTO) previously built payments infrastructure at Stripe and Index. He holds two Computer Science degrees from Stanford. 🙏🏼 **ASK:** We’re onboarding customers and looking to grow! We’d love intros to CEOs/CFOs/COO of companies selling physical products - ecomm brands, wholesalers, retailers, dropshippers. 🔗 **LINKS:** Connect with us on these platforms in the meantime! [**Company Linkedin**]( [**Peter Linkedin**]( [**Zach Linkedin**]( [**Peter Twitter**](

Moreta

moreta

https://moretapay.com

San Francisco, CA, USA

At Moreta, we simplify payments for international travelers, letting you pay for anything using just your phone. Say goodbye to the hassle of searching for the best cash exchange rates or ATMs on your next adventure. With Moreta, paying local merchants is easy, and the best part? Transparent fees. We came up with Moreta while on our own backpacking adventures. Watching locals pay with a simple scan while we fumbled with cards and cash. So, we created a solution. By syncing with local payment networks, Moreta lets you pay like a local—securely and digitally—without needing a bank account in every country. We’re starting in Asia, where QR code digital wallets are taking over and are now the top choice at checkout. As these wallets start to join forces for cross-border payments, Moreta will be the first to bridge the gap for international travelers from card-centric markets to seamlessly tap into these local networks. Moreta will target 67 million travelers arriving in the region each year, who together have $127 billion in everyday expenditure.

Payments made easy for international travelers.

Fintech

Fintech -> Payments

S24

Active

Early

https://www.ycombinator.com/companies/moreta

https://yc-oss.github.io/api/batches/s24/moreta.json

Title: Moreta: Payments made easy for international travelers. | Y Combinator URL Source: Markdown Content: ### Payments made easy for international travelers. At Moreta, we simplify payments for international travelers, letting you pay for anything using just your phone. Say goodbye to the hassle of searching for the best cash exchange rates or ATMs on your next trip. With Moreta, paying local merchants is easy, and the best part? Transparent fees. We came up with Moreta while on our own backpacking adventures. Watching locals pay with a simple scan while we fumbled with cards and cash. So, we created a solution. By syncing with local payment networks, Moreta lets you pay like a local—securely and digitally—without needing a bank account in every country. We’re starting in Asia, where QR code digital wallets are taking over and are now the top choice at checkout. As these wallets start to join forces for cross-border payments, Moreta will be the first to bridge the gap for international travelers from card-centric markets to seamlessly tap into these local networks. Moreta will target 67 million travelers arriving in the region each year, who together have $127 billion in everyday expenditure. Moreta Founded:2024 Team Size:2 Location:San Francisco ### Active Founders ### Walid Mahd, Founder/CEO Walid is the co-founder and CEO of Moreta. He was previously a card and fintech payments consultant at Mastercard Advisors and is an Ivey Business School HBA graduate. He is an avid world traveler and backpacker, having explored over 30 countries across Europe, Southeast & Eastern Asia, Africa, and the Middle East. ### Sam Yu, Founder/CTO Sam is the co-founder and CTO of Moreta. He studied Computer Science at the University of Waterloo and has extensive experience building and shipping consumer products. As the Head of Research & Development at Stems, he led the development of their first mobile app and generative AI product. During his internships at Apple, Google, Meta, and Uber, Sam worked on high-impact projects, including developing the instagram.com search feature used by over 50 million people. ### Company Launches [### Moreta: Payments made easy for international travelers]( Hi everyone, we’re Walid and Sam! We are avid travelers and backpackers who’ve been to over 40 countries combined. 🌏🎒 **TLDR:** Moreta enables international travelers to pay for anything using their phone anywhere in the world. We’re launching first to allow users to pay in Southeast & East Asian markets. [Join our waitlist for early access.]( ❌ **THE PROBLEM** When traveling abroad, do you find yourself asking if merchants accept card? How much cash do you need? Where can you get the best rate? Payments around the world are very fragmented and rely heavily on cash, or an alternative like QR code digital wallets. And for foreigners traveling, this typically means you need some combination of card and cash. In many countries, merchants may not accept card payments, and when they do, they often charge extra fees. Cash payments can also be costly due to high foreign exchange and ATM fees. ✅ **THE SOLUTION** Moreta simplifies payments for international travelers, letting you pay like a local using just your phone. Users download Moreta, upload documentation verifying their identity, connect their bank account to load funds, and pay, all within minutes. Moreta will launch first in the Southeast and East Asian markets. Users can scan existing merchant QR codes and pay like locals. 👥 **THE TEAM** Walid (CEO) and Sam (CTO) have been friends since high school. [Walid]( was a payments consultant at Mastercard Advisors and experienced the payment pains of an international traveler when he backpacked through Southeast & Eastern Asia. [Sam]( was previously the Head of Research & Development at Stems and developed the Instagram web search feature used by over 50 million people. 🎉 **BONUS** Moreta launch party social in Bangkok on October 25th— feel free to join us if you’re in the area! [

Brighterway

brighterway

https://www.brighterway.ai/

At Brighterway, we use AI to assist physicians in the workers' compensation claims review process. Physicians typically spend significant time reviewing thousands of pages of medical documents– our technology highlights the 10-15% of pages that are relevant, streamlining their review process and saving them time. Brighterway structures and deduplicates disorganized medical records (PDFs) and summarizes them down to content related to a doctor's specialty. In our online portal, physicians can view our summaries side-by-side with the original documents, and can filter, edit, and search for specific body parts or visits (i.e. “Clinical Notes” or “Ankle”) that can get them the answers they need in seconds, rather than hours.

Automating the workers-compensation claims review process

Healthcare

Healthcare -> Healthcare IT

S24

Active

Early

https://www.ycombinator.com/companies/brighterway

https://yc-oss.github.io/api/batches/s24/brighterway.json

Title: Brighterway: Streamlining the record review process for physicians in medlegal… | Y Combinator URL Source: Markdown Content: ### Streamlining the record review process for physicians in medlegal… At Brighterway, we use AI to speed up the medical record review process for physicians in medlegal exams. Physicians spend hours reviewing thousands of pages of medical documents– our technology highlights the information that is most relevant to a doctor's specialty, streamlining their review process and saving them time. Brighterway structures and deduplicates disorganized medical records (PDFs) and extracts the content that is medically relevant and related to a doctor's specialty, so they can focus on what matters most. Brighterway Founded: Team Size:2 Location: ### Active Founders ### Walker Wells, Founder Before Brighterway, Walker studied Statistics and Data Science at Yale University. He has also previously led a team of interns on a data science project focused on tracking user engagement and feature impressions at another startup. Fun fact? He solo-backpacked for 5 months at the ripe age of 18. ### Jason Apostol, Founder Improving the Workers Compensation review process with AI. Previously studied Computer Science at Yale. ### Company Launches [### Brighterway – AI-powered review of medical records]( Are you spending all of your time reading unnecessary medical documents? Yea, we didn’t think so. But many people are. At [Brighterway]( we use AI to speed up the record review process for physicians in medical legal exams. Physicians typically spend significant time reviewing thousands of pages of medical documents– our technology highlights the **10-15%** of content that is most relevant, streamlining their review process and saving them time. **OUR TEAM** [**Walker**]( **(left)**: Walker was a Statistics and Data Science major at Yale University. Previously, he led a data science team at another startup and was in charge of some of the largest student organizations at Yale – he is writing this post 🙂 [**Jason**]( **(right)**: Jason was a Computer Science major at Yale, completing a joint BS/MS degree, and was president of Yale Artificial Intelligence (YAI) – he is watching Walker write this post 🙂 **💥💥⚠️ THE PROBLEM ⚠️💥💥** When someone is injured on the job, in an auto accident, etc., they may file a claim for compensation or disability benefits. In such cases, a physician is often appointed to conduct an Independent Medical Examination (IME) or, in certain jurisdictions, a Qualified Medical Examination (QME) to provide an objective assessment of the worker’s condition. To make an informed evaluation, the examining physician must review the injured worker's medical records. These records can be extensive—often comprising **thousands of pages** (we’ve seen as high as 45,000 pages for one patient!). However, not all of this information is directly relevant to the physician’s specialty or to the questions of the claim, so they **spend hours** sifting through the records to identify the most pertinent information. **PHYSICIANS BEFORE USING BRIGHTERWAY** **⭐️⭐️ THE SOLUTION ⭐️⭐️** Brighterway’s system structures and deduplicates disorganized medical records (PDFs) and breaks them down to content related to a doctor’s specific specialty. So, **physicians can see what is most meaningful for them** without having to sift through thousands of pages of unimportant text. This saves physicians **hours each day**. Summaries are created and can be viewed side by side with the original records, so physicians can edit and type directly within the outputs (like a Google Doc) so they don’t have to leave the page when writing notes or a report. **PHYSICIANS AFTER USING BRIGHTERWAY** **\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_\_** **How can you help?** If you have any friends, family, or connections to somebody doing med-legal exams or IME/QME work (or wanting to start doing such work) either as a physician or a medical exam management company (a business that coordinates and manages medical evaluations for a group of physician examiners), send them our way! ### Hear from the founders #### How did your company get started? (i.e., How did the founders meet? How did you come up with the idea? How did you decide to be a founder?) Jason and Walker met at Yale University their first year. They realized their product could work when Walker broke his ankle, and while speaking with an orthopedic surgeon Walker told him what they were working on. The surgeon ended up sharing that he used to do workers comp cases and stopped because it was too time intensive, but that if he could use Brighterway’s product he would like to start doing it again.

Zuni

zuni

https://zuni.app

Remote

Smarter email for founders

B2B

B2B -> Productivity

S24

Active

Early

https://www.ycombinator.com/companies/zuni

https://yc-oss.github.io/api/batches/s24/zuni.json

Title: Zuni: Smarter email for founders | Y Combinator URL Source: Markdown Content: ### Smarter email for founders Zuni Founded:2024 Team Size:2 Location: ### Active Founders ### George Seabridge, Founder Building tools to make the web real-time (so you don't have to). ### Will Taylor, Founder Former YC CTO and startup builder, now making the productivity tools that I've always wanted in my career. Previously early eng with GitHub co-founders on their second venture.

Zenbase AI

zenbase-ai

https://zenbase.ai

San Francisco, CA, USA; Remote

Zenbase helps developers focus on programming by automating prompt engineering and model selection. We’re core contributors of Stanford NLP’s DSPy, the #1 LLM optimization framework used by Meta, Microsoft, Google, among many others. But it's hard to use outside of academia. We're building the production version.

Continuous prompt optimization for LLMs from DSPy core contributors

B2B

B2B -> Engineering, Product and Design

S24

Active

Early

https://www.ycombinator.com/companies/zenbase-ai

https://yc-oss.github.io/api/batches/s24/zenbase-ai.json

Title: Zenbase AI: Continuous prompt optimization for LLMs from DSPy core contributors | Y Combinator URL Source: Markdown Content: ### Continuous prompt optimization for LLMs from DSPy core contributors Zenbase helps developers focus on programming by automating prompt engineering and model selection. We’re core contributors of Stanford NLP’s DSPy, the #1 LLM optimization framework used by Meta, Microsoft, Google, among many others. But it's hard to use outside of academia. We're building the production version. Zenbase AI Founded:2024 Team Size:2 Location:San Francisco ### Active Founders ### Cyrus Nouroozi, Founder & CEO Cyrus is the Co-Founder @ Zenbase AI and a core contributor to Stanford NLP's DSPy. He went from 0 to PhD-level AI researcher in 4 months in 2023, and taught himself how to code when he was 14. He enjoys the little moments in a day with coffee, ecstatic dance, and tai chi. ### Amir Mehr, Co-Founder & CTO With a Master’s in Computer Science from the University of Calgary and as a key contributor to StanfordNLP's DSPy project, Amir has deep expertise in AI and software engineering. He played a crucial role in engineering at HubMeta and OVOU. His early success running a profitable web hosting business in high school showcases his technical acumen and entrepreneurial drive. ### Company Launches [### Zenbase: Continuous prompt optimization from DSPy core contributors]( **TL;DR:** Zenbase helps developers focus on programming by automating prompt engineering and model selection. We’re building developer tools and cloud infrastructure for teams to save time, never get stuck in prompt hell, and create AI apps that get smarter over time. Hey there! We're [Cyrus]( & [Amir]( In the past, we've both been lead engineers and founding CTOs. We became contributors to [DSPy]( and discovered the future of programming with language models. This is the story of how we came to this insight, our glimpse into the future, and 2 case studies on how Zenbase has helped companies escape prompt hell and scale prompt engineering. Problem ------- Prompt “engineering” is the most time-consuming, stressful, and uncertain part of programming with LLMs. With DSPy, we had found something profound. It promised to save us from the all-too familiar user journey we — like so many others — had experienced. DSPy kept growing. It became Stanford NLP's #1 GitHub repo with 16K stars. We started hearing of folks in Microsoft, Amazon, Google, and 40+ other companies using DSPy to prototype apps with it. We began hearing the same things all over again. Although many found DSPy elegant and intuitive, countless folks found it impossible to grok. Those who managed to build something with it had headaches productionizing it; finding it difficult to scale, make reliable, and make performant. So, we set out to create the productionized DSPy. Solution -------- Zenbase lets you optimize your prompts and models. We offer: 1. [zenbase/core]( is an open-source Python library that you can use to optimize your existing LLM pipelines using DSPy’s optimizers (versus having to rewrite your pipelines in DSPy) 2. A hosted API for creating AI functions that get smarter with time. We ingest user feedback to continuously optimize the prompt and model. We use the latest tricks from DSPy, our own custom optimizers, and fine-tuning as appropriate to execute your intents in a way that's good, fast, and at a reasonable price. 3. An on-prem API for businesses with data privacy requirements. Use Cases --------- ### How Zenbase saved Vera from Prompt Hell > **Zenbase came into the trenches with us to improve our evals from 10% to 80%.** It really felt like they were a part of our team. > > — Taeib, Cofounder @ [Vera-Health.ai (YC S24)]( They were staying up until 3am on multiple nights trying to prompt engineer their RAG query generator to retrieve the correct documents. Their progress was uncertain. It was stressful. We call this prompt hell. Prompt engineering is the most uncertain, risky, and stressful part of programming with LLMs. There didn’t seem to be a way out, but with Zenbase, they saw the light at the end of the tunnel. Zenbase makes prompting systematic and peaceful. **We helped Vera go from demo to production, by optimizing the prompt of their query generator.** With a product that could handle doctors’ stress tests, they could focus on selling, and go to bed at a good time. ### How Zenbase helped Superfilter create an AI that’s personalized to its users > I’ve seen a lot of AI Devtools and Zenbase is solving a problem that everyone building with AI will have when going to production. **The best part is their product is so easy to use that it’s a no brainer.** > > — Scott, CEO @ [Superfilter.ai]( (YC S24) It was all going great. Superfilter had just tested their AI email copilot with their beta users of investors and startup founders, and their users were excited. They onboarded a new cohort, and their prompts broke down. It worked well for the investors and startup founders, but not everyone. Scott and his cofounder Travis realized that prompt engineering wasn’t going to scale to accurately categorize user emails into important, action required, or ignore. Superfilter used our hosted API to create email categorizers that learned from users’ existing behaviour. With automatic prompt engineering, they were able to scale personalized experiences for every user. **Zenbase makes personalized AI apps easier to build and scale with automated prompt engineering.** Asks ---- 1. Are you in prompt hell, and do you want to feel the Zen? Are you trying to scale the personalization of prompts for every account?Learn more [on our website]( and [schedule a demo]( with us so we can understand where you are and where you want to be. Let us be your LLM doctors and wizards as we guide you to where you want to go. 2. Use our [MIT-licensed Python SDK]( to optimize your existing prompts with DSPy’s optimizers. 3. Kindly share this post with anyone you know who could benefit 🙏 ### Hear from the founders #### How did your company get started? (i.e., How did the founders meet? How did you come up with the idea? How did you decide to be a founder?) Cyrus and Amir have been working in startups as lead engineers and CTOs for 10+ years.Cyrus and Amir were introduced by a mutual friend 3 years ago. Worked on a handful of consulting projects together. Cyrus would sell, Amir would build. They were building AI apps and wondered, what if there were a way to use AI to write the prompt. Then, Cyrus discovered [DSPy]( the home of automated prompt engineering. He met up with its creator and took on a major role in the project. He persuaded Amir to join in. They became core contributors.Being startup guys at heart, they started exploring opportunities, and things just started clicking into place. #### What is the core problem you are solving? Why is this a big problem? What made you decide to work on it? AI systems are static prompts and models. AI systems need to be able to learn on the job. We build AI systems that can learn. #### What is your long-term vision? If you truly succeed, what will be different about the world? **Developers save time and effort:** With Zenbase, they can ship high quality AI features in hours and not days.**Users receive a better product:** They can confidently delegate knowledge tasks with more because they have an AI that gets smarter and faster the more you teach it.**Businesses drive quality and margins:** The highest performing, most cost-effective prompt and model for the task are always used.

Offstream

offstream

https://www.useoffstream.com/

San Francisco, CA, USA

Offstream is Vanta for carbon compliance. Today, every new infrastructure project, from data centers to biomass power plants has a complicated mess of local, state, federal, and private sector standards that they must understand and comply with to make their projects profitable. Offstream solves this by using LLMs to parse regulations, generate technical reports for customers, and manage 100,000+ operational data points on project’s lifecycles. We cut the time customers spend on compliance by 90% and at 1/5th of the cost of alternatives.

Vanta for carbon compliance

B2B

B2B

S24

Active

Early

https://www.ycombinator.com/companies/offstream

https://yc-oss.github.io/api/batches/s24/offstream.json

Title: Offstream: Vanta for carbon compliance | Y Combinator URL Source: Markdown Content: ### Vanta for carbon compliance Offstream is Vanta for carbon compliance. Today, every new infrastructure project, from data centers to biomass power plants has a complicated mess of local, state, federal, and private sector standards that they must understand and comply with to make their projects profitable. Offstream solves this by using LLMs to parse regulations, generate technical reports for customers, and manage 100,000+ operational data points on project’s lifecycles. We cut the time customers spend on compliance by 90% and at 1/5th of the cost of alternatives. Offstream Founded:2023 Team Size:3 Location:San Francisco ### Active Founders ### Varsha Ramesh Walsh, Founder CEO and co-founder at Offstream. HBS '23. Previously business lead at Indigo Ag where I launched a traceable cotton business where I closed contracts that got the business to $50M revenue bookings in a year. At Indigo, I experienced the pain point we are now solving at Offstream first-hand. We hired 80 ppl and spent 50M to just get carbon compliant. ### Zach Janicki, Founder CTO and cofounder of Offstream. CS from Notre Dame. Previously 1st engineering hire at Repool (YC S21) building hedge fund software, and early engineer at Pathpoint building insurance brokerage software. I've spent my career turning paperwork into software ### Company Launches [### Offstream - Streamlined carbon compliance for project developers]( We’re Varsha and Zach, the co-founders of [Offstream]( We’re automating carbon compliance for project developers. 👥 **The Team** 5 years ago, [Varsha]( struggled through carbon compliance as a project developer at Indigo Ag. While she was at HBS, she spoke to hundreds of developers with the same challenges she faced, who all asked for a solution to their compliance bottleneck. [Zach]( brings many years of full-stack software engineering experience at early-stage startups across compliance, legal, and accounting. He brings much-needed insight into building scalable products that replace people-intensive workflows. ❗**The Problem** Despite a growing amount of funding (to the tune of $2T annually) flowing into new decarbonization projects, it is incredibly challenging for project developers to complete the paperwork needed to access that funding. Today, carbon project developers navigate a complex web of credits, regulations, and market standards. This information comes from local, state, and federal governments, NGOs, industry coalitions, and investors. Tracking and reporting on compliance data is time-consuming, opaque, and expensive. The standards are dense, and most teams manage millions of data points in Excel and emailed PDFs, creating admin headaches. 🏆 **The Solution** Offstream is a one-stop shop for carbon compliance, making it simple for carbon project developers to get compliant and stay compliant with regulatory and voluntary market compliance standards. [ Users set up and generate life cycle assessments (LCA), which are a core component of all compliance standards. Offstream builds off the LCA to generate compliance documentation at a fraction of the cost and time of alternatives. [Are you ready to stop spending time on paperwork and spend more time scaling decarbonization? Join us.](

Proxis

proxis

https://www.proxis.ai/

San Francisco, CA, USA

Proxis is the first dedicated platform for LLM distillation and serving, unlocking production ready models at 1/10th the cost. Fine-tuning frontier models like GPT-4 on proprietary data is expensive, and locks customers into one external closed source model provider. To solve this, enterprise AI teams can use Proxis to distill state-of-the-art Llama 405b into an efficient model tuned on their own data, which is 5x faster and 10x cheaper than closed source alternatives. Jackson (CTO) optimized the Gemini model at Google for efficient deployment at massive scale. Liam (CEO) built zero-to-one systems as a software engineer.

The platform for LLM distillation and serving.

B2B

B2B -> Engineering, Product and Design

S24

Active

Early

https://www.ycombinator.com/companies/proxis

https://yc-oss.github.io/api/batches/s24/proxis.json

Title: Proxis: The developer platform for LLM distillation and serving. | Y Combinator URL Source: Markdown Content: ### The developer platform for LLM distillation and serving. Proxis is the first dedicated platform for LLM distillation and serving, unlocking production ready models at 1/10th the cost. Jackson (CTO) optimized the Gemini model at Google for efficient deployment at massive scale. Liam (CEO) built zero-to-one systems as a software engineer. Fine-tuning frontier models like GPT-4 on proprietary data is expensive, and locks customers into one external closed source model provider. To solve this, enterprise AI teams can use Proxis to distill state-of-the-art Llama 405b into an efficient model tuned on their own data, which is 5x faster and 10x cheaper than closed source alternatives. Proxis Founded:2024 Team Size:2 Location:San Francisco ### Active Founders ### Liam Collins, Founder CEO and cofounder of Proxis. I built zero-to-one systems as a software engineer for two years and prior to that was an investment banker in renewable energy. I received my bachelors from Columbia University and City University of Hong Kong (on a Dual Degree), and left my MBA at Wharton to build Proxis. ### Jackson Stokes, Founder CTO and cofounder at proxis; I optimize ML models to run more efficiently. Previously, I worked on kernel level optimizations for Gemini at Google, and framework level optimizations on video models at Google Research. ### Company Launches [### Proxis - the first dedicated platform for LLM distillation and serving]( **Tl;dr:** [Proxis]( is the first dedicated platform for LLM distillation and serving, unlocking production ready models at 1/10th of the cost. ### The Problem * Fine-tuning frontier models like GPT-4 on proprietary data is expensive, and locks customers into one external closed source model provider. * These large models are 100x the price per-token of smaller Llama 3.1 models, and run nearly 10x slower. This makes them impractical to deploy at scale. * Closed source model lock-in means customers can’t tweak or tune their own model, or deploy on-prem for sensitive data solutions. ### The Solution Model Distillation: * In model distillation, a large **teacher model** effectively fine tunes a smaller **student model** without the need for _labeled_ or _structured_ datasets. This ‘condenses’ large models down to the compute cost of small models. * Distillation results in near-frontier model quality, with efficient performance. This means we can achieve 5x the speed at 1/10th the cost. * With the release of Llama 405b in late July, open source frontier distillation became available to the public for the first time. ### Growth & Market * Llama model downloads have grown 10x in the last year and are approaching 350 million downloads to date. There were 20 million downloads in the last month alone, making Llama the leading open source model family. * Llama usage by token volume across major cloud service providers has more than doubled in just three months from May through July 2024 when Llama 3.1 was released. * Monthly usage (token volume) of Llama grew 10x from January to July 2024 for some of the largest cloud service providers. Source: [Meta]( ### The Team [Jackson]( (CTO, on the left) **optimized the Gemini model at Google** for efficient deployment at massive scale. [Liam]( (CEO, on the right) built zero-to-one systems as a software engineer. ### The Ask Give it a go! Sign up to our waitlist [here]( to access Proxis-hosted Llama models at a cheaper cost than current offerings. #### YC Sign Photo

Focal

focal-2

https://usefocal.ai

San Francisco, CA, USA

Focal enables large enterprises to effectively leverage AI to automate traditionally manual workflows. Our initial focus is geared towards helping financial institutions automate compliance, risk, and dispute operations.

AI automation for enterprises

B2B

B2B -> Operations

S24

Active

Early

https://www.ycombinator.com/companies/focal-2

https://yc-oss.github.io/api/batches/s24/focal-2.json

Title: Focal: AI automation for enterprises | Y Combinator URL Source: Markdown Content: ### AI automation for enterprises Focal enables large enterprises to effectively leverage AI to automate traditionally manual workflows. Our initial focus is geared towards helping financial institutions automate compliance, risk, and dispute operations. Focal Founded:2024 Team Size:2 Location:San Francisco ### Active Founders ### Katia Bazzi, Co-Founder Katia is the CTO of Focal, building AI automation for enterprises. Previously, Katia was an engineering leader at Stripe -- there, she launched new financial products such as credit cards and banking as a service, processing \>$6B per year. Before that, she led engineering teams at a distributed tracing startup (Lightstep) and Eventbrite after graduating from CMU. ### Ray Morgan, CEO / Co-Founder Ray is the CEO of Focal, building AI automation for enterprise companies. Before Focal, Ray spent 10 years at Stripe leading and scaling the core payments engineering teams (<$10B payments volume to well over $100B a year), launching new products (Stripe Issuing, new payment methods), and worked on Stripe’s enterprise readiness efforts to enable enterprise accounts like Amazon. Prior to Stripe, Ray started a company that got acquired by Ebay and led the frontend engineering team at Zappos.

FINNY AI

finny-ai

https://www.finnyai.com/

New York, NY, USA

Growth is the number one priority for all Financial Advisors. However, the organic growth process is highly inefficient and full of noise. Advisors often waste hours building prospect lists by aggregating information across horizontal solutions like Linkedin, ZoomInfo, Pitchbook, qualifying these leads, setting up meetings, and spending time on unqualified prospects. Advisors spend an average of 58 hours on business development for every 1 new client converted, and have <1% conversion rates. Due to the time-intensive nature of organic growth, advisors often turn to inorganic avenues for growing, via M&A, despite those being far less capital efficient. At FINNY, we are certain we can do better with AI. With our tool, advisors can: - Identify prospects within their target niche, aggregating thousands of data points per lead - Prioritize prospects based on their predicted likelihood of converting, a score unique to each advisor and prospect pair - Automate the outreach and meeting scheduling with their high priority prospects In short, FINNY is the one-stop shop for advisors when growing their book of business, starting by solving prospecting. FINNY launched paid pilots in May to select RIAs, and our waitlist grew to 30 firms in under one month.

Using ML to supercharge organic growth for Financial Advisors

Fintech

Fintech -> Asset Management

S24

Active

Early

https://www.ycombinator.com/companies/finny-ai

https://yc-oss.github.io/api/batches/s24/finny-ai.json

Title: FINNY AI: The OS for independent financial advisors | Y Combinator URL Source: Markdown Content: ### The OS for independent financial advisors FINNY is building a new operating system for independent financial advisors, and starting by helping them grow. Our goal is to empower the best financial advisors (the independents!) to serve all Americans who want financial advice. We are starting with growth, which is the most existential problem advisors face. The organic growth process is highly inefficient and full of noise. Advisors often waste 60 hours of business development to convert 1 new client. At FINNY, we are certain we can do better with AI. With our tool, advisors can: - Identify prospects within their target niche, aggregating thousands of data points per lead - Prioritize prospects based on their predicted likelihood of converting, a score unique to each advisor and prospect pair - Automate the outreach and meeting scheduling with their high priority prospects In essence, doing all the work for the advisors, and letting them focus on what actually matters, the client relationship. ### Jobs at FINNY AI New York, NY, US $100K - $150K 1.00% 1+ years FINNY AI Founded:2024 Team Size:4 Location:New York ### Active Founders ### Eden Ovadia, Founder Eden is the Co-founder and CEO of FINNY. She studied Software Eng. at McGill University with a focus on ML. Prior to starting FINNY, Eden was at BCG where she worked in Private Equity & Financial Services. ### Theodore Janson, Founder Theo is the co-founder and CTO at FINNY. Theo holds a Master's degree in Artificial Intelligence from Ecole Polytechnique, Paris, and a bachelor's degree in Electrical Engineering and Math from McGill University. During his studies, Theo held research roles in high-energy physics and explainable Machine Learning. Prior to starting FINNY, he led a team of machine learning engineers developing tools for financial institutions at an early Fintech startup. ### Victoria Toli, Founder Victoria is the co-founder and CPO of FINNY, an AI-driven growth engine for financial advisors. Before FINNY, she was a Product Manager at Uber, where she led Membership Growth for Uber's subscription product, Uber One. She graduated from Stanford in 2020, with a BS in Symbolic Systems, Artificial Intelligence. Prior to coming to the US for college, Victoria grew up in Athens, Greece. ### Company Launches [### ⚡FINNY - Supercharge organic growth for Financial Advisors]( Hey everyone, we are Theo, Eden, and Victoria — the founders of [FINNY]( We help independent financial advisors find and convert new clients, using AI. If you’re confused as to why the three of us are so excited about building for financial advisors, keep reading. 🧑‍🔬 **The Team** ------------------ **🏦The Sales Whisperer** [Eden]( was most recently at BCG NYC, where she primarily worked in the Tech, Financial Institutions & Private Equity practices. She holds a BEng in Software Engineering from McGill (ML focus). **👾The Model Man** [Theo]( worked as a ML engineer at an early Fintech startup and AI researcher in Explainable ML. He holds a BEng in EE and Math from McGill and MS in AI from Ecole Polytechnique, Paris. **👩🏼‍🎨The Visionary** [Victoria]( came from Uber, where she led membership growth for Uber One as a Product Manager. She holds a BS from Stanford in Symbolic Systems (AI focus). **🫶🏻 Why are we obsessed with Financial Advisors?** ----------------------------------------------------- * **The Great Wealth Transfer** - $80T of wealth will be transferred from the Boomer/Silent Gen to the next generation over the next 20 years. * **Retiring advisors** - 45,000 financial advisors are over the age of 65 in the US, managing $3T in wealth that will change management when they retire. * **Wealth creation** - Wealth is being created in unprecedented ways (E.g., Innovation, Entrepreneurship, Social media, Racial wealth gain), but the institution serving the wealth has not changed * **Consolidation** - Private equity driving consolidation of a historically long tail of independent advisors * **Price Inertia:** Financial Advisors are used to paying 25 bps annually IN PERPETUITY for referrals. That’s $200B annually from the wealth transfer alone. 💀 **The Problem FINNY is Solving** ----------------------------------- Financial advisors manage \>$100T in assets globally. Yet, bringing on new clients is highly inefficient. Their best bet is to repurpose horizontal solutions - like Linkedin, ZoomInfo, and Pitchbook - to create long and noisy prospect lists. However, without a proper way to qualify these leads or even reach their particular niche at scale, advisors end up spending an average of **58 hours** going after the wrong, often unqualified prospects before converting one new client. 💡 **The Solution** ------------------- FINNY is reimagining organic growth for financial advisors with AI. Our first product, a multi-channel Sales agent, allows users to sit back and relax while it goes “out in the wild” (aka our massive database of high-net-worth individuals) and prospects for them. It’s like your intern, except it’s always on, running in the background. FINNY’s goal is for advisors to wake up in the morning to meetings with prospects on their calendars, entirely booked by FINNY. Our value proposition is three-fold: 1. **_Prospect identification:_** * _We identify prospects within their target niche, aggregating thousands of data points per lead and constantly monitoring for life triggers_ 2. **_Prospect prioritization_:** * We prioritize prospects based on their predicted likelihood of converting, a score unique to each advisor and prospect pair 3. **_Prospect outreach & engagement:_** * _We automate the outreach and meeting scheduling with the high-priority prospects, intelligently selecting the optimal channel for the specific prospect_ ✅ **Progress so far** --------------------- In just 3 months, we have: * Released the **1st** and **2nd iteration** of the product to 15 advisory firms. * Grown revenue by \>**11x** MoM. * Got **70** firms to join our waitlist, without any acquisition efforts or outbound. * **Won** Morningstar’s Fintech Showcase last month. * Have been getting some **buzz**! [Here]( and [here]( **❓One Ask for You (If you made it this far)** ---------------------------------------------- If you ever thought about working with a financial advisor, evaluated one, currently work with one _(or want to get matched with one_ **😉**_),_ we’d love to talk to you. Book 15 min with us [here.]( If you’re in SF, we’ll buy you a coffee. If not, let’s call it an IOU.

Village Labs

village-labs

https://villagelabs.app/

New York, NY, USA

At Village Labs, we help America's businesses become employee owned by helping them set up and manage an employee stock ownership plan (ESOP). ESOPs are a federal retirement plan that enable businesses to give their employees a stake in the company in exchange for significant tax benefits. With 2.9M Baby Boomer business owners set to retire in the coming years, there's a growing "Silver Tsunami" of businesses across the country in search of new ownership. For many of these owners, selling to their employees is a socially-minded yet financially pragmatic solution. By selling ownership of their business to an ESOP, owners get a fair price, tax benefits, and the satisfaction of knowing their employees and legacy will be taken care of — things not guaranteed with traditional buyers like private equity. In turn, employees retain their jobs and get shares in the business that can help them create life changing wealth by the time they retire — imagine an office secretary retiring as a millionaire. Thousands of businesses across America like Bob's Red Mill, Harpoon Brewery, and Recology already have an ESOP, benefiting over 14 million workers and creating 2.1 trillion in employee wealth. We're excited about the opportunity to massively grow these numbers by making it simple for every business to become employee owned.

Helping America's businesses become employee owned

B2B

B2B -> Human Resources

S24

Active

Early

https://www.ycombinator.com/companies/village-labs

https://yc-oss.github.io/api/batches/s24/village-labs.json

Title: Village Labs: Helping America's businesses become employee owned | Y Combinator URL Source: Markdown Content: ### Helping America's businesses become employee owned At Village Labs, we help America's businesses become employee owned by helping them set up and manage an employee stock ownership plan (ESOP). ESOPs are a federal retirement plan that enable businesses to give their employees a stake in the company in exchange for significant tax benefits. With 2.9M Baby Boomer business owners set to retire in the coming years, there's a growing "Silver Tsunami" of businesses across the country in search of new ownership. For many of these owners, selling to their employees is a socially-minded yet financially pragmatic solution. By selling ownership of their business to an ESOP, owners get a fair price, tax benefits, and the satisfaction of knowing their employees and legacy will be taken care of — things not guaranteed with traditional buyers like private equity. In turn, employees retain their jobs and get shares in the business that can help them create life changing wealth by the time they retire — imagine an office secretary retiring as a millionaire. Thousands of businesses across America like Bob's Red Mill, Harpoon Brewery, and Recology already have an ESOP, benefiting over 14 million workers and creating 2.1 trillion in employee wealth. We're excited about the opportunity to massively grow these numbers by making it simple for every business to become employee owned. Village Labs Founded:2023 Team Size:2 Location:New York ### Active Founders ### Roger Cawdette, Co-Founder & CEO These days, I'm working towards expanding employee ownership which I view as the greatest wealth creation opportunity for middle class Americans today. Before this, I co-founded Finary (YC W21) where we were building a social network for the stock market (before WallStreetBets went viral!). I studied Technological & Social Innovation at Harvard and left after my sophomore year to do YC. ### Turja Chowdhury, Co-Founder & CTO Co-founder and CTO @ Village Labs ### Company Launches [### Village Labs - We help America’s businesses become employee owned]( Hello YC community & world, It’s an exciting time! Turja and I are building [Village Labs]( a new firm that helps businesses become employee-owned by helping them set up and manage an employee stock ownership plan (ESOP). ESOPs are federal retirement plans that enable companies to give their employees shares in the business in exchange for major tax benefits. They are NOT stock option plans, and no, we are NOT doing cap table management! **EMPLOYEE OWNERSHIP WILL BE BIG** There are currently about 6,000 companies with an ESOP across the country — the popular names include Publix Supermarkets, Harpoon Brewery, and Recology (the maker of those black, blue, and green trash bins you see on every street in SF). Employee ownership has been around for decades in America, but there’s going to be an explosion in the number of employee-owned businesses that are created in the next few years. There are a couple of reasons for this: 1. **The Silver Tsunami:** 2.9M Baby Boomer business owners are going to sell their businesses in the next few years. Many of them will sell to their employees because they get a fair price, tax benefits, and the comfort of knowing their employees will be taken care of - things not guaranteed by traditional buyers like private equity. 2. **Government support**: Congress and state legislatures are passing more legislation to encourage the creation of ESOPs. This largely looks like more tax benefits for ESOP businesses and owners who sell to an ESOP. 3. **Mainstream awareness**: There’s been a surge in awareness of employee ownership in the last few years because of new press coverage. See: [Pete Stavros on 60 Minutes]( [Blackstone announcing its employee ownership initiative]( [major businesses announcing becoming employee owned.]( **WHAT IS VILLAGE LABS DOING?** At Village Labs, we’re setting out to help create as many employee-owned businesses as possible. In terms of what we do, we help owners sell to their employees via an ESOP and then provide them with the tools they need to manage their plans. **HOW YOU CAN HELP** Most business owners don’t know that selling to their employees is an option (and one that commonly comes with the most financial benefits). We’re working to change that by building a nationwide network of partners with close connections to the many Boomer business owners who will retire soon. We work with these partners to educate owners on selling to their employees. If you’d like to help, you could introduce us to… * Owners and executives at mid-sized businesses with 20+ employees and stable cash flows * Business brokerage firms (e.g. Transworld, Sunbelt Brokers, etc.) * Exit/estate planning firms + wealth managers And if you happen to be one of these, we’d love to talk to you. **WHY WE’RE EXCITED ABOUT EMPLOYEE OWNERSHIP** The short answer is that being employee-owned is just a better way of doing business. Here’s the story of CHI Overhead Doors, a garage door manufacturer in rural Illinois. They’re not an ESOP, but their story highlights what makes employee ownership so awesome. This isn’t an isolated, one-off thing. There are many stories out there of businesses that benefited from becoming employee-owned - and we’ve seen them personally. Turja and I have been on the road for a lot of YC this summer, visiting employee-owned businesses in places like Portland, Tuscon, and other cities across the country. One of the main things you notice is that the employees are noticeably happier, and the quality of service is excellent - likely because the employees have ownership in the business and, therefore, an incentive to see it succeed. It’s the same reason we, as founders, give shares to our employees. Considering 70% of Americans [don’t like their jobs,]( and ~50% [have no retirement savings]( we think widespread employee ownership would bring those numbers down. That’s why our goal is to create as many employee-owned businesses as possible. That means more people who enjoy their jobs, more retirement savings, and a more productive economy. Again, most people don’t know that employee ownership is a thing, but wouldn’t we all be better off if they did? If you’d like to help us with this or are just curious to learn more, please do reach out! Thanks for reading! \-Roger & Turja #### Company Photo

Ionworks

ionworks

https://ionworks.com/

Pittsburgh, PA, USA; Remote

Every battery company spends tens of millions of dollars a year on experiments. Simulations that replace experiments reduce these costs but are currently underutilized in industry because they are hard to create and run. We make simulations accessible to every engineer, by helping them to create models from their own data, then run large scale simulations in the cloud to explore the full design space.

Simulation software used by leading battery and automotive companies

Industrials

Industrials -> Energy

S24

Active

Early

https://www.ycombinator.com/companies/ionworks

https://yc-oss.github.io/api/batches/s24/ionworks.json

Title: Ionworks: Simulation software used by leading battery and automotive companies | Y Combinator URL Source: Markdown Content: ### Simulation software used by leading battery and automotive companies Every battery company spends tens of millions of dollars a year on experiments. Simulations that replace experiments reduce these costs but are currently underutilized in industry because they are hard to create and run. We make simulations accessible to every engineer, by helping them to create models from their own data, then run large scale simulations in the cloud to explore the full design space. Ionworks Founded:2023 Team Size:6 Location:Pittsburgh, PA ### Active Founders ### Valentin Sulzer, Founder Created PyBaMM, the world's leading open-source battery simulation software. Now building Ionworks to make it accessible to every single battery engineer around the world. Previously PhD (Maths) at Oxford + 4 years engineering research at University of Michigan and Carnegie Mellon ### Rob Timms, Founder Co-Founder and CTO at Ionworks, building the modern software stack for batteries, powered by PyBaMM. Core developer of the leading open-source battery modelling package PyBaMM. Ex-Oxford mathematician. ### Company Launches [### Ionworks - Build and run battery simulations at scale]( We provide battery-simulation-as-a-service to help battery engineers accelerate their battery design cycles by replacing experiments with simulations. We’re building this on top of [**PyBaMM,**]( the world’s leading open-source battery modeling software, which was created by our team and is now used by most battery engineering teams around the world. **Our ask:** If you are a battery engineer or know one, reach out to [**[email protected]**](mailto:[email protected]) for a demo **Meet the team** We ([**Valentin**]( [**Rob**]( [**Tom**]( have PhDs in Maths/Chemical Engineering from Oxford/East Anglia/Leeds & UCL, and a collective 20 years of experience working on physics-based battery simulations. As academics, we were frustrated by the scientific reproducibility crisis, so we created an open-source project to share our research and hold ourselves accountable. This turned out to be something people wanted, and PyBaMM has now grown to become the leading open-source battery simulation package used by engineers in battery companies and academic labs around the world. We’re now building Ionworks to scale up this success and bring the power of digital design to every single battery engineer. **The Need** Batteries are a crucial technology underpinning the electrification of every application ranging from EVs to grid storage. This makes the battery market worth _$150b today (1.5% of global GDP)_, with projected growth to $680b in 10 years (4% of projected global GDP). However, batteries still need to improve massively to meet a difficult set of demands: **higher energy and power density, faster charge times, lower cost, longer lifetimes, and better safety.** **Why designing better batteries is difficult** Batteries come in all shapes and sizes, with complex and nonlinear effects of design inputs on performance outputs. As a result, battery engineering teams build and test thousands of physical prototypes to find the optimal design for their application. This is an expensive and time-consuming process (up to $1000 and 2 years for each individual test), which leads to **$100m/5 year battery development cycles**. This makes batteries the bottleneck in many product development cycles, such as new laptops and electric cars. **The Solution** We’re doing for batteries what CFD (computational fluid dynamics) did for aerodynamics and CAD (computer-aided design) did for structural and electrical engineering: **move the design problem into the digital world**. Generic engineering software already exists but requires specialist knowledge and decades of experience to build battery-specific models. We’re building software that makes this process available to any battery engineer. Our solution is built for the cloud, providing the interfaces and cost models to do this at scale. **How it works** We move battery design into the digital world using the following process: 1. Collect a small amount of experimental data 2. Parameterize (i.e. train) the physics-based model to find the parameter values that best match the experimental data 3. Use the generated parameters to virtually explore the entire space of designs and operating conditions 4. Pick the best candidates from the virtual experiments and test those in the physical world to validate the simulation results The underlying physics-based models are old-school partial differential equations that describe the chemical reactions, lithium transport, and heat exchange happening inside a battery. These models generalize well from their training data, which allows us to build accurate models from small amounts of data. **Our secret sauce is that we’ve built software that makes steps 2 and 3 fast and robust.** **Ask** * If you are a battery engineer or know one, reach out to [**[email protected]**](mailto:[email protected]) for a demo * Submit a [**PyBaMM Testimonial**]( for a chance to win some [swag]( * Star [**PyBaMM on GitHub**](

Patchwork Technologies

patchwork-technologies

https://getpatchwork.io

London, England, United Kingdom

Patchwork automates logging from code to storage. We make it simple to implement actionable, structured logging across your entire codebase

The next generation logging and observability platform

B2B

B2B -> Engineering, Product and Design

S24

Active

Early

https://www.ycombinator.com/companies/patchwork-technologies

https://yc-oss.github.io/api/batches/s24/patchwork-technologies.json

Title: Patchwork Technologies: The next generation logging and observability platform | Y Combinator URL Source: Markdown Content: ### The next generation logging and observability platform Patchwork automates logging from code to storage. We make it simple to implement actionable, structured logging across your entire codebase Patchwork Technologies Founded:2024 Team Size:3 Location:London, United Kingdom ### Active Founders ### Ben Hall, Founder Prior to Patchwork, Ben spent almost 5 years at Palantir, mixing code and sales. Most recently he led the company's zero-to-one effort in the EMEA Utilities sector – an engineer in the deep end of sales. He was founding Principal Engineer at deep tech startup, Oxford Flow. The startup spun out of Oxford Uni's aerospace lab, where he had carried out his PhD research in jet engine aerodynamics. Ben has a patent and research papers with collectively thousands of downloads. ### Sam Kramer, Founder Prior PatchWork, Sam spent six years at Palantir, splitting his time equally between product and business development. Sam has done everything from fixing availability bugs in Paxos protocol implementations to running relationships with customer IT. ### Alex Maass, Founder Prior to Patchwork, Alex spent 9 years at Palantir as a Forward Deployed Engineering Lead and as a Production Engineering Manager. He has led globally distributed teams to deliver client facing projects across diverse industries. As a Production Engineering Manager, Alex led and grew the EMEA team and was responsible for special projects with some of Palantir's most critical clients. Prior to Palantir, Alex had hardware engineering internships at Nvidia and MSI. ### Company Launches [### Patchwork Technologies: Automated logging from code to storage]( [Patchwork]( automates logging from code to storage. Developers no longer need to manually write or re-write logs — Patchwork does it for them. Actionable, context-rich, secure logs can become the new standard across your entire codebase. The product currently works with Java, Go, Python, Ruby, TypeScript, JavaScript, Scala, and Kotlin. [ **The Problem with Logs** ------------------------- Manually instrumenting code with logging is time-consuming and tedious. Unlike building features, logging doesn’t earn promotions. Logs are written by developers but consumed by SREs and security engineers once in production. Picture this: a critical incident strikes at 2 AM. Your SREs are scrambling, frantically searching through a mess of inconsistent logs, trying to piece together what went wrong. Sound familiar? That’s because logging, as it stands today, is broken. Empires like Splunk have been built on the premise that messy logs are a fact of life and you therefore need to fork out for best-in-class indexing and querying. We say it’s time to tackle this problem at the source. **Patchwork: Logging, _Un_\-imagined** With Patchwork, developers can focus on building features while our product ensures that every log is context-rich, actionable, and secure. Developers can un-imagine logs, forever. Patchwork works in the background, reviewing your existing codebase and stepping in during continuous integration to ensure that your new code meets your engineering standards. Our product is flexible, but in our experience, good logs: * Are context-rich and actionable, * Explain what the failed method was attempting (the intent), * Provide the context in which it failed (the state of all relevant variables), * Use a structured format, * Handle sensitive information gracefully (never include tokens/passwords, label or exclude, e.g., personal information). Our mission is to provide a few-click, automated journey to implement logging from code through to storage. **Our Ask: Give an SRE a Hug** ------------------------------ And then, help them out by giving them our email – [[email protected]](mailto:[email protected]). We’d love to connect with Senior SREs and VPs of Engineering – in our experience, they really get it. Check out how we think at our [blog.]( **Our Story** ------------- The Patchwork founders ([Ben]( [Sam,]( and [Alex]( met while working at Palantir. That translates to _met while delivering product and outcomes to demanding customers, at massive data and infra scale, under tight timelines_. Late-night debugging sessions at Palantir taught us that good logs are priceless.

Rastro

rastro

https://rastro.sh

San Francisco, CA, USA

Rastro is a personal shopper for home decor. Our AI agents find you products from any retailer on the internet: we aim to build Perplexity for shopping. Built by ex-Shopify AI engineers, we're doing $20k, and growing, monthly gmv. Decorating your home can cost tens of thousands for an interior designer, or involve hours of scrolling retailer sites. Tell Rastro about your vision, and send our agents to source for you. Receive dozens of products in your inbox in minutes, buy only what you love, and pay no markup on retail. We make money on trade discounts and referral fees. Under the hood, we're building general purpose agents capable of sourcing products matching your taste and constraints from any retailer on the internet. Our algorithms consider images and descriptions, as well as product style, materials, brand and location.

AI personal shopper for home goods

Consumer

Consumer

S24

Active

Early

https://www.ycombinator.com/companies/rastro

https://yc-oss.github.io/api/batches/s24/rastro.json

Title: Rastro: AI personal shopper for home goods | Y Combinator URL Source: Markdown Content: ### AI personal shopper for home goods Rastro is a personal shopper for home decor. Our AI agents find you products from any retailer on the internet: we aim to build Perplexity for shopping. Built by ex-Shopify AI engineers, we're doing $20k, and growing, monthly gmv. Decorating your home can cost tens of thousands for an interior designer, or involve hours of scrolling retailer sites. Tell Rastro about your vision, and send our agents to source for you. Receive dozens of products in your inbox in minutes, buy only what you love, and pay no markup on retail. We make money on trade discounts and referral fees. Under the hood, we're building general purpose agents capable of sourcing products matching your taste and constraints from any retailer on the internet. Our algorithms consider images and descriptions, as well as product style, materials, brand and location. Rastro Founded:2024 Team Size:3 Location:San Francisco ### Active Founders ### Baptiste Cumin, Founder CEO @ Rastro: Amazon-search over every auction in Europe. Previously built Shopify's search + recommendations for 6 years as a Staff ML Engineer. ### Augustin Baudoin, Founder COO @ Rastro: Amazon-search over every auction in Europe. Previously Ops @ Uber and DS @ Criteo / French Gov. ### Papa Sougou Wele, Founder CTO @ Rastro, a search engine for art and antique auctions in Europe. Previously Backend engineer at Zapper and Shopify. ### Company Launches [### Rastro: Your personal interior designer]( Should it cost you 50k to decorate your home? --------------------------------------------- Decorating your home? You have 2 choices: hire an interior designer for $50k, or spend 50 hours doing it yourself. A beautiful home, time, money: pick two. Rastro is your hassle-free personal shopper and interior designer: 1. We craft your taste profile in a 15-minute call. 2. We find you matching decor from 46 sources. 3. We buy what you love, delivered to your door. Are you decorating your space? [**Book your 15-minute onboarding call here.**]( 💡 How we got here ------------------ In our first week at YC, we built a tool to help interior designers source furniture for less. Early on, a designer told us, “Getting an $8k chair for $4k isn’t material, we do $75k jobs.” Not sure their customer would agree, but it does explain their fee. Most of us know our taste, but scrolling dozens of websites is a waste of time. Niche sourcing channels are jealously guarded by professional dealers, and the best inventory is difficult to find. ✅ Rastro’s Solution ------------------- Rastro is your personal interior designer, with no up-front costs. 1. In an onboarding call, we craft a taste profile for you. Show us your space and tell us what you’re looking for. Share specific pieces you love, styles that inspire you, and how your home is laid out. 2. Our sourcing agents automatically scour through 46 different sites to find decor matching your tastes. We look through auctions and listing sites as well as classic sources like 1stDibs or West Elm. You receive our top picks and share your feedback directly on Rastro. 3. We deliver everything you purchase to your door, leveraging retailer partnerships and referrals to ensure a same-price guarantee. Let us help you decorate your home: book your onboarding call today at [**rastro.sh**](

Mica AI

mica-ai

https://usemica.com/

San Francisco, CA, USA

Mica builds AI Agents that automate B2B sales. Our agents analyze sales calls, generate tailored collateral, and follow up until the deal is closed. The founders met at UC Berkeley, and at our last jobs, we saw that the best sales reps weren’t just great at selling—they were elite content creators. They crafted tailored collateral like decks and highlight videos for each buyer, and their champions used this info to advocate for purchasing their product within their companies. However, because this takes more than 30 minutes per call, even the best reps could only do tailored follow-ups for a few clients, leaving millions in lost deals. Mica’s agents reduce that time to less than a minute. This lets top performers do A+ work on every deal and helps the rest of the team sell like a top performer. Sales collateral creation and follow-up is a $11b market in the US but this is just our beachhead. We’ll ship a fleet of agents that automate and eventually replace every other facet of the $80b enterprise sales market.

Al Agents that Automate B2B Sales

B2B

B2B -> Sales

S24

Active

Early

https://www.ycombinator.com/companies/mica-ai

https://yc-oss.github.io/api/batches/s24/mica-ai.json

Title: Mica AI: Al Agents that Automate B2B Sales | Y Combinator URL Source: Markdown Content: ### Al Agents that Automate B2B Sales Mica builds AI Agents that automate B2B sales. Our agents analyze sales calls, generate tailored collateral, and follow up until the deal is closed. The founders met at UC Berkeley, and at our last jobs, we saw that the best sales reps weren’t just great at selling—they were elite content creators. They crafted tailored collateral like decks and highlight videos for each buyer, and their champions used this info to advocate for purchasing their product within their companies. However, because this takes more than 30 minutes per call, even the best reps could only do tailored follow-ups for a few clients, leaving millions in lost deals. Mica’s agents reduce that time to less than a minute. This lets top performers do A+ work on every deal and helps the rest of the team sell like a top performer. Sales collateral creation and follow-up is a $11b market in the US but this is just our beachhead. We’ll ship a fleet of agents that automate and eventually replace every other facet of the $80b enterprise sales market. Mica AI Founded:2023 Team Size:3 Location:San Francisco ### Active Founders ### Achyuta Iyengar, Founder/CEO Achyuta is the co-founder and CEO of Mica AI. Prior to Mica, Achyuta worked as a Product Manager at Lob (YC S13) where he led the New Product Development, Address Verification, and Developer Experience teams. He has a BS in Business Administration from the Haas School of Business at UC Berkeley. ### Bharadwaj Swaminathan, Founder/Head of Eng Bharadwaj is the founder and Head of Engineering at Mica AI. Previously, he worked as a Data Scientist at Hewlett Packard Enterprise, working on creating machine learning products in the networking space. He has a Bachelor's in Data Science and Economics from UC Berkeley. ### Jai Yarlagadda, Founder/CTO Jai is the co-founder and CTO of Mica AI. Before Mica, Jai was Senior Machine Learning Engineer at GenerAI focused on building video understanding AI models, scaling software, and understanding client needs. Before that, he was a Software Engineer at Booz Allen Hamilton, building scalable software used by tens of thousands of users. He studied Physics and Computer Science at University of California, Berkeley. ### Company Launches [### Mica AI - AI Agents that Automate B2B Sales]( Hi everyone, we’re Achyuta, Jai, and Bharadwaj! **TLDR:** [**Mica**]( helps B2B sales teams close more deals by automating their entire sales follow-up process. Our AI agents analyze sales calls, generate tailored collateral, and follow up with prospects until the deal is closed. **The Problem:** The best sales reps aren’t just great at selling—they are elite content creators. They crafted tailored collateral like decks, 1-pagers, and product demos to help champions advocate internally. However, creating this content often takes over 30 minutes per call, and with multiple calls a day, even the best reps can only prioritize a few clients, leaving others with subpar follow-ups and millions in lost deals. **Our Solution:** Mica automates the sales follow-up process by generating deal-closing collateral in less than a minute. Our AI analyzes sales calls, extracts key insights, builds tailored collateral, and sends this to prospects in a centralized deal summary page. **Key Benefits** 1. **Quicker Follow-up Time:** Mica creates personalized follow ups in under 1 minute. This lets top performers do A+ work on every deal and helps the rest of the team sell like a top performer. 2. **Faster Sales Cycles:** The point of contact can easily review important information from the call, share it internally, and make decisions faster. **How it Works** 1. **Upload:** Mica automatically pulls in your sales call from existing recording or notetaking tools. 2. **Extract Key Insights:** Mica identifies the key points the prospect was most interested in. 3. **Generate Tailored Collateral:** Mica creates tailored decks, product demos, and 1-pagers for each key topic. 4. **Share:** Mica generates a shareable link containing all the collateral, which can be sent directly to the prospect in follow-up emails that Mica also creates. Here’s a demo of Mica that we created from an actual Mica sales call: [ **Our Ask:** We’re looking to onboard customers who want to automate and improve their sales follow-ups. We’re looking for: * Sales reps and heads of sales at B2B SaaS companies. * Founders who are doing founder-led sales. Anyone who is interested can book a demo [**here**](

Soff

soff

https://soff.ai

San Francisco, CA, USA

Soff is a supply chain operating system for manufacturing companies. We unify supply chain data and automate core workflows, such as data entry, requesting quotes, comparing quotes, and issuing purchase orders. Our customers save dozens of hours a week spent on repetitive manual tasks.

Supply chain operating system for manufacturing companies.

B2B

B2B -> Supply Chain and Logistics

S24

Active

Early

https://www.ycombinator.com/companies/soff

https://yc-oss.github.io/api/batches/s24/soff.json

Title: Soff: Supply Chain OS for manufacturing companies. | Y Combinator URL Source: Markdown Content: ### Supply Chain OS for manufacturing companies. Soff is a supply chain operating system for manufacturing companies. We unify supply chain data and automate core workflows, such as data entry, requesting quotes, comparing quotes, and issuing purchase orders. Our customers save dozens of hours a week spent on repetitive manual tasks. Soff Founded:2024 Team Size:2 Location:San Francisco ### Active Founders ### Raban von Spiegel, Founder Co-founder at Soff ### Berni Hausleitner, Founder Co-Founder at Soff ### Company Launches [### 🚛 Soff: Supply Chain Operating System for Manufacturing Companies]( **⛓️ TL;DR** [Soff]( is a supply chain operating system for manufacturing companies. We automate data entry, requesting quotes, comparing quotes, and issuing purchase orders. Our customers save dozens of hours a week spent on repetitive manual tasks. 👬 **Team** We met five years ago while studying at the CDTM in Munich. Most recently, [Berni]( worked as a full-stack engineer at Tesla on the Autopilot team, whilst [Raban]( finished his engineering master’s at Stanford. At Tesla, Berni built tons of internal tooling and saw the supply chain inefficiencies firsthand. During the batch, Raban worked as a purchasing manager for one of our customers. 🌋 **Problem** Supply chain data is spread across multiple platforms, such as ERPs, custom spreadsheets, and email inboxes, leading to inefficiencies and a lack of visibility. Supply Chain Managers spend hours each week searching through emails and spreadsheets to track the status of parts procurement, from request to quote to award. This process is not only time-consuming but also prone to errors and delays. For example, changes in production schedules require Supply Chain Managers to follow up with all suppliers individually to adjust their purchase orders. Quotes that aren’t used are lost in the abyss of email inboxes. Suppliers are reluctant to upload quotes to yet another portal and often fail to notify about delays, leading to costly production standstills. 🚛 **Solution** Soff connects suppliers, parts, and quotes into a single platform. We integrate with ERPs and email to automate supplier communication, send RFQs, and issue purchase orders. We turn unstructured email and PDFs into SQL-ready data, injecting quotes directly into the ERP system. We eliminate the need for manual data entry and dedicated supplier portals, thus increasing supplier response rates. Our platform stores quote data, ensuring no loss of information when switching suppliers or moving from single-source to dual-source. It also tracks parts from the initial request through to the final receipt, providing a complete overview for supplier management. Soff enables manufacturers to deliver higher-quality products faster. Companies like SpaceX (Warpdrive), Rivian (Basecamp), and Apple (Incline) spend millions on building their in-house supply chain tools—we bring that level of efficiency to any manufacturer. **❤️ How you can help** If you work in Supply Chain or work(ed) at a manufacturing company we’d love to connect. Send us a message at [[email protected]](mailto:[email protected]). 📨

Zigma - by NextUI

zigma-by-nextui

https://zigma.io

San Francisco, CA, USA; Remote

We began with an open-source library for React developers that enables them to create beautiful and accessible interfaces (NextUI). We are now building Zigma a complete frontend-as-a-service platform to make building the frontend part of products easier, allowing teams to focus more on their product and less on design systems and frontend complexities (zigma.io).

We connect your design files to production code in minutes

B2B

B2B -> Engineering, Product and Design

S24

Active

Early

https://www.ycombinator.com/companies/zigma-by-nextui

https://yc-oss.github.io/api/batches/s24/zigma-by-nextui.json

Title: Zigma - by NextUI: We connect your design files to production code in minutes | Y Combinator URL Source: Markdown Content: ### We connect your design files to production code in minutes We began with an open-source library for React developers that enables them to create beautiful and accessible interfaces (NextUI). We are now building Zigma a complete frontend-as-a-service platform to make building the frontend part of products easier, allowing teams to focus more on their product and less on design systems and frontend complexities (zigma.io). Zigma - by NextUI Founded:2023 Team Size:3 Location:San Francisco ### Active Founders ### Junior Garcia, Founder Junior Garcia is the Founder and CEO of Zigma and the creator of NextUI. With nearly 10 years of experience as a software engineer, including his time at Deel, he is passionate about creating design systems and frontend development, as well as contributing to open-source and enhancing user interfaces ### Company Launches [### Zigma by NextUI - Design systems, simplified]( **TL;DR** [**Zigma**]( simplifies creating and managing design systems, effortlessly syncing your design tokens between Figma and source code. With AI-powered updates and seamless synchronization, **Zigma** ensures consistency across all projects and speeds up the development, while reducing errors and manual handoffs. Our mission is to make design systems & frontend development accessible to everyone, regardless of expertise. **Team** I’m [Junior Garcia]( the founder of Zigma and creator of [NextUI]( I’ve spent nearly 10 years as a software engineer, including my time at Deel, working to improve design systems and frontend development for teams, with a passion for open-source and user interfaces. Zigma and NextUI were built with the dedicated effort of an amazing team that came to the company from the open-source community, including: [Tianen Pang]( [Ryo Matsukawa]( [Marcus Wong]( [Winches He]( [Tsuki]( [Daniel Oiteiro]( [Andres Gonzales]( **🚨 Problem** Design systems are often complex and challenging to maintain. Teams struggle with: * Reusability and scalability * Keeping alignment between design and development * Maintaining design consistency * Documentation that’s often incomplete or inaccessible **🚀 Solution: What is Zigma?** Zigma is the platform that bridges the gap between design and development. It allows teams to: * Create and manage design tokens easily * Use AI-powered updates to modify tokens quickly * Seamlessly sync tokens between Figma and source code to ensure consistency and save time Companies can now rebrand or update their design system effortlessly, with changes automatically applied across their designs and applications. **Video:** [ \*\*Watch this video to see the process:\*\* [ **✅ With Zigma you can:** * **Create and manage design tokens:** Easily organize tokens into groups and collections, covering everything from colors to typography. * **AI-powered updates:** Zigma’s AI suggests and applies token changes 10x faster than manual updates in Figma. * **Seamless synchronization:** Automatically sync design tokens between Figma and source code, ensuring team alignment without manual updates. With **CLI** and **GitHub** integration, Zigma streamlines the process. **We aim to save companies up to $500k by automating the creation, maintenance, and documentation of design systems** **📢 Ask** * We’re looking for early adopters to try out **Zigma**, provide feedback, and share how we can help you, contact us at [[email protected]]( or request a demo [here]( * Follow us on [X/Twitter](

Superfilter

superfilter

https://superfilter.ai

Remote

Superfilter is an email companion that manages your inbox like a human executive assistant would. We use AI to organize your high priority and actionable emails, summarizes you latest inbox activity, give you enough context to help you respond quickly, and take low leverage tasks like coordinating meetings and signing documents off of your plate. It's a reliable companion for founders who value responsiveness without sacrificing focus and peace of mind. Superfilter is always working, improves as it learns from you, and will only leave to work for another person if you cancel your subscription.

Expert assistant that manages your email.

Consumer

Consumer

S24

Active

Early

https://www.ycombinator.com/companies/superfilter

https://yc-oss.github.io/api/batches/s24/superfilter.json

Title: Superfilter: Expert assistant that manages your email. | Y Combinator URL Source: Markdown Content: ### Expert assistant that manages your email. Superfilter is an email companion that manages your inbox like a human executive assistant would. We use AI to organize your high priority and actionable emails, summarizes you latest inbox activity, give you enough context to help you respond quickly, and take low leverage tasks like coordinating meetings and signing documents off of your plate. It's a reliable companion for founders who value responsiveness without sacrificing focus and peace of mind. Superfilter is always working, improves as it learns from you, and will only leave to work for another person if you cancel your subscription. Superfilter Founded:2024 Team Size:3 Location: ### Active Founders ### Albrey Brown, Founder Co-founder at Superfilter. I'm really good at helping early stage start ups go to market. I built and sold my second company in 2016. Since then I've had the pleasure of working with amazing people at startups like DocuSign, and Airtable. Superfilter is an assistant that saves you time. Mainly by managing your email inbox. Send me an email if you'd like to use it or want to talk sales, marketing, parenting, or crossfit. ### Scott Moss, Founder Product eng and former VC that loves making amazing experiences that solve obvious problems. ### Travis Ueki, Founder Co-Founder of Superfilter, Previously Tipe (W18), Coinbase (S12) Obsessed with building product. Born and raised on Oahu, Hawaii, my journey’s been anything but conventional. ### Company Launches [### Superfilter - An executive assistant in your pocket]( **TLDR;** We built a companion that manages your inbox with the intelligence of an executive assistant. It never misses an important email, suggests follow up actions, and handles things like meeting coordination and quick replies in one click. [**Sign up here to get access.**]( ### **Who we are?** Hey everyone, we’re [Scott]( [Travis]( and [Albrey]( We’re serial entrepreneurs who’ve worked together for almost a decade. Over various roles, we discovered the worst part of work is managing our email inbox. So we built an [assistant]( that handles it for us and are now offering it to you. **The Problem** --------------- In [_A World Without Email,_]( Cal Newport explains that the average knowledge worker spends 30% of their workday managing their inboxes rather than engaging in meaningful work. Cutting this time in half could unlock hundreds of hours of productivity. We interviewed founders and executives to understand how they deal with this. We typically heard one of three strategies: 👀 Ignoring email altogether. Super efficient, but results missed opportunities and annoyed customers and friends. 🤝🏿 Hiring an executive assistant. Saves a lot of time, but is very expensive. ✉️ Clients like Superhuman. Affordable, but marginally more efficient than traditional email clients. We couldn’t find a solution that was as reliable as an executive assistant and as affordable as a suped-up email client, so we set out to build it. **Enter Superfilter** --------------------- Superfilter doesn’t replace your email client. Rather — it connects to your inbox and manages it like your executive assistant. While you’ll use email for strategic work, Superfilter keeps you on top of the tactical stuff. Here are some of its capabilities users love so far: **Overviews make sure you never miss an important email** **Feeds keep track of your to-do’s and FYI’s** **Tasks let you delegate repetitive work to Superfilter** Superfilter’s purpose is to give you back time and peace of mind. It’s always working, improves as it learns from you, and will only leave to work for another person if you cancel your subscription. **The Ask** ----------- Use the product and give us feedback. We only support iPhone and Gmail (Superhuman and Hey included). * [**Sign up on our waitlist for early access**]( * **Follow us:** Give us a follow on [Twitter &]( [LinkedIn](

Usul

usul

https://www.usul.co/

San Francisco, CA, USA

Usul helps contractors win more DoD contracts. We do this by using LLMs to interpret government policy, track funding lines, and find new business opportunities.

AI Insights for Defense Contractors

B2B

B2B

S24

Active

Early

https://www.ycombinator.com/companies/usul

https://yc-oss.github.io/api/batches/s24/usul.json

Title: Usul: Win Defense Contracts with AI. | Y Combinator URL Source: Markdown Content: ### Win Defense Contracts with AI. Usul helps companies win defense contracts with AI. We do this by using LLMs to interpret government policy, track funding lines, and find new business opportunities. Usul Founded:2024 Team Size:4 Location:San Francisco ### Active Founders ### Jarren Reid, Founder CEO & Co-Founder @ USUL | Prev: Stanford CS, Palantir, MITRE ### Oliver Gomez, Founder CPO & Co-Founder @ USUL | Prev: Stanford CS AI + Econ, Palantir, Product at ex-YC company ### Joonghyun Lee, Founder Chief Engineer & COO @ USUL (S24). Ex: UMich ECE and Stanford Aero/Astro. ### Pranav Sriram, Founder CTO & Co-Founder @ USUL. Prev: ML Research at Apple, NASA, Stanford Vision and Learning Lab ### Company Launches [### Usul - AI for defense contractors 🇺🇸]( ### **TLDR:** [Usul]( allows defense contractors to easily interpret government policy, track funding lines across budgetary documents, and strategize the best opportunities for new business. ### **Problem:** Winning DoD contracts can take years, even if you’re well-connected and highly familiar with government acquisition processes. Unfortunately, many companies lack both and go out of business before ever landing their first contract. Adding fuel to the fire, critical government information is randomly dispersed across a multitude of archaic websites, databases, and spreadsheets, making it even harder for new entrants to compete against established primes. As a result, defense contractors are currently spending weeks finding answers to crucial policy, budgetary, and contracting questions and are completely neglected in terms of tooling. ### **Solution:** We allow defense contractors to quickly track and query complex DoD demand signals under a unified workspace. Contractors can now: * Identify novel contract opportunities * Track DoD funding lines and analyze project budgetary trends * Identify relevant acquisition and compliance regulations * Find the most relevant DoD buyers for existing capabilities All in one place. ### **Coming Soon:** Stake your own company data against government demand signals to: * Predict future funding lines for your capabilities * Identify personalized avenues for future opportunities * Pull custom acquisition and compliance matrices ### **About Us:** We’re a group of AI nerds who’ve all worked for government contractors like Palantir, Apple, Reliable Robotics, etc. We met in a class at Stanford called Hacking 4 Defense, where we talked to 200+ people in the DoD/defense sector about their biggest problems. Our insights from these incredible conversations inspired us to form Usul. And yes, it’s a Dune reference… [Jarren Reid]( left high school to start his first company. He then worked for 4 government contractors (Palantir, MITRE, etc.) and dropped out of Stanford to help his past colleagues win more contracts. [Oliver Gomez]( started building as a PM at Finnt (YC W22). He then worked at Palantir and leveraged LLMs for military operational legal review at Stanford’s Gordian Knot Center. He dropped out last quarter. [Joonghyun Lee]( is a commercial pilot who has been working on autonomous aircraft research for the past 3 years. He has experienced numerous pain points dealing with regulation across different government agencies. [Pranav Sriram]( has been engrossed in AI research for the past four years at Apple and the Stanford Vision and Learning Lab. He previously worked at NASA and saw firsthand how time-consuming getting answers to basic technology, security, and policy questions across government could be. ### **Our Ask:** We’re Live! Learn more about what we’re building at [usul.co]( If you know any: * Business Development or Capture teams at Defense Contractors * Defense/Government consultants * Companies looking to expand to Defense/Government We’d love to chat! Please feel free to contact us at [[email protected]](mailto:[email protected]) or book a demo [here]( #### Company Photo

The Forecasting Company

the-forecasting-company

https://www.theforecastingcompany.com

Paris, Île-de-France, France

The Forecasting Company builds planning systems based on our in-house foundation models for time series. We are starting with demand forecasting for large retailers.

We build foundation models for time series.

B2B

B2B

S24

Active

Early

https://www.ycombinator.com/companies/the-forecasting-company

https://yc-oss.github.io/api/batches/s24/the-forecasting-company.json

Title: The Forecasting Company: Foundation models for time series | Y Combinator URL Source: Markdown Content: ### Foundation models for time series The Forecasting Company builds planning systems based on our in-house foundation models for time series. The Forecasting Company Founded:2024 Team Size:2 Location:Paris, France ### Active Founders ### Geoffrey Negiar, Founder & CEO Co-founder and CEO at The Forecasting Company, helping enterprise customers improve their operational decision making. PhD grad from UC Berkeley's BAIR. I spent time with Amazon Forecasting Science in SCOT, Google Brain, and Bloomberg LP where I worked on forecasting, optimization and deep learning research. ### Joachim Fainberg, Founder & CTO PhD in ML for speech and full-stack developer. Interned at Sonos where I built an on-device wake word detector and at Bloomberg building a large ASR system. Worked full-time at JPMorgan in their Machine Learning Center of Excellence in New York, and then at two startups, team lead ML/NLP at Consigli, a prop-tech startup, and sr algorithm engineer and full-stack at Vind AI, a SaaS for designing off-shore wind parks. ### Company Launches [### The Forecasting Company: Foundation models for time series]( **tl;dr:** The same core technology that led to the LLM explosion can be adapted to build accurate and easy to use time series models. We’re building the enterprise-grade, plug and play forecasting systems of tomorrow. [Talk to us]( if you have forecasting and planning use-cases! Hi everyone, we’re Geoff and Joachim. We’ve built ML and forecasting systems at Amazon, Google, Bloomberg LP, JP Morgan, and Sonos. We’re The Forecasting Company! ### Problem 📉 * Existing time-series forecasting solutions produce **unreliable predictions** with sometimes catastrophic results - they have a hard time handling activity spikes like Black Friday. * Developing and maintaining useful forecasting and planning models requires **large and costly teams** of data scientists and domain experts. * Data science teams are backlogged, leaving operators **waiting for months** before a use-case to be put into production. ### Solution 📈 * **An easy to use, flexible and accurate forecasting system. No training required!** * A **massive training dataset of time-stamped data** from many industries and domains (energy, transportation, etc.). * A forecasting model that can leverage **user-provided input variables** or our own **curated data streams**. ### What we’re forecasting these days Bacteria levels in the Seine 🦠 🏊‍♂️ ------------------------------------- Our forecasts show that swimmers lucked out today 🍀! (Data provided courtesy of the [Fluidion Open Data Initiative)]( ### Ask * If you have forecasting problems in-house, or know of someone who does, **please reach out - we’d love to help!** * **Try our sandbox model** on [ and let us know what you think [[email protected]](mailto:[email protected]) ### YC S24 Application Video

Exa Laboratories

exa-laboratories

https://exalaboratories.com

San Francisco, CA, USA

Exa is making reconfigurable chips for AI, offering superior speed and energy efficiency compared to traditional GPUs/TPUs/LPUs. Our chips automatically adapt themselves to each specific AI model, overcoming the von Neumann bottleneck, (a common limitation in standard systems). By optimizing for each AI architecture, we significantly boost both inference and training speeds while reducing energy consumption. Imagine having hardware tailored to your AI model, whilst still maintaining the flexibility of a GPU. This way, you won't need to re-manufacture your hardware each time you change your model architecture, all while still benefiting from the speed and efficiency of specialized hardware, saving data centers and compute clusters hundreds of millions of dollars every year!

Energy efficient chips for AI

Industrials

Industrials

S24

Active

Early

https://www.ycombinator.com/companies/exa-laboratories

https://yc-oss.github.io/api/batches/s24/exa-laboratories.json

Title: Exa Laboratories: Energy efficient chips for AI | Y Combinator URL Source: Markdown Content: ### Energy efficient chips for AI Exa is making reconfigurable chips for AI, offering superior speed and energy efficiency compared to traditional GPUs/TPUs/LPUs. Our chips automatically adapt themselves to each specific AI model, overcoming the von Neumann bottleneck, (a common limitation in standard systems). By optimizing for each AI architecture, we significantly boost both inference and training speeds while reducing energy consumption. Imagine having hardware tailored to your AI model, whilst still maintaining the flexibility of a GPU. This way, you won't need to re-manufacture your hardware each time you change your model architecture, all while still benefiting from the speed and efficiency of specialized hardware, saving data centers and compute clusters hundreds of millions of dollars every year! Exa Laboratories Founded:2024 Team Size:2 Location:San Francisco ### Active Founders ### Elias Almqvist, Founder/CEO Elias is the CEO of Exa Laboratories. A self-taught engineer since age 9, he studied CS & CE at Chalmers University of Technology, worked part-time as an embedded software engineer, and was involved in various PhD aerospace projects as an undergrad. After dropping out in his second year, he founded Exa to accelerate AI innovation and elevate human consciousness. Greater and more sustainable compute drives better AI and scientific breakthroughs. ### Prithvi Raj, Founder CTO and co-founder of Exa Laboratories. Before Exa, I was pursuing my MEng as part of a world leading lab, (the Computational Stats & ML Lab at Cambridge). Here, I fell in love with scientific machine learning - a field that requires bespoke neural network architectures and extreme hardware efficiency. I'm here to provide these so that AI can start solving the game-changing problems that it was originally promised to address. ### Company Launches [### Exa Laboratories - 27.6x more efficient chips for AI 🔥🚀]( ### TL;DR: We're building reconfigurable chips for AI that are up to 27.6x more efficient and powerful than the H100 GPUs. This could save data centers hundreds of millions to billions in annual energy costs. ### Meet the Team Hello! We're [Elias]( and [Prithvi]( from [Exa]( We're developing **reconfigurable chips for AI** that are up to around **27.6x\* more efficient and performant** than the modern H100 GPUs. \*: [_Read our litepaper!_]( [**CEO, Elias Almqvist**]( (right): Self-taught engineer who also studied computer science and computer engineering (dropped out and founded Exa, btw) at Chalmers University of Technology. Previously worked in the embedded software space but also worked on various aerospace projects at university. [**CTO, Prithvi Raj**]( (left): Holds an MEng from the world-leading Computational Stats & ML Lab at Cambridge. During his time there, he fell in love with scientific machine learning, a field that demands bespoke neural network architectures and extreme hardware efficiency, and also interned at Microsoft as a software engineer. ### The problems! The AI industry faces critical challenges threatening its sustainable growth: 1. **Unsustainable Energy Consumption**: Modern GPUs consume 600-1000 W per unit, creating massive scaling issues for data centers. Large data centers face energy costs in the hundreds of millions to potentially billions each year. **GPU power draw seems to be increasing with each new release**, while **compute per area has remained the same** for the past 5 years. 2. **Exponential Compute Demand**: With AI advancements, computational power demand is rapidly increasing. Unchecked, this trend could lead to an energy crisis, impeding AI progress and costing data centers billions of dollars. 3. **Hardware Limitations**: **Current fixed architectures constrain AI innovation**. They lack the versatility to efficiently support diverse AI architectures and custom neural network designs crucial for solving real-world problems. ### The solution. Exa's polymorphic computing technology addresses these challenges: * **Reconfigures for each AI model architecture**, maximizing efficiency and versatility * Supports diverse approaches, from **transformers** and **GPTs** to novel AI architectures (e.g., the new [**Kolmogorv-Arnold Networks (KAN)**]( * Early simulations indicate potential efficiency gains of up to **27.6x** over the H100 GPUs **This technology could save data centers hundreds of millions to billions in annual energy costs**, significantly reducing operational expenses and environmental impact. For a somewhat deeper technical dive, refer to our [litepaper]( ### Asks :) * Read our [litepaper]( All feedback welcome! * Introduce us to anyone in the scientific machine learning space and/or someone conducting research in AI, **particularly those who have very “_cursed model architectures.”_** * Get us in contact with any data center, AI research organization, or GPU cloud provider (i.e., AWS, OpenAI, Anthropic, DeepMind, Lambda). * Give us intros to semiconductor industry professionals, particularly those interested in **bringing back chip manufacturing to the US**! Feel free to reach us at [[email protected]](mailto:[email protected]), **we would love to hear your feedback and answer your questions!** #### Company Photo ### Hear from the founders #### How did your company get started? (i.e., How did the founders meet? How did you come up with the idea? How did you decide to be a founder?) Elias, originally from Sweden, was motivated by the challenge of running large language models on FPGAs to combat the high power consumption of modern GPUs. During a week-long event in London, Elias met Prithvi, whose expertise in electrical engineering, generative AI, and scientific research perfectly complemented his own. They quickly recognized that advancing science and technology required a new computing paradigm - one that was both more powerful and sustainable. Driven by a shared passion for advancing humanity and accelerating technological progress, Elias, who had recently dropped out of university, and Prithvi, who had just graduated, founded Exa to revolutionize computing and push the boundaries of (artificial) intelligence. #### What is your long-term vision? If you truly succeed, what will be different about the world? Exa's long-term vision is to eliminate hardware constraints in computing and artificial intelligence by creating hardware that is **fully reconfigurable**. This would remove the need to replace underlying hardware to support the latest AI models - a crucial consideration given the anticipated chip and silicon shortages in the coming decades. Given its reconfigurability, our hardware would therefore open up new possibilities in scientific machine learning (_beyond just sentence generation_), where flexible AI design is essential. This capability could lead to the **discovery of new equations** and **scientific breakthroughs**, fulfilling one of the original promises of artificial intelligence and machine learning in advancing human knowledge, and solving the worlds most pressing problems. We aim to enable individuals, organizations, and governments to run large-scale AI models on their own hardware, making AI **fully decentralized** and **safe** while significantly reducing energy costs. With modern GPUs, this would be impossible. As the AI revolution accelerates, models are becoming larger, and AI usage is surging exponentially. This trend is leading data centers to consume increasingly massive amounts of energy, potentially requiring the installation of dedicated power plants to meet demand in the future. For humanity to advance, we must discover a more sustainable way to run AI models. Exa is dedicated to providing this solution. Without it, we risk excessive energy consumption, which could potentially lead to the end of humanity as we know it. _Exa computing is the most optimal way forward._

Lighthouz AI

lighthouz-ai

https://lighthouz.ai/

San Francisco, CA, USA; Remote

Lighthouz AI is building AI procurement specialists for manufacturers. 500,000 procurement professionals waste their entire day using emails to communicate with suppliers, and entering data on spreadsheets and 90s looking ERP software. That's $40B wasted annually. Lighthouz's AI procurement specialists disrupts this entire profession. The AI agent automates all manual tasks of procurement -- data extraction from emails & documents, data entry on ERP systems. It 15x'es the productivity of procurement teams. Our AI agents will replace 500k jobs in the next 5 years. Our team has 20+ years of combined AI experience: - Srijan has been a GaTech CS professor, ex- Google, Stanford, UMD, IIT. He has built systems now in production at Meta, Home Depot, etc. He has written 60+ papers that are cited 6000+ times. - Sonali was a lead AI engineer at Progressive insurance, American Family insurance, Halliburton, did her MS at University of Washington, and undergrad at IIT.

AI procurement specialists to 15x productivity