Applications are now open for the Anthropic Residency, a paid, fixed-term research program for academics, technical founders, and staff-level engineers who want to make AI safe and beneficial for everyone. No ML background is required.
Apply by November 22: alignment.anthropic.com/2026/anthropic…
Love to see @PabloTorre bring up @DropSiteNews reporting on Morning Joe, highlighting the wild detail that two Mike Rogers staffers are registered foreign agents for the state of Israel
An AI phone to replace your iPhone.
That’s what we’re building at Cyclon (YC F26).
If you can do it on your phone, you should be able to ask your phone to do it.
AI uses the apps you already have, builds the ones you’re missing, and handles tasks in the background.
Our team used Gemini 4 Argon to build and edit this entire video, then displayed it on a real vintage CRT TV 📺
We used Argon in @Antigravity to script the 3D Blender boot animations and retro SuperCollider audio, then piped the signal via Raspberry Pi to a real TV screen.
“There is no moat.”
But moats are everywhere:
Ambition.
Taste.
Agency.
Curiosity.
Judgment.
Courage.
Obsession.
Persistence.
Integrity.
Patience.
This era belongs to those who refuse to be mediocre.
For AGI to benefit all of humanity, it should help people with the everyday tasks that matter most to them. So we built LifeVal, a new benchmark shaped by feedback from thousands of ChatGPT users on where they most want help from a personal assistant—from finding childcare to accessing government services to managing personal finances.
Over the past year, we’ve used LifeVal to measure and improve how well our models handle these everyday tasks.
openai.com/index/lifeval/
Congratulations to OpenRouter for crossing 10 QUADRILLION annual token run rate today!
Token volumes continue to 2x every 10.5 weeks.
In the last 9 weeks, OpenRouter processed more tokens than all of time before that.
At @ycombinator, we're excited to fund more crypto startups solving financial needs including AI capital raising, stablecoins, trading, and agents.
Here's our latest crypto request for startups. Our next application deadline is Monday.
The age of agents is going to be amazing for most programmers, but it'll be hard times for the cantankerous ones.
Specialized knowledge and rare skills will no longer offer immunity for those who are a pain to work with.
Best learn some basic people skills.
Saying please and thanks to your agent feels like a waste of tokens… but our brain can’t really tell the difference between talking to your agent and talking to a human.
So if you command your AI all day without being polite, it'll transfer to how you talk to colleagues, your wife, your kids and so on.
Don't talk politely for the AI's sake. Do it for your own
The Trump administration is openly meddling in European democracies to promote far-right political parties that support Putin and to advance the commercial interests of Trump allies.
Omarchy on Nvidia N1X, running Nemotron 3.5 Bolt at 90 tps, diagnosing a sleep issue while offline, feels like the future. I know the N1X has gotten a bit of a rough reception, but imo it's almost all down to a terrible out-of-the-box software experience. We can fix everything!
What’s up in the night sky for November? This month, you can look forward to:
- Multiple meteor showers reaching their peak
- Jupiter meeting up with Mars in the wee hours
- A supermoon lighting up the night sky
More skywatching tips: go.nasa.gov/4iuD8FZ
Today we’re releasing GDPx, a benchmark for measuring agentic capabilities across the most valuable domains.
GDPx is a true frontier-level benchmark: even the best model scores just 12% (1/7)🧵
Open-source "codes of conduct" are cancer and need to be retired; I deeply regret championing them. The inevitable outcome is an unproductive mob controlling projects through intimidation. Given enough time, it pushes away people who would contribute patches. Eventually, all that remains is an unproductive military-police class playing social status games.
If you’re under 26, you have never lived a day without someone orbiting above you.
26 years ago today, the first resident crew arrived at the ISS. Humanity has had a home in space ever since.
The research and advances made possible by our crews, teams on the ground and international partners have helped prepare us to live and work on the Moon and Mars.
As we transition to the next era of our presence in low Earth orbit, we build on everything they have made possible.
I remember when FIVE MINUTES seemed like an ambitious goal. Now we are down to 7 seconds!!
Incredible what's possible if you refuse to be satisfied with whatever happens to be the current standard.
We're so, so close now on 4.5. And the improvements coming in 4.6 are already stacking up. Insane acceleration on development, compatibility, performance happening. Everything all at once!
Omarchy v4.5.0 RC3 has been released! 🎉
Just a few more fixes before final release and the first release including the new Apple Silicon installer for M1, M2, M3, and Neo.
github.com/omacom/omarchy…
I re-OCR'd 14,483 pages of fossil fuel industry documents using Saturate. One GPU + $1.47.
8.9 pages/s once running via @huggingface Jobs.
Saturate is a library I've been working on for running models over a dataset or bucket against any OpenAI-compatible endpoint/inference engine. I didn't tune concurrency; saturate settled on 512 by itself.
The image is one line from one page I checked: existing OCR vs Xiaomi-OCR-0
Curious if this tool is at all useful for anyone else?
github.com/davanstrien/sa…
Tutor Intelligence co-founder and CEO Josh Gruenstein is putting AI robots to work in American factories and warehouses, taking on the back-breaking jobs they can't hire for.
Tutor builds its robots in Massachusetts and charges by the hour, with no upfront cost. When the AI gets stuck, a human "tutor" takes over and teaches it the task, creating training data along the way.
Josh joins a16z's Martin Casado and Raghu Raghuram to discuss how Tutor goes from a first sales call to a working robot in about three months, why so many robot startups died building for a single customer, why anyone will be able to start a business in the physical world, and a whole lot more.
0:50 Deploy robots, then build the data
2:50 Why smart robots aren't enough
4:38 The ultimate systems problem
6:45 Why robot companies go vertical
9:07 How remote humans tutor robots
11:00 Will robots outgrow their tutors?
14:54 How AI labs sped up robotics
16:57 Meet Cassie and Sonny
20:55 What factory work actually looks like
23:18 A robot for $16 an hour
25:51 When robots become coworkers
29:04 Why robot startups keep dying
31:21 When 1% cheaper flips a market
34:52 Robots cost more to run than to buy
38:18 China's automation bet may backfire
42:02 Real-world RL on a 100-robot fleet
YouTube: youtu.be/lXkjXSul6yg@joshgruenstein@tutorintel@martin_casado@RaghuRaghuram
We raised a $75M Series B to deploy intelligent robots now, led by @a16z, with participation from @usv and @neo.
@tutorintel robots work real jobs across the United States, directed by human “tutors” and the foundation models we’ve trained on the work done. This round accelerates frontier model training, robot manufacturing, and deployment into industry.
We’re in almost 4x as many sites as a year ago; nearly a third of this year’s deployments went to existing customers coming back for more; and we’re on track to have more than 1,000 robots under contract by the end of 2027.
Our mission is to proliferate robotics for the common good. If you want to make robots reliably do things others think are impossible, for people who’ve never had access to this technology, join us.
Introducing Exa Monitors V2
Get alerts across the largest index of web + proprietary data
- 130B+ web pages
- 211M businesses
- 1.3B products
- 1B people
Most monitors just rerun search, missing results. Exa runs entity recognition on the whole web. Now 7x cheaper at $2/1k runs
Every model upgrade makes AI apps cheaper to run, and easier for customers to replace.
Still, 94% of enterprise LLM buyers expect their spend to rise.
Three rules for winning that spend over the build-it-yourself option:
1. Operate the workflow better than the customer can
2. Make context cheaper to rent than to rebuild
3. Prove the value, then price to capture it
Who's doing it:
- Legal AI platform Harvey beat a general-purpose AI assistant and energy giant Repsol's own in-house tool in a blind test
- Notion considered building its own support AI, then bought Decagon’s AI support agents so its team could stay focused on the core product
"Model progress continually improves the customer's alternative. Your advantage must widen faster for the application to keep earning its premium."
Full piece from a16z's Tugce Erten: a16z.news/p/how-ai-appli…
It's raining OCR models again!
This time, @OpenBMB released MiniCPM-OCR, a 1B-parameter model for document parsing
I made a @huggingface Space so you can compare its outputs against Baidu's new Unlimited OCR v2
Find the link below
Today we’re launching Superserve (YC F26).
Every agent needs a computer. We are building the cloud to give each one their own.
Already running over 100,000 machines every week for teams like Lindy and Y Combinator.
Fastest cold and snapshot starts. Beats E2B, Daytona, Modal, Cloudflare, Vercel, and Google Cloud Run.
curl -fsSL superserve.ai/install | sh
Today I'm announcing Workers IO to solve the slowest part of engineering, verification!
AI agents can now write years’ worth of code in days, but it still takes significant engineering effort to verify what’s being built.
It’s impossible to predict all scenarios that could lead to a failure and that’s why most expensive failures are found in production.
Workers IO runs millions of simulated production scenarios on your entire system, including dependencies. These simulations run inside a deterministic hypervisor which injects faults like network delays, disk errors, clock skew, and process failures to verify how your system handles these conditions.
Deterministic execution allows us to systematically explore the state space of the system and perfectly reproduce failures.
We are already helping customers do years worth of production hardening in days and build confidence in what they’re shipping.
If you are working on systems that you can never afford to break, DM me and we can help you battle test them!
Introducing Infinite Storage on Prized.
Every box can now grow to 40 TB, runs as fast as a local disk, and Docker just works. One more step toward the best dev box for you and your agents.
Adler is a strong compliment to Stoicism, Nietzsche's amor fati, and Camus' "one must imagine Sisyphus happy". Have the courage to take control of your mind, then decide to be happy/optimistic/energetic. Great book!
Yes. Adlerian psychology is a psychology of courage. Your unhappiness cannot be blamed on your past or your environment. And it isn’t that you lack competence. You just lack courage. One might say you are lacking in the courage to be happy
— The Courage to Be Disliked by Ichiro Kishimi and Fumitake Koga
An AI phone to replace your iPhone.
That’s what we’re building at Cyclon (YC F26).
If you can do it on your phone, you should be able to ask your phone to do it.
AI uses the apps you already have, builds the ones you’re missing, and handles tasks in the background.
introducing @tesser_sh (YC W27)
tesser is the simplest way to offload heavy workloads like dev servers and builds to cloud VMs. every worktree gets its own machines, with automatic port forwarding to your laptop
I’m pleased to welcome Andy Baker as an advisor to a16z Global Affairs. I’ve valued Andy’s perspective on national security and look forward to working together to help our founders build partnerships that strengthen America and our allies.
We’re thrilled to lead Tutor Intelligence’s Series B.
@tutorintel is a robotics research and deployment company that builds generally capable robot hardware. They deploy into real production environments ahead of full autonomy, using human teleoperators (“tutors”) to train the robot foundation models and make the fleet more capable over time.
To date, Tutor robots are in operation across some of the country’s largest manufacturing and logistics networks, with plans to have more than 1,000 robots under contract by the end of 2027.
Co-founders Josh Gruenstein and Alon Kosowsky are lifelong robotics enthusiasts who met as undergrads at MIT. They combine excellent research chops with a practical, ROI-driven orientation that is more typical in enterprise software.
Congrats @joshgruenstein, @KSachsAlon, and the Tutor Intelligence team!
By @RaghuRaghuram, @martin_casado, and @jamiedsully
We raised a $75M Series B to deploy intelligent robots now, led by @a16z, with participation from @usv and @neo.
@tutorintel robots work real jobs across the United States, directed by human “tutors” and the foundation models we’ve trained on the work done. This round accelerates frontier model training, robot manufacturing, and deployment into industry.
We’re in almost 4x as many sites as a year ago; nearly a third of this year’s deployments went to existing customers coming back for more; and we’re on track to have more than 1,000 robots under contract by the end of 2027.
Our mission is to proliferate robotics for the common good. If you want to make robots reliably do things others think are impossible, for people who’ve never had access to this technology, join us.
Introducing Jinko, the travel search and booking infrastructure for agents
50% of transactions volume made through personal agents are for travel, but the infrastructure is still broken.
With Jinko, agents can:
> Answer all travel questions
> Avoid getting blocked by Airlines (United and American Airlines)
> Create a new revenue stream by earning commissions
Travel will be a core primitives for agents.
We are working with agents such as @wajo_ai , @pally , @ExaAILabs and more.
Connect Jinko today - gojinko.com
Book a demo - calendly.com/kevin-jinko/30…
In updating a graphic showing U.S. company AI adoption over time according to a long-running U.S. Census Bureau survey, I found out that the Bureau's work has been sullied by that asinine "Super Intelligence" executive order
Today we're launching Hera Canvas on Product Hunt.
Generate images, design, and animate on one canvas. We've been using it for all our own ads.
Link to our Product Hunt launch is in the comments.
Today, the @lettercompany is releasing the beta of Letterpress-1, the world’s first coding agent, harness, and IDE purpose-built for growth engineering.
tl;dr: check out the 🧵 to see the product in action.
Growth is re-emerging as an engineering problem, but today’s coding tools weren’t designed for it. Your website isn’t isolated like your app: it reacts to the outside world (e.g. agent behavior, Google rankings, observed human interactions, and competitor activity).
Leading coding agents were built to scale complex codebases, not to autonomously build for outside signals. Letterpress-1 is.
Letterpress-1 is designed to tackle growth problems of every size. It’s also proactive: it suggests the nth next thing to build for growth. It'll issue pull requests for de-anonymization, experimental landing pages, warmed emails, blog automations, agent telemetry, etc.
The Letterpress IDE is also built into your site. Press ⌘+L, authenticate, and edit text directly on your website. Letterpress-1 automatically opens a pull request with the change.
Letterpress-1 ships with readily available infrastructure to stage futuristic growth strategies that win over agents. These include:
=> Proofs: autonomous evidence chains that verify customer success data for agent visitors
=> Simulations: staged mirror applications where agents can safely try solutions before buying
=> Benches: self-hosted, reproducible benchmarks with open-source, independent governance
=> Phantoms: autonomous websites to sandbox growth strategies
Build, automate, and refine your growth process with the Letterpress family of tools. We're proud to count some of the world's most forward-thinking startups as our customers.
Sign up today at letter.dev.
I’ll share a few examples of Letterpress-1 in action in the thread below.
1/ Explore what’s possible with Gemini 4 Argon in @antigravity.
Take a look at some community experiments across 3D generation, physics, and reverse engineering to inspire your next build 🧵
Prospecting tools are useless if nobody is ready to buy.
That's why we built Clean.
The world first AI that predicts which companies are ready to buy, right now.
Here's how it works:
Introducing Darwin Robotics.
We're building general-purpose robot technicians that assemble, commission, and maintain data centers.
AI infrastructure is scaling rapidly, but bringing compute online still requires enormous amounts of physical work. Skilled technicians must install servers, connect cables, commission equipment, and maintain the infrastructure.
We create robot technicians that can take a work order and execute the physical tasks requirement to assemble, configure, and repair hardware. Instead of explicitly programming each workflow, our robots learn complex physical tasks from human demonstrations and improve through real-world deployment.
Here’s an early demonstration of Darwin performing learned cable manipulation on a server rack.
Our mission: Robots to build civilization.
If you're building or operating data centers, let's talk about deploying Darwin at your facilities.
darwinrobotics.ai
Today, we launch Ryhla (YC F26)
Solar, batteries, EV chargers, data centers: everyone wants to connect to the power grid. There isn't room for everyone, so developers wait months, sometimes years, for an answer.
Ryhla builds models of the distribution grid to find the best spots for your project and get you connected faster.
Introducing Zenith. Our in-house designed flight computer, purpose-built for advanced reentry.
From Mach 20+ hypersonic flight to a precise, soft landing, Zenith orchestrates nearly every system on Arc.
Bridgit Mendler of @NorthwoodSpace says the builders of the early internet bet on a direction without knowing exactly what it would become, and the space industry is in that same moment now:
"For folks that were early with the internet, there's no way they would've foreseen what the internet is today. But they were betting on a direction... and look what we have today."
"I think we're in a similar moment with the space industry where we don't know what is going to be built. A year ago today, would we think that everybody would be talking about orbital data centers? No way."
"But it's captivated the public imagination and who knows what's gonna come next... The whole thing that we're about is taking companies further faster. Taking space missions further faster."
"If you analogize to the internet, you have a couple of kind of platform infrastructure plays that really wind up becoming enmeshed across the internet... There are a few that wind up becoming really influential and fundamental."
"Cloud is one that we analogize to a lot and we are inspired by. And we wanna be one of those companies."
@BridgitMendler
Today, Northwood is announcing a $375M Series C.
In February of 2024, less than three years ago, we announced Northwood as a small six-person team.
We’ve grown a lot, but the goal stays the same – solve connectivity between Earth and space to take missions further faster.
The world is hungry for more data movement through space, and that is what we're building for, with scalable network capacity for the most challenging use cases.
We are proud to add new partners to our cap table and continue the journey with existing ones as we enter this new chapter.
Thanks to @a16z@DFJvc@WashingHarbour Alpine Space Ventures @stepstone and many others.
Read more: northwoodspace.io/blog/northwood…
Holy moly, @particle_api 's model landed on our pareto frontier for data analytics tasks, beating out Terra-6.1 and Haiku 5.5.
First time a new player has landed on our frontier in years. Congrats @Suhail and team!
Full results here: plotly.com/blog/particle-…
See how AI agents help create a simulated obstacle course for Microduck, using NVIDIA Omniverse libraries for 3D asset creation, Newton for physics simulation, NVIDIA Isaac Lab for policy training, and Isaac Sim on NVIDIA RTX Spark for validation. nvda.ws/4zo7dOM
Blacksmith (W24) is building a faster cloud for continuous integration (CI), helping developers test and ship code as AI coding tools generate more of it.
As software production accelerates, testing and validation are becoming the new bottleneck. That shift has helped @useblacksmith grow 20x over the last year, and the company recently raised a $66 million Series C at a $1 billion valuation.
The founders, @adijayaprakash, @aayush_shah15, and @adityamaru27, started as outsiders to DevOps and built their first infrastructure using the same high-performance CPUs found in gaming PCs, renting servers originally designed for Minecraft and Call of Duty. Today, they run a much larger fleet with a team of just 30 people, helping companies cut CI waits from minutes to seconds.
They share how they found the idea, won their first 200 customers, and why faster testing will become even more important as agents write more of the world’s code.
00:00 — Making CI Faster With Gaming CPUs
03:46 — Winning Over Investors and Developers
06:46 — Scaling CI for the AI Coding Era
One month's progress for a successful startup. The previous month's towering revenue number now looks small, and the early part of the graph is starting to read as flat. This is what exponential growth feels like.
25,000+ of you joined the waitlist. Today, we’re welcoming our first traders. ☕
Mochatrade’s first batch of invites is rolling out now.
Here’s what’s waiting:
→ Trade US markets from India, in rupees.
→ Get early access before our public launch.
→ ₹500 in trading credits. T&Cs apply.
Still waiting for your invite? We’re opening access in batches over the coming days. Keep an eye on your inboxes
Thanks for being here from the start. 🤎
app.mochatrade.com
Josh Gruenstein of Tutor Intelligence says a perfect robot brain wouldn't solve robotics, because the bottleneck is building robots at scale and finding customers, not intelligence:
"There are eight billion people on Earth who all have this baseline physical intelligence that we want our robots to have."
"Imagine you had infinitely powerful software that could connect any human being on Earth to any robot and embody that robot with intelligence... Suddenly you have unlimited infinite robot intelligence... Now what?"
"The problem isn't actually solved, right? You still need to actually manufacture robots at scale... Robots are a rounding error even in the most automated industries like manufacturing and logistics."
"None of the software exists and none of the customers exist because the robots can't do any of the jobs that customers want robots to do."
"So robotics is not any one layer of the stack. It's not just about intelligence, it's not just about hardware, it's not just about the customer. It's about kind of figuring out how to put all these three pieces together."
@joshgruenstein@tutorintel
Tutor Intelligence co-founder and CEO Josh Gruenstein is putting AI robots to work in American factories and warehouses, taking on the back-breaking jobs they can't hire for.
Tutor builds its robots in Massachusetts and charges by the hour, with no upfront cost. When the AI gets stuck, a human "tutor" takes over and teaches it the task, creating training data along the way.
Josh joins a16z's Martin Casado and Raghu Raghuram to discuss how Tutor goes from a first sales call to a working robot in about three months, why so many robot startups died building for a single customer, why anyone will be able to start a business in the physical world, and a whole lot more.
0:50 Deploy robots, then build the data
2:50 Why smart robots aren't enough
4:38 The ultimate systems problem
6:45 Why robot companies go vertical
9:07 How remote humans tutor robots
11:00 Will robots outgrow their tutors?
14:54 How AI labs sped up robotics
16:57 Meet Cassie and Sonny
20:55 What factory work actually looks like
23:18 A robot for $16 an hour
25:51 When robots become coworkers
29:04 Why robot startups keep dying
31:21 When 1% cheaper flips a market
34:52 Robots cost more to run than to buy
38:18 China's automation bet may backfire
42:02 Real-world RL on a 100-robot fleet
YouTube: youtu.be/lXkjXSul6yg@joshgruenstein@tutorintel@martin_casado@RaghuRaghuram
Planning your trip to MAX POWER? We've got a tool for that.
Event schedules, maps, and FAQs are at your fingertips with the MAX POWER app. It's free and available to download from official app stores on iOS and Android devices: go.nasa.gov/3V2iiWX
The deadline to apply to the YC W27 batch is tonight at 8 pm PT!
If you’re making something people want, we’d love to hear from you.
ycombinator.com/apply
Exactly a year ago today, we launched Starcloud-1 to space. To celebrate its birthday 🥳, we have put together some previously unseen footage we have from around the launch. I particularly like the footage we have of the @Starcloud_ kids! 😅👌
Starcloud-1 is operating today with the only @NVIDIA H100 in space, still the most powerful SI compute on orbit. With it we became the first entity to do training and high powered inference on orbit, and we now have 3 more launches coming up in the next year! We are planning to move almost all compute off Earth 🚀
See the careers page on our website if you want to join this awesome team!
ladies and gents, we did it - Omarchy on the iPad Mini 4!! 🎉
massive thanks to @henrikklee for accelerating the journey … and my god, switching from Codex to Claude made it 10x easier!
now to get this merged upstream!
Election Tuesday is just two days away. For those eligible to vote in the U.S., please make time to go to the polls to vote (or mail in your ballot if you haven't already). Voting is a great privilege of living in a democracy and how we all make our collective voices heard. 🗳️
This walkthrough shows OpenSEO + OpenCRM + OpenMail all working together in EveryApp.
Easily self host all 3 in your @CloudflareDev account, then use them through 1 MCP gateway.
The most common fork of OpenSEO is building a CRM for a backlink outreach system.
This article shows how to easily do this with EveryApp by combining: OpenSEO, OpenCRM and OpenMail. x.com/i/article/2116…
YC's own agent harness runs on Superserve. So do Lindy and 50+ other teams, spinning up 100K+ machines every week. This is a critical piece of infrastructure for any company building serious agentic experiences.
It’s been awesome working with Pavitra!
Today we’re launching Superserve (YC F26).
Every agent needs a computer. We are building the cloud to give each one their own.
Already running over 100,000 machines every week for teams like Lindy and Y Combinator.
Fastest cold and snapshot starts. Beats E2B, Daytona, Modal, Cloudflare, Vercel, and Google Cloud Run.
curl -fsSL superserve.ai/install | sh
Josh Gruenstein of Tutor Intelligence on how to value a robot: "You take how much work a robot can do and you divide it by its cost, and that's it. Everything else is commentary."
If robots can do that work for less, he says the market is "all physical work done."
One Wall Street firm's view of the scale, in a forecast of annual robot sales:
- 2026: 29 million
- 2030: 93 million
- 2040: 605 million
- 2050: 1.4 billion
"We're just gonna have way more factories in the same way that we have way more businesses because the cloud and all of the software tooling let people build new businesses."
"Anybody will be able to start a business in the physical world. Anybody will be able to scale a business in the physical world if we kind of have this infinite flexible labor substrate."
@joshgruenstein@tutorintel
Tutor Intelligence co-founder and CEO Josh Gruenstein is putting AI robots to work in American factories and warehouses, taking on the back-breaking jobs they can't hire for.
Tutor builds its robots in Massachusetts and charges by the hour, with no upfront cost. When the AI gets stuck, a human "tutor" takes over and teaches it the task, creating training data along the way.
Josh joins a16z's Martin Casado and Raghu Raghuram to discuss how Tutor goes from a first sales call to a working robot in about three months, why so many robot startups died building for a single customer, why anyone will be able to start a business in the physical world, and a whole lot more.
0:50 Deploy robots, then build the data
2:50 Why smart robots aren't enough
4:38 The ultimate systems problem
6:45 Why robot companies go vertical
9:07 How remote humans tutor robots
11:00 Will robots outgrow their tutors?
14:54 How AI labs sped up robotics
16:57 Meet Cassie and Sonny
20:55 What factory work actually looks like
23:18 A robot for $16 an hour
25:51 When robots become coworkers
29:04 Why robot startups keep dying
31:21 When 1% cheaper flips a market
34:52 Robots cost more to run than to buy
38:18 China's automation bet may backfire
42:02 Real-world RL on a 100-robot fleet
YouTube: youtu.be/lXkjXSul6yg@joshgruenstein@tutorintel@martin_casado@RaghuRaghuram
This is what re-industrialising America looks like: Sinter helps hardware teams go from prototype to production in days.
Congrats on the launch @dishwashertree and Justin Kwon!
Ever build a web app, hit refresh, and watch all your data vanish into thin air? 👻 Watch how to make sure your data doesn’t ghost you using Firestore in Episode 4 of Don't Be Afraid of the Cloud!
Try it here: goo.gle/4zrFE7o
Josh Gruenstein of Tutor Intelligence says he watched robot startups die the same way "5 to 10 times in a row," and explains what they got wrong:
"We started this company at a time we're seeing all of these very academic supply chain oriented robotic startups get started and then not succeed... One of the big things I think that robotics companies get wrong is they're not actually product companies."
"They're like, 'Okay, some really big company has a very specific problem... We're gonna spend two years building a very custom system to solve their problem.'"
"Chances are it won't work... doing product requires iteration. Even if it does work, who's to say that customer still has that problem? So boom, the company's done."
"Working with a diverse range of customers is this beautiful forcing function for building the right shape of product, which to us is a general robot."
"If you can sign a contract in two weeks with a mid-market customer and have a robot running immediately after that... you know you're being successful."
@joshgruenstein@tutorintel
Tutor Intelligence co-founder and CEO Josh Gruenstein is putting AI robots to work in American factories and warehouses, taking on the back-breaking jobs they can't hire for.
Tutor builds its robots in Massachusetts and charges by the hour, with no upfront cost. When the AI gets stuck, a human "tutor" takes over and teaches it the task, creating training data along the way.
Josh joins a16z's Martin Casado and Raghu Raghuram to discuss how Tutor goes from a first sales call to a working robot in about three months, why so many robot startups died building for a single customer, why anyone will be able to start a business in the physical world, and a whole lot more.
0:50 Deploy robots, then build the data
2:50 Why smart robots aren't enough
4:38 The ultimate systems problem
6:45 Why robot companies go vertical
9:07 How remote humans tutor robots
11:00 Will robots outgrow their tutors?
14:54 How AI labs sped up robotics
16:57 Meet Cassie and Sonny
20:55 What factory work actually looks like
23:18 A robot for $16 an hour
25:51 When robots become coworkers
29:04 Why robot startups keep dying
31:21 When 1% cheaper flips a market
34:52 Robots cost more to run than to buy
38:18 China's automation bet may backfire
42:02 Real-world RL on a 100-robot fleet
YouTube: youtu.be/lXkjXSul6yg@joshgruenstein@tutorintel@martin_casado@RaghuRaghuram
Gemini 4 Argon in Antigravity is an absolute beast 🔥
Inspired by yesterday’s Gemini 4 Argon demo by George Kenwright, I wanted to recreate a similarly high-quality and accurate 3D structure.
I fed the demo video to Antigravity and asked it to reverse-engineer the skill behind it, using the visible workflow as a reference. I then upgraded my existing skill based on the process used in the official demo.
4 hours in, and Antigravity is absolutely cooking. Let's see how long it takes, but so far, it's looking really good!
More details soon :)
Josh Gruenstein of Tutor Intelligence says the machine-run factory most people picture is just the top .01%, and humans still box almost everything you buy:
"Most people have this mental picture of what the industrial base or manufacturing logistics looks like in the United States... You see these Rube Goldbergian machines of levers, and you don't see any humans in those pictures, right?"
"That's the top .01% of American manufacturing logistics. What's below that? You walk into these million square foot facilities and you see a bunch of people standing around on conveyor belts or folding tables doing things by hand."
"It's kind of shocking for anybody who doesn't come from this industry and community."
"A huge amount of work, which is back-breaking work that's really hard to hire for. There's a three-month turnover where your average age of the worker doing that work is 55 years old and about to retire."
"Basically anything you ever bought that came in a box, a human had to put that in a box, which is a crazy revelation for most people... Even iPhones, you would assume it's a robot putting that packaging together. No, it's a person even at the highest level of scale."
@joshgruenstein@tutorintel
Tutor Intelligence co-founder and CEO Josh Gruenstein is putting AI robots to work in American factories and warehouses, taking on the back-breaking jobs they can't hire for.
Tutor builds its robots in Massachusetts and charges by the hour, with no upfront cost. When the AI gets stuck, a human "tutor" takes over and teaches it the task, creating training data along the way.
Josh joins a16z's Martin Casado and Raghu Raghuram to discuss how Tutor goes from a first sales call to a working robot in about three months, why so many robot startups died building for a single customer, why anyone will be able to start a business in the physical world, and a whole lot more.
0:50 Deploy robots, then build the data
2:50 Why smart robots aren't enough
4:38 The ultimate systems problem
6:45 Why robot companies go vertical
9:07 How remote humans tutor robots
11:00 Will robots outgrow their tutors?
14:54 How AI labs sped up robotics
16:57 Meet Cassie and Sonny
20:55 What factory work actually looks like
23:18 A robot for $16 an hour
25:51 When robots become coworkers
29:04 Why robot startups keep dying
31:21 When 1% cheaper flips a market
34:52 Robots cost more to run than to buy
38:18 China's automation bet may backfire
42:02 Real-world RL on a 100-robot fleet
YouTube: youtu.be/lXkjXSul6yg@joshgruenstein@tutorintel@martin_casado@RaghuRaghuram
today we're launching People Infrastructure for Agents
the problem: every people stack you pay for was built for a human with a mouse, and your agents hate using your mouse, with all those separate logins and invoices
introducing Lantern's People Infrastructure for Agents:
-> 70+ popular tools (people data, company signals, deep research, outreach) behind one connection
-> CLI native, built for codex, claude code and grok bot
-> your agent describes the job, Lantern picks the tools, finds the human, explains the fit and reaches out via Linkedin, email, iMessage
-> one wallet instead of 6 separate subscriptions
Tell your agent "setup people.lantern.md", and get started with our generous free tier
let's give agents the ability to contact ANYONE in the world ;)
@LanternAI__@ycombinator
In this powerful conversation, Kurtis Blow, The D.O.C., Roxanne Shanté, and QD3 explore the generations of artists who shaped the culture, the pioneers who waited decades for recognition, and the importance of passing that legacy forward.
From the roots of Black music to mentorship and the future of Hip-Hop, one thing is clear: every generation stands on the shoulders of those who came before.
#PaidInFullFoundation#HipHopNobelPeacePrize#HipHopHistory#HipHopCulture#HipHopLegends#KnowYourHistory#a16z
No, thank you, this was good work and I think good engineering is about making others in the field better and being critical about each others work. You did that for Ghostty today, and I hope that can continue.
Josh Gruenstein of Tutor Intelligence on the factories their robots work in: "None of these places can hire."
The year-over-year data agrees:
- Hours per worker are up in all 10 manufacturing categories
- Hours per worker are growing faster than headcount in 7 of 10
"Back-breaking work that's really hard to hire for. There's a three-month turnover and your average age of the worker doing that work is 55 years old and about to retire... Have you ever considered a robot for 16 bucks an hour?"
@joshgruenstein@tutorintel
Tutor Intelligence co-founder and CEO Josh Gruenstein is putting AI robots to work in American factories and warehouses, taking on the back-breaking jobs they can't hire for.
Tutor builds its robots in Massachusetts and charges by the hour, with no upfront cost. When the AI gets stuck, a human "tutor" takes over and teaches it the task, creating training data along the way.
Josh joins a16z's Martin Casado and Raghu Raghuram to discuss how Tutor goes from a first sales call to a working robot in about three months, why so many robot startups died building for a single customer, why anyone will be able to start a business in the physical world, and a whole lot more.
0:50 Deploy robots, then build the data
2:50 Why smart robots aren't enough
4:38 The ultimate systems problem
6:45 Why robot companies go vertical
9:07 How remote humans tutor robots
11:00 Will robots outgrow their tutors?
14:54 How AI labs sped up robotics
16:57 Meet Cassie and Sonny
20:55 What factory work actually looks like
23:18 A robot for $16 an hour
25:51 When robots become coworkers
29:04 Why robot startups keep dying
31:21 When 1% cheaper flips a market
34:52 Robots cost more to run than to buy
38:18 China's automation bet may backfire
42:02 Real-world RL on a 100-robot fleet
YouTube: youtu.be/lXkjXSul6yg@joshgruenstein@tutorintel@martin_casado@RaghuRaghuram
yc applications close nov 2 (tomorrow!). yc helped us pivot giga, unlocked 50+ leads off one retweet and got us intros to top public companies. if you've been sitting on an idea, just apply
The deadline to apply for the YC Winter 2027 batch is coming up in 2 days!
YC founders get:
🔸 $500K in funding
🔸 $12M+ in tokens, credits & deals
🔸 Partners who were founders first, and will always have your back
🔸 Access to the best founder community in the world
🔸 Help launching your company across YC’s channels, with a combined audience of 6M+
🔸 Connections to 100s of potential enterprise buyers through YC’s Early Access Network, months before Demo Day
🔸 Hiring through Work at a Startup, with 1.25M+ candidates across engineering, design & product. (Founders have already made 2,000+ hires)
🔸 Demo Day in front of thousands of top investors
Apply by Nov 2: ycombinator.com/apply.
In 2012, I asked @ericmigi whether YC was good for hardware startups, and this was his answer.
The answer to all of these questions is now yes. Today, 25% of @ycombinator startups build hardware. YC is excited to fund even more, including supply chains and machines, and including robots, data centers, and aerospace. Here's the latest request for startups. The next application deadline is today!
I think its obvious in my message but I should be explicit that I focused on VT throughput here and not your other benchmarks like startup time and input latency and stuff. I just want to be explicit because I don't want you to think I'm dodging that.
libghostty-vt is the main performance crown jewel for Ghostty and the biggest impact industry-wide (since so many things embed libghostty).
libghostty-vt does include a _render state_ API which is a very high performance API to feed renderers (of any paradigm, CPU or GPU, we have both as users!). Details here: libghostty.tip.ghostty.org/group__render.… Realistically, I've measured Ghostty's frame times to be a fraction of a 120Hz frame but the thing I care about most is the render state times.
I don't dedicate as much time or energy in the other aspects of performance and I accept that we're likely slower there, and I don't think I've ever claimed otherwise so this should be consistent with how I talk about performance.
Watching YC as a young aspiring applicant in college, to a YC founder in my 20s, and now a YC partner in my 30s - there is no other institution that has this legacy.
I'm blown away by the way companies help each other during the batch and reach back through time as alumni as well. The network effect of decades of founders who want to give back is irreplaceable.
I hope you take a chance and apply.
Deadline is 8pm PT tonight, hope to see you on campus.
The deadline to apply for the YC Winter 2027 batch is coming up in 2 days!
YC founders get:
🔸 $500K in funding
🔸 $12M+ in tokens, credits & deals
🔸 Partners who were founders first, and will always have your back
🔸 Access to the best founder community in the world
🔸 Help launching your company across YC’s channels, with a combined audience of 6M+
🔸 Connections to 100s of potential enterprise buyers through YC’s Early Access Network, months before Demo Day
🔸 Hiring through Work at a Startup, with 1.25M+ candidates across engineering, design & product. (Founders have already made 2,000+ hires)
🔸 Demo Day in front of thousands of top investors
Apply by Nov 2: ycombinator.com/apply.
Rather skip the download? You can access the app's features and start planning your MAX POWER experience by visiting maxpower.kennedyspacecenter.com on any mobile device.
@SUNY partnered with Google Cloud to provide secure access for hundreds of researchers to cutting-edge technology like Gemini and AI Hypercomputer, furthering its mission to advance solutions for the public good faster ↓
So excited to introduce @RoutantAI (YC F26): The best way to let AI agents into your apps, on your rules.
Agents need access to get real work done. Teams need control over what they can see and do. routant.ai builds the infrastructure to handle both.
Check it out👇
