@ycombinator
We help founders make something people want. Subscribe to our newsletter: https://t.co/sjqjxxBeLc
Paul Graham on what surprised him most about AI: “The way we thought it was all going to play out was you start out with a perfect fly.” “Instead, what we got was basically a full-on human, but full of shit.” “Instead of starting with perfect and then working your way up to human, you start with human and then work your way towards perfect.”
Congrats to @hylight_aero (S23) on their $14.5M raise! They’re bringing back the airship, this time autonomous, hydrogen-powered, and built to inspect critical infrastructure. Power lines, pipelines, and railways are still mostly inspected by helicopter, which is expensive, polluting, and risky for crews. HyLight’s 12-meter airships hover low and slow with LiDAR, cameras, and methane detectors, capturing imagery down to 1 millimeter per pixel. Four are flying missions in France today, with 15 planned to operate at once by 2027. t.co/g4CfPmyQot
Jeetu Patel is Cisco’s President and Chief Product Officer, helping turn one of the world’s largest technology companies into what he calls “the world’s largest startup.” Cisco is now positioning itself as the picks-and-shovels company for the AI era, spanning networking, security, silicon, and infrastructure. Its AI infrastructure business went from zero to $9.3B in orders in just two years. In this episode of Main Function, Jeetu joins YC's Jared Friedman to talk about bringing a founder mindset to the company, and what the massive data center buildout looks like from the inside. He also shares the unlikely path that took him from waiting tables at Sizzler to leading product at Cisco, and the lessons he’s learned about ambition, compounding, and making the most of the opportunities in front of you. 00:00 — Cisco’s Bet on AI 03:38 — Running Cisco Like a Startup 06:45 — The Infrastructure AI Agents Need 10:00 — Getting 32,000 Engineers to Embrace AI 14:07 — Bringing AI to Hardware and Silicon 17:15 — Inside the AI Data Center Buildout 21:27 — Are We Overbuilding AI Infrastructure? 25:10 — Building Trust in AI Agents 28:18 — From India to Waiting Tables at Sizzler 32:12 — Choosing Learning Over Ego 35:31 — Ambition, Teamwork, and Compounding 39:31 — Don’t Squander Your Opportunity
Tune in: youtu.be/ofxAtQhV-xA
Congrats to @vitalize_care (YC W23) on their $31M Series A! They're building the AI operating system for hospital staffing, scheduling, and capacity. Nurse managers often lose two to four hours every day building schedules by hand, time that could be going to patients. Vitalize now manages millions of shifts a week across 15+ health systems, saving clinical leaders more than 2,000 hours a day. St. Luke's Health cut its overtime and agency labor spend by 54% within 12 weeks. t.co/tAeMC5tJMo
Congrats to @multiplylabs (YC S16) on their $75M Series B! They build robots that manufacture the world's most complex drugs, including gene therapies and mRNA treatments. AI is speeding up drug discovery, but someone still has to physically make the medicines, and today that work mostly happens by hand. Manual production is slow, vulnerable to contamination, and can cost more than $1 million per dose for some gene therapies. Multiply's robots run that production inside facilities at AstraZeneca and Legend Biotech. t.co/O3Rf1tZGrH
Congrats to @sensible_bio (YC S21) on their $47M Series A! They engineer living cells to manufacture mRNA, turning the cell itself into the factory. Today's mRNA is made in cell-free processes that work fine for vaccines, where doses are small. The next generation of mRNA medicines, like in vivo CAR-T and gene editing, needs far higher doses and far purer mRNA than those methods can deliver. Sensible is closing that gap with a platform that can already produce around 1,000 mRNA sequences per week. t.co/zZiEHJpHx9
Congrats to @kiwicopple, @AntWilson, and @supabase on their $150 million round at a $10.65 billion valuation! More than 13 million developers use Supabase as the backend for their apps. They're adding more than 1 million users and over 4 million databases every month, and 70% of new databases are now created by agents or AI-driven tools. t.co/EKUQrA00q5
Applied Electrodynamics is building cameras that can see through walls and other opaque materials. Using radio waves, they can locate studs, pipes, cables, and junction boxes behind surfaces like drywall or marble and turn it into a dimensionally accurate 3D model.
Gumloop (YC W24) is building a platform that lets employees across a company build and share AI agents, while giving IT control over the data, permissions, and infrastructure behind them. Today it’s used by companies including Shopify, Gusto, and Instacart to automate work across sales, customer success, operations, and more. In this fireside, @gumloop co-founders @MaxBrodeurUrbas and @rbehal1729 joined @gustaf to share how they built Gumloop before models were ready for agents, then found product-market fit as the technology caught up. They talk about landing early enterprise customers, making companies AI-native, and why the people closest to the work should be the ones automating it. 00:00 — The AI Agent Builder for Every Team 03:38 — From Workflow Automation to Enterprise AI 07:20 — What It Takes to Win Enterprise Customers 11:02 — How the Founders Met and Got Into YC 15:00 — How Gumloop Uses Its Own Agents 18:02 — Landing Shopify and Rethinking Pricing 21:23 — Fundraising and the Limits of Staying Lean 24:35 — Letting Employees Automate Their Own Work 27:43 — Building Agents Companies Can Control 31:23 — Getting IT to Say Yes
This week’s Paper Club is all about alternative compute. Modern AI has been shaped by a tight coupling between transformers, backpropagation, and GPUs. But as compute efficiency starts to plateau, there may be very different ways to build the machines that come next. We explore what AI compute could look like beyond conventional digital hardware through optical systems that use light to perform computation, neuromorphic architectures inspired by the brain, and experiments teaching living brain cells to play Doom. 00:07 — @FrancoisChauba1: Why Alternative Compute? 03:38 — Beyond Backpropagation 08:08 — Zero-Order Optimization and SOMA 15:06 — Ilker Oguz: Computing With Light 19:27 — Why Optical Computing Isn’t Everywhere Yet 22:23 — Building a Diffusion Model With Light 30:03 — Optical Computing Q&A 39:00 — @AlokVasudev: What Is Neuromorphic Computing? 45:23 — What the Brain Can Teach Us About Chips 49:35 — Where Neuromorphic Computing Stands Today 54:24 — Neuromorphic Computing Q&A 1:04:24 — @vytalow: Teaching Brain Cells to Play Doom 1:07:48 — Training Biological Neurons With Reinforcement Learning 1:12:28 — How Brain Cells Learn From Feedback 1:15:40 — Biological Computing Q&A
Tune in: youtu.be/xc2FTBGRSJo
Startup School is coming to Taipei on November 12! 🇹🇼 Hear from founders like @andyfang of DoorDash, @jeffchanginvest of Vest, @moseslo of Xendit, @mchiang0610 of Ollama, and more. And join builders and hackers from across Asia for a day of talks and sessions with YC partners.
RSVP here: events.ycombinator.com/startup-school…
Startup School is coming to Taipei on November 12! 🇹🇼 Hear from @andyfang of DoorDash, @jeffchanginvest of Vest, @moseslo of Xendit, @mchiang0610 of Ollama, and more. Join the best builders & hackers from across Taiwan and Asia for a day of talks and sessions with YC partners.
RSVP to attend: events.ycombinator.com/startup-school…
We’re excited to welcome @vivianmshen and @raphaelschaad as YC’s newest General Partners! Vivian and Raphael first came to YC as founders. Vivian co-founded Juni Learning (W18), which teaches kids coding and math, and later founded Acely, an AI tutor for college prep. Raphael founded Cron (W20), the calendar app that became Notion Calendar. They built products used by millions of people, then returned to YC as Visiting Partners to help other founders. Now they're doing it full-time. t.co/8dvzPwxZlU
Halluminate is building reinforcement learning environments and benchmarks to help AI models do knowledge work beyond coding, starting with finance. With a team of fewer than 10 people, the company works with four of the top five closed-source US AI labs and recently raised a $30M Series A. In this fireside, co-founders @Jerr_Wu and @wgm752 sit down with YC's @xuster to share how testing browser agents led them to building the environments models now use to practice tasks like financial modeling. They explain why better verification, not just more training tasks, is critical to making models smarter, and how that work could eventually lead to simulated companies where teams of AI agents learn to work together. 00:00 — Halluminate’s Pivot From Evals to Training Environments 03:48 — Teaching AI to Do Financial Work 07:13 — Why Verification Matters More Than Volume 10:13 — From Finance to Simulated Companies 13:30 — AI Safety and Building the Team
Live from DC with NASA Administrator Jared Isaacman (@NASAAdmin) x.com/i/broadcasts/1…
We're hosting an evening in SF with @oakhcft and π for early-stage builders in physical AI: robotics, autonomy, and intelligent hardware. Join us for a fireside chat with @physical_int co-founder @QuanVng on what it takes to build in the physical world. events.ycombinator.com/yc-oak-happyho…
The Startup Industrial Base: Building for the Next 250 — Washington, D.C. x.com/i/broadcasts/1…
Tensr is building robotic factories that build robots. Their 12,000-square-foot factory is already making humanoids, end effectors, and even robots for the International Space Station. Every order becomes training data for the factory to manufacture the next one better.
“Tradeoffs that only pay off in 10 years will usually be underpriced.” Play the long game. When you think in decades rather than quarters, you’ll see that your potential customers have trajectories, not fixed sizes. Paul Graham on making startups more powerful: t.co/ZKNMGomfnn
One of the biggest surprises in AI over the last few years has been how well coding agents generalize beyond software. In a recent essay, MIT professor Philip Isola argued that we may be entering the era of robot-use agents: general-purpose models that can control different robots, write policies, and learn new physical tasks with little or no robot-specific training. In this episode of Decoded, we're joined by the founders of @theWaddleLabs and @Robocurve, two of the startups whose work helped drive this realization. They're working at the frontier of using general-purpose models to control robots, and their recent demos helped inspire the growing conversation around robot-use agents. Together, we dig into the research behind that idea, from code-as-policies and vision-language-action models to the harnesses and evals needed to make these systems work in the real world. 00:00 — What Are Robot-Use Agents? 02:11 — From Vision-Language-Action Models to General LLMs 04:07 — The Bitter Lesson for Robotics 07:04 — How Coding Agents Learned to Control Robots 10:41 — How Robots Learn From Experience 14:22 — The Harness as a Form of Robot Intelligence 16:14 — Watching Astra Control a Robot 20:55 — Why General Models May Win in Robotics 26:05 — How Close Are General-Purpose Robots?
Hop Aero is building rockets that can deliver cargo anywhere on Earth in minutes. They’re starting with military logistics and other time-sensitive payloads, including temperature-controlled materials with short shelf lives, where getting something there much faster can materially change the outcome. They’ve already built and successfully hot-fired a new rocket engine, and signed $1.37B in commercial LOIs.
Vapi (@Vapi_AI) is the platform for building and deploying voice AI agents. It now serves about a billion calls a year for companies including Amazon, Uber, and Intuit, helping businesses automate phone calls with AI. Voice is becoming the default interface between businesses and their customers, but getting agents reliable enough for production is still much harder than it looks. In this episode of Founder Firesides, co-founders @jordan_dearsley and @nikhilro_ sit down with @gustaf to share the long path to product-market fit, from pivoting more than a dozen times to building an AI therapist that eventually became the foundation for Vapi. They explain why voice was the first idea that felt inevitable, and how sticking with that conviction before the technology was ready eventually paid off. 00:05 — Powering a billion calls a year 02:12 — Why voice AI suddenly started working 04:57 — More than a dozen pivots 09:08 — From calendar app to AI note taker 13:23 — Starting over in San Francisco 14:14 — The AI therapist that became Vapi 17:56 — Finding PMF 22:47 — Why the model labs won’t replace Vapi 26:49 — The future of voice AI
A lot of construction work is still done by hand simply because the machines to automate it don’t exist. That’s become a bottleneck for building things like solar farms, data centers, and other critical infrastructure. @cosmic_robotics is building those machines. Their robots have already installed 11,000 solar panels, and they just signed a major contract with the largest construction company in the country. They’re building the construction robots we’ll need on Earth today, and eventually on the Moon and Mars. @JamesAEmerick
Congrats to Alex, Ben, and @hubble_network (YC W22) on their $200M Series C at a $1.6B valuation! Hubble's satellite network lets any Bluetooth device connect directly to orbit with just a software update. Cellular covers about 15% of Earth's landmass, and tracking anything beyond that has meant satellite hardware costing $25 or more per device. Hubble brings connectivity under 50 cents per device, which changes the math for billions of pallets, shipping containers, water pumps, and farm equipment that were never worth connecting before.
Congrats to @SpamRoss, @mduvall_, and @numeral (YC W23) on their $100M Series C! Numeral automates sales tax compliance end to end, from registrations and calculations to filings and payments. They handle US sales tax plus VAT and GST across more than 90 countries, all in one system. More than 3,500 businesses run their compliance on Numeral, including Supabase, Eight Sleep, and Graza, with transaction volume up 327% in the past year.
Attack drones are reshaping modern warfare, but many countries still lack an effective way to defend their airspace. Earendil Robotics is building autonomous interceptors, along with the software and training needed to stand up that capability from scratch. Their systems can take off, transit, and intercept targets on their own, reducing months of pilot training to a few clicks. The goal is air defense that’s affordable enough to deploy in large numbers and simple enough to operate at scale.
Attack drones are reshaping modern warfare, but many countries still lack an effective way to defend their airspace. Earendil Robotics is building autonomous interceptors, along with the software and training needed to stand up that capability from scratch. Their systems can take off, transit, and intercept targets on their own, reducing months of pilot training to a few clicks. The goal is air defense that’s affordable enough to deploy in large numbers and simple enough to operate at scale.
Attack drones are reshaping modern warfare, but many countries still lack an effective way to defend their airspace. Earendil Robotics is building autonomous interceptors, along with the software and training needed to stand up that capability from scratch. Their systems can take off, transit, and intercept targets on their own, reducing months of pilot training to a few clicks. The goal is air defense that’s affordable enough to deploy in large numbers and simple enough to operate at scale.
The YC Early Access Network is filling up fast. On Oct 22, tech leaders from high-growth startups to the Fortune 500 meet the best companies building enterprise AI. Be in the room. Request one of the remaining spots in Early Access now. events.ycombinator.com/yc-early-acces…
Congrats to @CalebPeffer, @ericciarla, @nickscamara_, and @firecrawl on their $75M Series B! Firecrawl turns the web into clean, structured data for AI agents, with more than 1.5 million people building on their API. They're also launching Alexandria, a knowledge library where agents can search the live web, official data providers, and dedicated indexes of scientific papers, developer docs, and government records. Providers get paid when agents use their data, and Firecrawl plans to open that up to anyone with knowledge worth sharing. t.co/sScCgFgvkZ
Confido (@confidotech) is building the AI workforce for consumer brands, automating everything from accounting to sales and demand planning. They work with brands like Olipop and Dude Wipes, as well as divisions of Mars and Unilever, and recently raised a $55 million Series B led by Insight. In this episode of Founder Firesides, Confido co-founders Justin Hunter and Kara Holinski join @sdianahu to share how they found a huge opportunity in an industry still running on spreadsheets and manual processes, and why selling the work rather than the software helped them win large enterprise customers. 00:04 — What Confido Does 01:16 — A Year of Pivoting 03:14 — Finding a Painful Problem in CPG 04:16 — Selling the Work, Not the Software 07:13 — The Future of Confido
Congrats to @angle_health (YC W20) on their $200M Series C at a $2.7B valuation! They're an AI-native health benefits platform built for small businesses. Health insurance is one of the hardest, most regulated industries a startup can enter, and Angle is already profitable on a net-income basis with revenue more than doubling year over year. They oversee nearly $1 billion in annual premium equivalents for more than 5,000 small businesses, some with as few as two employees. t.co/z3Wreizo9f
Kastle is building AI employees for banks. Its agents automate operations across mortgage servicing, consumer lending, and other back-office functions. Today, Kastle works with 10 of the top 25 mortgage servicers and has processed more than $2 billion in transactions. The company recently raised a $24M Series A, just two years after going through YC. In this episode of Founder Firesides, founders @therishic and Nitish join @sdianahu to share why they started over just one month before Demo Day, how they found their way into one of the hardest industries for deploying AI, and what it takes to make agents reliable enough to handle real financial transactions. 00:05 — What Kastle Does 01:00 — Why Banks Need AI Employees 02:50 — Going Back to Zero Before Demo Day 05:00 — Finding the Idea for Kastle 06:43 — Landing Their First Customer 08:18 — Moving Into a Customer’s Office 08:59 — The Future of AI Employees in Banking
YC works with thousands of founders every year, which gives us an early look at how startups are changing. Right now, the shift is striking: startups are moving from bits to atoms, nearly one in five YC companies has a solo founder, and companies are reaching meaningful revenue faster than ever. In this episode of the Lightcone, @garrytan, @snowmaker, @sdianahu, and @harjtaggar dig into what’s driving these changes and what they mean for founders. They discuss how AI is making it possible for smaller teams to take on more ambitious problems, why experienced founders are having a resurgence, and why knowing what to build is becoming more important than simply knowing how to build it. 01:06 — Startups Are Moving From Bits to Atoms 03:29 — Why AI Is Making Hard Tech Easier 05:21 — Defense, Manufacturing, and the Return of Hard Tech 09:48 — AI Compute Is Becoming a Physical Infrastructure Problem 12:34 — Robotics Is Approaching Its ChatGPT Moment 15:40 — Software Isn’t Dead. It’s Becoming the Harness 17:59 — Why AI Startups Are Growing Faster 22:52 — The Hidden Boom in Data and RL Environments 27:08 — Why Robotics Will Need Specialized Models 29:38 — The Rise of the Solo Founder 32:54 — Why Experienced Founders Are Back 34:59 — What Founders Should Do Right Now
Tune in: youtu.be/yslXlV2BP_Y
Raindrop (@raindrop_ai) is building the safety layer for AI agents. As agents get more capable and take on more complex work, Raindrop detects when they go wrong in production, from failed tool calls and hallucinations to problems companies didn’t even know to look for. Now they’re launching Raindrop Simulations, which brings that same approach earlier in the development process to catch failures before agents reach production. Already used by companies including Vercel, Clay, Framer, and Speak, Raindrop just raised a Series A bringing its total funding to $50 million. In this episode of Founder Firesides, founders @snarkyzk, @benhylak, and @alexisgauba sat down with @sdianahu to share how a tool they originally built to debug their own coding agent became the company, and why better AI agents actually create more ways for things to go wrong. 00:05 — What Raindrop Does 01:32 — Why Better AI Agents Create More Problems 03:25 — Finding the Failures Evals Miss 04:38 — How Raindrop Got Started 06:32 — Why They Kept Going 07:58 — Catching Agent Failures Before Production
Attack drones are reshaping modern warfare, but many countries still lack an effective way to defend their airspace. Earendil Robotics is building autonomous interceptors, along with the software and training needed to stand up that capability from scratch. Their systems can take off, transit, and intercept targets on their own, reducing months of pilot training to a few clicks. The goal is air defense that’s affordable enough to deploy in large numbers and simple enough to operate at scale.
The US CLARITY Act has an initial vote this week in the Senate. We’re bullish on the Clarity Act - for the future of finance and for American fintech startups. By specifying regulators and clarifying rules, Clarity makes it easier for American founders to build with blockchains.
Blockchains allow instantaneous asset movement for pennies, global access, 24/7 trading, and open APIs. What the internet did for information, blockchains do for every asset, including money. Adding an API to every asset is especially a big deal. Suddenly, anyone can build on top.
The best outbound starts before you automate anything. Founders need to learn who actually has the problem, what gets their attention, and what makes them reply. YC Visiting Partner @ChristinaG325 recommends doing at least your first 100 outreaches by hand so you can learn those things yourself. Drawing from her experience generating and closing millions of dollars through founder-led sales at OneSchema, Christina shares eight ways founders can improve their outbound, from targeting the right people and writing emails worth reading to following up, debugging low reply rates, and using customer language to sharpen your message. 00:38 - Start manual before you automate 01:25 - Targeting beats perfect messaging 02:23 - How to find the right job titles 03:07 - Picking the right companies 03:47 - Intent signals 04:25 - Craft a message worth reading 06:00 - A cold email that works 07:19 - Subject lines 07:37 - Improve your LinkedIn profile 08:47 - What your reply rates are telling you 09:47 - Follow up, and reply fast 10:37 - Block out time for outbound 11:25 - Use the messaging from your customers 12:27 - Your slog compounds
Tune in: youtu.be/wr6PMD06hP0
Rare earth magnets sit inside nearly every electric motor, but 90% of the material is conventionally thrown away. Mass Magnetics (S26) is recycling magnets from EV motors into new ones for robotics and defense, avoiding much of the capital-intensive refining required by mining.
Ashton joined us at Demo Day!
Happy Alumni Demo Day!
Centralize (YC W24) is building the GPS for enterprise deals. It turns emails, calls, CRM data, and other customer interactions into automatic org charts that show sales teams who they know, who they’re missing, and the relationships that can help them win an account. They recently raised a $15M Series A and are already used by teams at companies including Cognition, Intercom, Brex, and Exa. In this Founder Fireside, @UseCentralize_ founders Rachit and Will joined @sdianahu to talk about building the trust graph behind the product, how one customer used it to turn a stalled six-month sales process into an eight-figure deal, and why relationships and institutional knowledge still matter in enterprise sales. 00:00 — GPS for Enterprise Deals 00:45 — Who's Already Using It 01:00 — Building the Trust Graph 03:19 — Real Customer Outcomes 05:52 — Six Months to Weeks 06:08 — Who Should Use Centralize 07:09 — Who Thrives Here
Congrats to @AndyLapsa, @Rkt_Da, and @stoke_space (W21) on their $1B Series E! They're building Nova, a rocket designed to be fully reusable, meaning both stages fly back and fly again. Today's rockets mostly reuse the booster and throw away the upper stage on every flight. Stoke's upper stage returns too, protected by a novel heat shield ringed with 24 thrusters. Nova's first orbital launch is targeted for early 2027, and a larger version carrying 15 metric tons to orbit is already in development. t.co/10q5ruOAIx
Introducing the YC Early Access Network. 4x a year, senior technology leaders get early looks at YC's enterprise AI companies and help shape their roadmaps. Invite-only, by application. Fall '26 kicks off Oct 22 in SF. Apply now: events.ycombinator.com/yc-early-acces…
Harnesses often get dismissed as just scaffolding, just prompt engineering, and not real research. But that couldn't be farther from the truth. The same model weights that score 30% on ARC-AGI score 95% with a better harness. So we gathered a group of researchers and founders working at the frontier to do a deep dive into the state of harnesses. We cover how we got to this point, the case for making your harness as expressive as possible, and what YC learned building an agent for every employee in the company. 00:00 - @FrancoisChauba1: Why harnesses matter 04:27 - Building an auto-researcher by accident 07:13 - A five minute history of harnesses 13:56 - Self-improving harnesses 18:35 - @sethkarten: Prime Agent, a self-improving RLM harness 21:50 - Context as an L1, L2, L3 cache 24:51 - From Turing machine to von Neumann computer 28:33 - Messaging between agents 30:04 - ARC-AGI results 33:09 - Emulator Bench and GPU kernels 37:30 - @JonSaadFalcon: OpenJarvis, personal AI on personal devices 38:26 - How far behind are local models 39:21 - The five primitives of a personal AI stack 42:47 - Letting cloud models optimize your local stack 43:53 - 800x cheaper than the cloud 45:58 - @josh__france and @jbellregan: QM, YC's agent harness for work 47:29 - A history of YC's internal agents 49:24 - OpenClaw and a fleet of 50 agents 51:04 - Pulling the brain out of the sandbox 54:43 - Letting the agent choose its own sandbox and model 57:16 - The grind tool: budgets on goals 58:50 - Agents don't understand social context
Tune in: youtu.be/n9xKblqyQ28
Open models are closing the intelligence gap, says @ollama's @jmorgan. That opens up a new opportunity for founders: coordinating fast, cheap models to solve more complex problems. “We’re maybe less than three months behind between the frontier closed models and the open models. But the next problem to solve is extreme efficiency.” “They’re good enough for 80% of the tasks, they’re really fast and they’re ultra cheap.” “That class of models, in my mind, will be the first ones that come down to this idea of the unlimited tokens.” “Being able to coordinate these flash models together to do different tasks can also yield great results that a bigger model can.”
Y Combinator@ycombinator·🦙 @ollama is used by 9 million developers and 85% of the Fortune 500, giving co-founder and CEO Jeffrey Morgan (@jmorgan) a unique view into which AI models people are actually using and how that’s changing. Right now, the biggest shift he sees is toward open models, driven by coding agents, falling costs, and capabilities that are rapidly catching up to the frontier labs. On Ollama Cloud, that shift has driven a 150x increase in token usage since the start of the year. In this episode of @LightconePod, Jeff joins @garrytan, @snowmaker, @sdianahu, and @harjtaggar to talk about the future of open models and the story behind Ollama, from two years of searching for the right idea to building one of the most widely used AI developer tools in the world. 00:43 — The Shift to Open Models 03:03 — How AI Agents Are Driving Token Usage 05:31 — Are Open Models Catching Up? 08:26 — What Happens When a New Model Launches 11:31 — Ollama as an Operating System for AI 14:05 — The New Opportunities Above the Model Layer 18:19 — Why 80–90% of Enterprise Tokens Could Be Open 20:57 — The Future Is Local and Cloud 26:40 — Why AI Is Coming Back to Your Computer 28:56 — The Coming Era of Unlimited Tokens 32:30 — Do We Still Need a “God Model”? 33:41 — Open Models and Geopolitics 36:14 — The Origins of Ollama 40:36 — Two Years Lost in the Wilderness 42:39 — The Pivot That Changed Everything 47:02 — How Ollama Found a Business Model 49:43 — Why Second-Time Founders Did YC
🦙 @ollama is used by 9 million developers and 85% of the Fortune 500, giving co-founder and CEO Jeffrey Morgan (@jmorgan) a unique view into which AI models people are actually using and how that’s changing. Right now, the biggest shift he sees is toward open models, driven by coding agents, falling costs, and capabilities that are rapidly catching up to the frontier labs. On Ollama Cloud, that shift has driven a 150x increase in token usage since the start of the year. In this episode of @LightconePod, Jeff joins @garrytan, @snowmaker, @sdianahu, and @harjtaggar to talk about the future of open models and the story behind Ollama, from two years of searching for the right idea to building one of the most widely used AI developer tools in the world. 00:43 — The Shift to Open Models 03:03 — How AI Agents Are Driving Token Usage 05:31 — Are Open Models Catching Up? 08:26 — What Happens When a New Model Launches 11:31 — Ollama as an Operating System for AI 14:05 — The New Opportunities Above the Model Layer 18:19 — Why 80–90% of Enterprise Tokens Could Be Open 20:57 — The Future Is Local and Cloud 26:40 — Why AI Is Coming Back to Your Computer 28:56 — The Coming Era of Unlimited Tokens 32:30 — Do We Still Need a “God Model”? 33:41 — Open Models and Geopolitics 36:14 — The Origins of Ollama 40:36 — Two Years Lost in the Wilderness 42:39 — The Pivot That Changed Everything 47:02 — How Ollama Found a Business Model 49:43 — Why Second-Time Founders Did YC
At @HeartAerospace, @AndersForslund1 built the company one material milestone at a time, using each one to earn the credibility and capital to build the next. “We really had no technology. I came to YC Demo Day with a little 3D-printed model.” “We went from having the LOIs to building the 400 kilowatt motor, which was the size of a small jet engine. That in turn attracted actual pre-orders, and then that attracted more capital.” Their first major airline customer came through the spam folder. “United Airlines came through an inbound email through our info email... we showed them the electric motor we built, and they kind of immediately got it.”
Y Combinator@ycombinator·YC W19's @HeartAerospace just flew the largest electric airplane ever flown. Seven years ago, co-founder & CEO @AndersForslund1 came to YC with a 3D-printed model you could hold in your hand. Today, the aircraft has a 100-foot wingspan, weighs 25,000 pounds, and costs just $5 in electricity to get off the ground. YC’s @gustaf visited their pilot plant in LA to learn how they built it -- and what cheap electric flight could unlock for the future of regional air travel.
Paul Graham on what drives founders through the hard parts: “Being dutiful is not enough. The obstacles are too fearsome for mere dutifulness.” “You know what actually motivates founders day to day? The fear of failure.” “If your server’s crashing, you’re not thinking, ‘I got to deal with this because if I deal with it, I’ll become a billionaire.’ You’re thinking, ‘Oh no, the server’s crashing. It will be a disaster. The engine of my model train set is falling off the edge of the table. I need to go and save it.’”
Y Combinator@ycombinator·YC visiting partner @vivianmshen sits down with Paul Graham (@paulg) at the original YC office in Mountain View to talk about startups, AI, ambition, and what makes great founders. 01:07 — Why Today’s Startups Are More Ambitious 04:09 — Frighteningly Ambitious Ideas 05:13 — What Actually Motivates Founders 09:28 — What Makes a Founder Formidable 12:54 — What Surprised PG About AI 17:58 — How YC Started and What Hasn’t Changed
YC visiting partner @vivianmshen sits down with Paul Graham (@paulg) at the original YC office in Mountain View to talk about startups, AI, ambition, and what makes great founders. 01:07 — Why Today’s Startups Are More Ambitious 04:09 — Frighteningly Ambitious Ideas 05:13 — What Actually Motivates Founders 09:28 — What Makes a Founder Formidable 12:54 — What Surprised PG About AI 17:58 — How YC Started and What Hasn’t Changed
Congrats to @ConveoAI (S24) on their $50M Series A! Their AI interviewer holds in-depth video conversations with thousands of real consumers, so brands learn in days what used to take months. Consumer research has always forced a choice: surveys reach lots of people but stay shallow, while focus groups go deep but are slow and expensive. Conveo removes that tradeoff, and more than 400 enterprises now use it, including 50+ Fortune 500 brands like Google, Unilever, and Canva. t.co/ThulQ8rCHz
Wafer (@wafer_ai) is building a fast AI inference cloud that uses agents to optimize GPUs and run open-source models at industry-leading speeds. The result is the low cost of open source with the latency of a much smaller model. Four months after launching the cloud, they went from zero to $8 million in ARR and just raised a $40M Series A. In this episode of Founder Firesides, YC's @snowmaker sits down with Wafer co-founders @gpusteve and @gpuemi to talk about going from UChicago undergrads to building one of the fastest-growing AI infrastructure companies in just over a year. They share how a college side project became Wafer, how they got GLM 5.2 running 2-3x faster than anyone else in the market, and why speed, not just cost, is pushing enterprises toward open-source models. 00:48 — What Wafer Does 01:35 — Going From $0 to $8M ARR in Four Months 04:21 — Scaling Faster Than They Could Keep Up 09:14 — How AI Makes AI Faster 11:21 — The Pivot That Changed Everything 14:41 — Why Open-Source Models Are Taking Off 18:03 — From a College Side Project to Wafer 23:33 — Advice for Starting and Joining Startups
YC W19's @HeartAerospace just flew the largest electric airplane ever flown. Seven years ago, co-founder & CEO @AndersForslund1 came to YC with a 3D-printed model you could hold in your hand. Today, the aircraft has a 100-foot wingspan, weighs 25,000 pounds, and costs just $5 in electricity to get off the ground. YC’s @gustaf visited their pilot plant in LA to learn how they built it -- and what cheap electric flight could unlock for the future of regional air travel.
Congrats to @ArintraHealth (W22) on their $25M Series B! They build AI that translates medical visits into the billing codes hospitals need to get paid. Every chart has to be coded before a claim goes out, and doing it by hand is slow and error-prone. Arintra codes charts autonomously across inpatient, outpatient, ambulatory, and emergency care. They work with large health systems representing more than $50 billion in combined patient revenue and process over $5 billion in claims a year. Customers report a 5.1% lift in compliant revenue capture, 32% lower costs, and 43% fewer coding-related denials. t.co/cuBEpl0L9k
"The way to get the very best startup ideas is not to look for startup ideas. If you’re consciously looking for them, you’ll be too conservative. You’ll lop off the outliers." Instead, work on projects with your friends. Build things you think would be cool. Paul Graham on How to Earn a Billion Dollars: t.co/Xa3fTRxPII
Outset (YC W23) is building an AI customer research platform used by companies like Google, Microsoft, and Nestlé. Its AI interviewers have now conducted millions of conversations, giving companies the scale and speed of a survey with the depth of a one-on-one interview. In this founder fireside, @OutsetAI co-founder and CEO @AaronLCannon sits down with YC's @harjtaggar to talk about building a product before the market was ready for it, how they helped create the category of AI-moderated research, and what comes next. Aaron shares how they're expanding beyond interviews with Outset's new Simulations Lab and Digital Twins, which let companies simulate their customers and get feedback on everything from messaging and pricing to new products. 00:24 — What Outset Does 01:18 — Building a Category Before the Market Was Ready 06:31 — Landing the First Enterprise Customer 08:14 — How Better AI Models Changed the Product 12:57 — When the Market Finally Caught Up 14:45 — From AI Interviews to Customer Simulations 19:40 — Aaron’s Path to Becoming a Founder
Tune in: youtu.be/x3XOeDckUX0
Congrats to @billythalheimer, @mikeklinker1, and @regentcraft (W21) on their $240M Series B! They build seagliders, electric craft that skim just above the water and combine the speed of an aircraft with the convenience of a boat. Their Viceroy Seaglider prototype is designed to carry 12 passengers at up to 180 mph, and its first human flight is imminent. Regent’s commercial order book spans six continents, and production is starting at their newly completed 255,000-square-foot factory in Rhode Island.
Today, more than 3% of the world's lawyers use Legora, and the company has grown from $1 million to $100 million in ARR since launching in October 2024. At Startup School 2026, @WeAreLegora co-founder and CEO @MaxJunestrand shares how they built one of the fastest-growing enterprise software companies in the world, from cold emailing lawyers and moving into a customer's office to freezing sales for six months to rebuild the product. He explains why building a company is ultimately about people, how to create a culture that wants to win, and why founders have to learn to love the hustle. 00:07 — From 3 Engineers to $100M ARR 03:15 — How Legora Got Started 05:09 — Learning the Legal Industry From Scratch 07:22 — Getting Rejected by YC 09:36 — Moving Into a Customer's Office 11:11 — Going From Zero to $1M ARR During YC 13:26 — Why We Froze Sales for Six Months 15:18 — Rebuilding the Product From Scratch 17:26 — Building a Company Is About People 21:51 — Creating a Culture That Wants to Win 25:26 — You Have to Love the Hustle 27:55 — Do Founders Need Domain Expertise? 29:14 — Betting on Models Getting Better 36:29 — When Should You Start a Company? 45:04 — How to Become More Ambitious 48:18 — The Skills Founders Need Today
Tune in: youtu.be/o0ORPbSEgd8 Transcript: ycrootaccess.com/p/max-junestra…
Big labs don’t have a monopoly on AI research. More than 20 YC startups have published research recently at top conferences like NeurIPS, ICLR, and ICML. empirical.health/blog/yc-startu…
Congrats to @hellomundoai on their $20M Series A! Mundo partners with frontier AI labs and companies to build the datasets, evaluations, and research that help AI better perceive and understand the real world across audio, video, and emerging modalities. mundoai.world/research/perce…
What does the stack of the most 'cracked' builders look like these days? @steipete’s latest progression: cli → app → web (multiplayer agent sessions) Watch the full conversation with @raphaelschaad linked below.
Raphael Schaad@raphaelschaad·Caught up backstage w ClawFather @steipete; this is a packed one: * Loops, Graphs, and Florps — what's latest in agentic engineering? * Outie vs. Innie — avoid his personal vs. work agent fighting? * Startup ideas in 2026 — what to build? Lots of incredible zingers in here:
This week's Paper Club is focused on data. As models have scaled across tasks and languages, old assumptions about training and evaluation are starting to break. So we gathered three domain experts to break down the challenges and frontiers of training data, benchmarks, and multilingual pre-training. Thanks to the following presenters: 00:00 - @FrancoisChauba1: Why a whole night about data? 9:51 - @vincentsunnchen: The Art & Science of Benchmarking Agents 28:44 - @volokuleshov: Inception: Diffusion Language Models for Production 40:26 - @ShayneRedford: ATLAS: Practical Scaling Laws for Multilingual Models
Tune in: youtu.be/IfoPg2QefF8
In 2013, @cjoneslevy created the SAFE at YC. It's since been used by YC founders to raise $15B+ and has become the standard way startups raise money. So we're launching Send a SAFE: a free tool to generate, sign, & send one in ~2 minutes, built to be used by founders (and their AI agents).
@cjoneslevy Learn more: ycombinator.com/safe
Michael Kratsios (@mkratsios47) has seen the AI boom from both sides: as COO of Scale AI and inside the White House. Today, as the director of the White House Office of Science and Technology Policy, he helps shape America’s national strategy on AI, science, and emerging technology. At Startup School 2026, he sat down with YC's @lutherlowe, to talk about how Washington makes technology policy, why the White House supports open source AI, and why little tech needs a seat at the table. 02:38 — From Tech to the White House 04:39 — How Technology Policy Actually Gets Made 05:51 — The White House on Open Source AI 08:28 — How Washington Sees AI Differently 09:53 — Regulating a Technology That Changes Every Six Months 11:37 — Which AI Risks Are Overblown? 13:03 — Giving Little Tech a Seat at the Table 17:31 — Regulation Without Creating Incumbent Moats 18:01 — Born Free vs. Born in Captivity Technologies 20:46 — What Working in the White House Is Actually Like 25:56 — A New Golden Age of American Science 31:32 — Quantum, Congress, IP, and What Comes Next 37:54 — Why Technologists Should Consider Public Service
Tune in: youtu.be/zLUZclThLhU Transcript: ycrootaccess.com/p/michael-krat…
Before joining Apple as the Macintosh artist, @SusanKare was an art history PhD who barely knew anything about computers. She went on to design many of the icons, typefaces, and symbols that helped make the original Mac feel understandable and human, defining a visual language for personal computing that still influences software today. At Startup School 2026, Susan shares the stories behind the Happy Mac, Command key, Chicago font, and more, along with the design lessons she learned from Steve Jobs, Paul Rand, and the original Mac team: make things meaningful and memorable, embrace constraints, iterate constantly, and use just enough detail to make an idea instantly clear. 00:00 — Making Things Meaningful and Memorable 01:51 — How Susan Joined Apple 05:01 — Designing the Original Macintosh 06:35 — A Computer Anyone Could Use 08:26 — Creating Chicago and the Mac Typefaces 11:47 — MacPaint and the First Icons 14:45 — The Story Behind the Happy Mac 17:08 — Why Simple Icons Work Better 18:25 — The Icons That Didn’t Work 21:23 — Designing Symbols That Last 22:44 — How the Command Key Got Its Symbol 25:20 — What Ancient Symbols Can Teach Designers 30:18 — Designing After Apple 36:32 — Lessons From Steve Jobs, Paul Rand, and More
Tune in: youtu.be/YEvLKzsEwMw Transcript: ycrootaccess.com/p/susan-kare-o…
