A few weeks ago @steipete told me he was thinking about canceling his TED talk. Too busy. No time to prep. Working on OpenClaw and OpenAI simultaneously with zero bandwidth.
He prepped the whole thing in a week, showed up to Vancouver still taking meetings between sessions, walked into the green room with a lobster in his pocket, and delivered one of the most talked about talks of TED2026 – no teleprompter.
Moments like this are why I love curating TED. The mad lad did it!

I want to share some difficult news. I was recently diagnosed with a rare form of cancer and I want you to hear it directly from me.
Treatment is starting immediately and will include multiple strategies over the course of about a year. While I will be stepping back from some of my usual activities, I will continue to support SV Angel founders, who I love with a passion.
SV Angel remains unchanged. Topher has made all of our investment decisions for the better part of the last decade, and Ronny joined as Managing Partner in 2024. They bring experience from nearly every major technology cycle in Silicon Valley and are now focused on partnering with founders building the future of AI. SV Angel has a deep, experienced team that remains fully focused on supporting exceptional founders.
With a more focused and balanced schedule, I can prioritize treatments while helping SV Angel founders at inflection points like we always do!
I’ve chosen not to share the specific type of cancer since I don't want speculation about my prognosis. I appreciate your understanding and respect for this.
I am optimistic about my prognosis. I am fortunate to have the best/amazing team of UCSF doctors in San Francisco, and as you know, I never back down from a fight.
Thank you for your support, it means a great deal to me.

To get LLMs to work effectively, you need to extract and document all the implicit "domain knowledge" in your business.
Once you do that, AI agents are magical.
But it also means that OpenAI/Anthropic have full access to your company's domain knowledge 😬

Lots of people being negative about @openclaw past couple of weeks…
And yes, it’s fun to play with other tech too… but let’s have reverence for who forced anthropic to release 3 years worth of their roadmap in 3 months 🦞
Good watch.
Garry Tan: Must watch. OpenClaw is a revelation. https://www.ted.com/talks/peter_steinberger_how_i_created_openclaw_the_breakthrough_ai_agent

BREAKING: @ShaunWMusic just landed his first paying customer on Levelyze.
Shaun Willis: first paying customer came through today!! @Replit @raymmar @MannyBernabe Let's gooo! You guys were the best. Enjoyed learning from you both

Tom Steyer using his billions to buy his way to Governor while counting on progressives who want to "tax the rich" and do a billionaires tax. And like lemmings, they're falling for it. The hypocrisy is astounding.
Rob Pyers: It's a Friday night, so that can only mean one thing--Tom Steyer dumping another $11.25 million into his campaign for California governor.
Updated total candidate spend for political reporters: $133,776,123

Run. Build. Ship.

The CEO of Y Combinator just open-sourced his entire AI development setup.
And it is already at 72,600 stars on GitHub.
Garry Tan runs Y Combinator. He has worked with Coinbase, Instacart, and Rippling when they were two people in a garage. Before that he was one of the first engineers at Palantir. He has seen more startups build product than almost anyone alive.
He is now shipping 10,000 to 20,000 lines of production code per day. Part-time. While running YC full-time.
In the last 60 days alone: 600,000 lines of production code. 35% of it tests.
That number is not a typo.
Here is exactly how he does it.
He built a system called gstack — 23 AI tools that turn Claude Code into a full engineering team. He open-sourced the entire thing. Free. MIT license. One command to install. And then he posted the quote that explains why he built it:
"I don't think I've typed like a line of code probably since December, basically, which is an extremely large change." — Andrej Karpathy, March 2026.
When Tan heard that, he wanted to find out how. The result is gstack.
Here is what the 23 tools actually do.
There is a CEO tool that challenges your product framing before you write a line of code. It does not just approve your idea. It finds the 10-star product hiding inside what you described and pushes back on everything you got wrong.
There is an engineering manager that locks architecture, draws ASCII diagrams of data flow, and forces hidden assumptions into the open before anything gets built.
There is a designer that rates every design decision on a 0 to 10 scale, explains what a 10 looks like, and edits the plan until it gets there. It also has AI slop detection. It catches the generic AI output that looks fine and ships badly.
There is a QA lead that opens a real browser, clicks through your actual app, finds bugs, writes regression tests, and verifies the fix. Not a simulation. A real browser.
There is a security officer that runs OWASP Top 10 and STRIDE threat modeling with 17 false positive exclusions built in, so you only see findings that actually matter.
There is a release engineer that syncs main, runs tests, audits coverage, pushes, and opens the PR. One command from approved to shipped.
And then there is something Tan says was the biggest unlock of all.
You can run 10 to 15 of these sprints in parallel. Each one in its own isolated workspace. One agent challenging a product idea. One implementing a feature. One doing QA on staging. Six more on separate branches. All at the same time.
Tan's GitHub contribution graph for 2026 is a vertical wall. In 2013, building Bookface at YC from scratch, he made 772 contributions in a year. In 2026, he is at 1,237 — and still climbing.
Same person. Different era. The difference is the tooling.
One more thing.
In the README, Tan quotes the number directly: 140,751 lines added. 362 commits. 115,000 net lines of code. In one week. Part-time.
That is not what a solo developer looks like. That is what a team looks like.
Except it is one person with 23 AI specialists and a GitHub repo you can clone right now for free.
http://github.com/garrytan/gstack

Second episode of the podcast. @davis7 spends most of this episode trying to gstack pill me. Did he succeed?
00:02:50 - Is Mythos legit?
00:24:20 - Post-AI Uncle Bob @unclebobmartin
00:36:23 - Praising Claude Code a bit
00:40:49 - GStack is actually good @garrytan
00:51:07 - Everything should be a md file

👀

Grateful to these incredible founders for speaking at Startup School India today.
Aadit Palicha (Zepto)
Harshil Mathur (Razorpay)
Vidit Aatrey (Meesho)
Lalit Keshre (Groww)
Mukund Jha (Emergent)
Varun Vummadi (GigaML)
Six YC-backed Indian founders. Combined company value: $25B+. Jobs created in India: over 250K.

📍 YC startup school

gbrain v0.11 is out. garry tan ships minions agent orchestration, 10x faster than openclaw's default subagents.
→ spawn storm defense
caps recursion depth and per-parent child count. max 5 levels deep by default, max_children is tunable per parent
→ tasks run only once
tag a job with an idempotency key. same key, only one run ever. a cron misfire or two pods submitting at once collapses to a single row
→ children auto-notify the parent
when a child finishes, the result lands in the parent's inbox. the parent just reads its mail, no polling needed
garry benchmarked: 19,240 posts across 36 months, single bash loop, 15 minutes, $0. the sub-agent spawn approach failed on 40% of runs and cost $1.08 in tokens.
v0.10 taught the agent which skill to use. v0.11 makes sure the agent's work doesn't get dropped.
Garry Tan: Now launching GBrain v0.11 with Minions
I got sick of OpenClaw's subagents timing out and not getting things done
So I built a queue/jobs system that uses GBrain's Postgres/PGLite based on BullMQ to give your OpenClaw/GBrain setup wings.
Minions are 10X faster, more reliable

Giga is one of the first great AI companies to come out of India. Here are some crazy stories @varunvummadi shared:
1). He came from a poor background, but made $100k winning kaggle competitions.
2). To start Giga he turned down an insanely high paying offer from a high frequency trading fund. He had to convince his parents.
3) Giga is a completely different idea from what they started YC with. They wound up building AI customer support after they talked to Aadit from Zepto and realized it was a big need.
4). Most people thought it was too late - many companies were already doing AI customer support and some of the had raised large amounts of money.
5). It wasn't. Giga was able to close huge customers like Zepto and Doordash very quickly. They just built faster and better.
6.). If AI coding agents didn't exist, he'd need 7x the team size to run Giga.

Just had a holy-shit moment with my @openclaw exec assistant while I’m traveling with the fam in Japan.
R2, my claw, has a cron job that runs every 15 minutes to sweep my inbox/cal and this task appeared in my todo list:
“Review Vercel new-device sign-in alert and confirm it was expected”
He connected the dots that I’m in Tokyo right now but the Vercel sign in was from Las Vegas - impressive. (It’s fine by the way. It was a CLI session I authenticated for @DevinAI who’s been working on a feature while I’ve been traveling.)
The one small line of text in the email about the location of the sign in would’ve been missed by most human assistants.
I open sourced this whole openclaw setup btw:
http://github.com/snarktank/clawchief

"Ron discovered how to be the investor of the future by accident. He didn't foresee the future of startup investing, realize it would pay to be upstanding, and force himself to behave that way. It would feel unnatural to him to behave any other way."
https://paulgraham.com/ronco.html

"Ron discovered how to be the investor of the future by accident. He didn't foresee the future of startup investing, realize it would pay to be upstanding, and force himself to behave that way. It would feel unnatural to him to behave any other way."
https://paulgraham.com/ronco.html

Had the pleasure of sharing the stage with @varunvummadi at Startup School today.
What an epic story.
Grew up without much money and spent time in college winning kaggle competitions.
He got into YC and was working on an idea around model finetuning. In shipping it, discovered that customer service was a big use-case in his user base.
Bet the farm on customer service. Landed Zepto and DoorDash as customers, beating out much bigger competitors.
Now, growing insanely fast and being a defining example of a new generation of Indian founders.

None of the critiques here of @germanrlopez's recent NYT essay even mentions the well-being of San Franciscans who don't use drugs: the kids who walk to school through open-air drug scenes, the spouses who are abandoned, the shop workers going bankrupt from shoplifting. The dismissal of the harm done to people who don't use drugs is a key reason why the perspectives here have been rejected by policymakers and the public. In public health, *everyone* should matter, not just people who use drugs but everyone else too.
https://www.nytimes.com/2026/04/18/opinion/san-francisco-addiction-homeless.html

California is projected to lose 15%+ of its student population while Florida may add 3.2% in the next 5 years.
Seems bad for school districts, teachers unions, the electoral college, the financial viability of various child-focused services that I, as a mom, rely on for my kids!
Marc Porter Magee 🎓: I don’t think people have fully absorbed just how big the declines in student enrollment are going to be.
Eight states are projected to experience DOUBLE DIGIT declines by 2031.

What I told 2,000 future founders in Bengaluru today:
1/ We believe we are at the start of a second wave of Indian companies that will build world-class AI native products for the global market. Emergent and Giga are the model of the future.
2/ Just because a space seems crowded doesn't mean it's too late. Zepto, Emergent, Giga - none were first movers. Second mover advantage is real.
3/ In fact, a good formula for finding startup ideas is to look at ideas that are showing some promise and just execute them better. Execution is everything: if you're an exceptional engineer, and you can build and move faster than your competitors, you'll win.
4/ There is every reason to believe Indian teams can beat US teams building global products. The level of engineering talent here is on a whole different level, and that's the key input.
5/ In the AI era, the best founders are the ones building at the edge of what's technically possible. You need to be experimenting wth the latest models, the latest open source projects.
6/ Stay in the flow of information. Watch the right podcasts, follow the right people on X. With AI changing this fast, you need to know what the smartest builders are thinking.
7/ Most of the best startups don't come from someone explicitly trying to start a company. They start from someone building a project just for fun, or tinkering with a new technology because they are curious. India needs more of this "tinkering" culture - this is how you have novel ideas when technology is shifting quickly.
8/ Founders are getting younger. Aadit was 18 when he started Zepto. The Giga founders were 20 when they came to SF. Young people who can learn very fast have the advantage right now.
9/ The best founders are pushing AI coding to the max. You can now write 20K lines of code / day. One person can do the work that just a year ago would take a 100 person team. The best builders are taking advantage and building at Garry Tan speeds.

Lots of people asking what’s so good about the new codex desktop computer use. Here’s 5 things that come to mind
1. operate Mac Apps without a great API: Slack, Google Sheets, Notes, IMessage without installing separate plugins. It instantly transforms all your apps into tools
2. If you need to operate your browser more visually it works really smoothly and fast (good for sites that are still human centric)
3. It uses its own cursor, keyboard etc so you can keep working.
4. Once you do any task once you can simply ask Codex to reflect on what it did and how it would accomplish the task next time with the benefit of hindsight and create a skill AND schedule an automation. It’s really nice that codex can just schedule and edit automations when asked! it’s very Claw like in this way. This last point is not computer use specific but is powerful when combined with computer use
5. The UI polish is insane: you get nice icons for any application you want to tag into computer use plus all the other built in new stuff like built in file viewer and browser so there is no context switching. So you can iterate really fast and not lose focus. Because of the polish it also feels nice and delightful to use.
Hamel Husain: Seriously stop everything you are doing and use codex desktop app new computer use. Absolutely mind blowing

Watched Peter Steinberger’s @TEDTalks today, and one idea stood out clearly:
The biggest AI shift may not be smarter models. It may be giving ordinary people the power to build.
When agents let non-technical users create products, automate workflows, and turn ideas into reality, innovation is no longer limited to engineers.
That changes who gets to participate.
Worth watching: https://www.ted.com/talks/peter_steinberger_how_i_created_openclaw_the_breakthrough_ai_agent?utm_campaign=tedspread&utm_medium=referral&utm_source=tedcomshare
@steipete

At some point, usually in your 20s, you'll notice that the people around you stop believing in themselves. And no matter how hard you try, you can't save them. By all means, do not let it infect your mind. Stay on your path.

Here's my new tutorial with a live demo of everything you can build with Claude Design.
I cover how to create videos, slides, websites, apps, and even an initial design system all in 16 minutes.
Claude Design is super fun and you'll burn through your usage fast :)
📌 Watch now: https://youtu.be/WMnk1LFBMqA

I went to Stanford's grad school of education where Boaler was working before I knew how bad the state of education science is and how among other things ed grad school doesn't replicably make teachers better. I wasn't teaching math so we never interacted, but she was talked about universally as the main celebrity there. Never saw any pushback and I didn't look into the specifics of what she was about until later.
Talking about how bad the state of education science is as a teacher can make you sound like a conspiracy theorist, it's hard to communicate in a way that sounds grounded how fake a lot of the things being promoted by prestigious institutions are. I'm really excited Kelsey's taken this problem on because there needs to be a lot more good thorough writing on it. Wish this wasn't paywalled so I could share it with more people but imo it's worth the subscription.
https://www.theargumentmag.com/p/education-research-is-weak-and-sloppy

The @nytimes and @wsj keep publishing photos of Sam Altman’s house over and over again. You can’t keep posting op-eds about the risk of stochastic terrorism and not be institutionally aware of what you might be enabling.

if you are the only one who can build this amazing tech and believe it is super dangerous, it’s very easy: you quit for the sake of humanity.
if you are not the only one who can do it, you are definitely over-valued.
you can’t claim to be both.

I haven’t given this the time it deserves yet, but my first impressions…let’s just say my new CMO and SEO agent found something pretty significant, like needle moving!
Matt Shumer: This is why https://agent-s.app exists

Introducing: Browser Harness. A self-healing harness that can complete virtually any browser task. ♞
We got tired of browser frameworks restricting the LLM. So we removed the framework.
> Self-healing — edits helpers. py on the fly
> Direct CDP — one websocket to Chrome
> No framework, no rails, complete freedom
> Drop-in for Claude Code and Codex
I challenge anyone to find a task that DOESN'T work. I couldn't yet.🔥
100% open source ↓

Terminal automation + e2e testing solved
Now as simple as snapshot, click, type:
– wterm renders terminal-in-html, every cell in the a11y tree
– agent-browser automates pages via the a11y tree
Here's opencode in one browser driving Claude Code in another

You can access 𝕏 APi via @OpenClaw.
We’re trying to make it affordable without giving away the shop.
Hopefully, this can be useful & fun 💫
Robert Scoble: Holy shit.
Now everyone will be able to use their @OpenClaws and all the other agentic platforms to build apps on top of X.
Here's the secret: build lists.
Lists are how you build apps.
The pattern:
Build a list of your favorite football team. Or whatever you are into.

Re @elonmusk @openclaw Grok has gone OpenClaw, unleash the lobsters 🦞

This makes any agent a full employee:
1. e.g. creates videos with seedance 2
2. uploads the videos on Tiktok every 4h
3. analyses best hooks and repeats.
- It can modify its own harness. Therefore no task is impossible.
- Shared domain specific skills - agents learn from each other to be faster next time.
Gregor Zunic: Introducing: Browser Harness. A self-healing harness that can complete virtually any browser task. ♞
We got tired of browser frameworks restricting the LLM. So we removed the framework.
> Self-healing — edits helpers. py on the fly
> Direct CDP — one websocket to Chrome
> No

It's time to learn to Build it. Ship it. Vibe it. Get it into production. For real. We'll make you an agentic expert.
Together with @Replit at 2026 http://SaaStrAIAnnual.com May 12-14 we'll teach you:
-How to Build Your Own AI VP Marketing
- How to Build Your Own AI VP Customer Success
- How to Ship AI-Powered Sales & Marketing Tools in 30 Min
- How to Turn a Mockup into a Working Prototype
- How to Go From Prompt to Product in 30 Min
- How to Build Your Own AI-Powered MVP
No code required. Just bring your laptop.
We'll give you the prompt.
http://SaaStrAIAnnual.com 2026. May 12-14 in SF Bay!!

1/ In the AI/robotics revolution, the biggest winner is actually going to be @ycombinator—because the world is about to need 100x more founders. And YC creates great founders at scale.
Sounds weird but hear me out for a hot second 🧵👇

i am forced to recommend viewing of sunil's talk if i want to retain healthy knees
https://www.youtube.com/live/O_IMsEg91g8?t=31662&is=WIMl-Di3-Xi6gtse
sunil pai: You should watch these 2 talks from AI Engineer Europe, from the "Vienna school of agentic coding":
@badlogicgames "Building pi in a World of Slop" https://www.youtube.com/watch?v=RjfbvDXpFls
@mitsuhiko / @cristinaponcela "The Friction is Your Judgment" https://www.youtube.com/watch?v=_Zcw_sVF6hU
They're very good.

Try our new 𝕏 API CLI tool with SKILL.md added to merge with @openclaw!
We will also make calling Owned Reads a fraction of the cost starting Monday 4/20.
Elon Musk: You can access 𝕏 APi via @OpenClaw.
We’re trying to make it affordable without giving away the shop.
Hopefully, this can be useful & fun 💫

recommended viewing. specifically because cristina is awesome.
first ever talk is a keynote in front of a huge audience and she didn't even flinch. inspiring and really happy she's on the team.
https://youtu.be/_Zcw_sVF6hU?is=CQFlWnWZk5qApBNz

Re The tensor engine was first implemented inside SN3 (before it was called Lush) in 1992 at Bell Labs by Léon Bottom and me.
The naming convention has survived to this day in PyTorch and other libraries.
The naming of the tensor operations was reused in EBlearn (C++ deep learning library written by Pierre Sermanet and me, with some help from @soumithchintala).
It was recycled in Torch5 and Torch7, which was written largely by Ronan Collobert, and my students Clément Farabet, and @koraykv ). Clément and Koray had been brought up on Lush (the open version of SN) and knew the nomenclature.
Then, Soumith used the same conventions in PyTorch.

There are two options: listen to people who actually study the effect of technological development, or to someone who wants to sell you the next Figama
TFTC: Anthropic CEO Dario Amodei: “50% of all tech jobs, entry-level lawyers, consultants, and finance professionals will be completely wiped out within 1–5 years.”

Dario is wrong.
He knows absolutely nothing about the effects of technological revolutions on the labor market.
Don't listen to him, Sam, Yoshua, Geoff, or me on this topic.
Listen to economists who have spent their career studying this, like @Ph_Aghion , @erikbryn , @DAcemogluMIT , @amcafee , @davidautor
TFTC: Anthropic CEO Dario Amodei: “50% of all tech jobs, entry-level lawyers, consultants, and finance professionals will be completely wiped out within 1–5 years.”

Indeed. The history of tech impact on labor is well-documented, including by those named. It's unpredictable, but usually improves productivity and leads to expansion. Law & white-collar workers aren't horse-buggy drivers or elevator operators. They will use AI and adapt.
Yann LeCun: Dario is wrong.
He knows absolutely nothing about the effects of technological revolutions on the labor market.
Don't listen to him, Sam, Yoshua, Geoff, or me on this topic.
Listen to economists who have spent their career studying this, like @Ph_Aghion , @erikbryn ,

🆕The Friction is Your Judgment
https://www.youtube.com/watch?v=_Zcw_sVF6hU
@mitsuhiko (creator of Flask) and Cristina Poncela Cubeiro (AI-native Engineer) break down what goes WRONG when you "ship without friction" — listing many areas of agentic coding where you are enticed to turn your brain OFF when you most need to turn it ON.

Interestingly, the tech centimillionaire pretending to be socialist Congressional candidate @saikatc is both 1) trying to buy the election with his Stripe wealth and 2) majority funded by astroturf donors that don’t even live in California, let alone San Francisco.
San Francisco Chronicle: California’s 11th Congressional District election is taking shape as Scott Wiener and Saikat Chakrabarti pull ahead on money and donor support. https://www.sfchronicle.com/projects/2026/ca-house-11-campaign-finance/?taid=69e37ba34d22860001045cfc&utm_campaign=trueanthem%2B3988&utm_medium=social&utm_source=twitter