if you can master the meta-skill of figuring out what problems in arbitrary domains are computationally tractable, you will have the opportunity, for at least a year two, and maybe longer, to be a kind of meta-genius. you will not know the answer to anything, or even how to find it, but you’ll have refined heuristics for the right questions to ask about everything to make meaningful progress along the margin. this is probably the skill to have optimized for in the last three years, though I readily admit I don’t know how long it will remain a human advantage. it is for now though.

Excited to take part in Swyx's Help Kill My SaaS hackathon this weekend 🫡
Follow along for updates as I attempt to prompt my agents from my phone while playing dnd tomorrow 😂
swyx: $10,000 kill my saas in a weekend competition is live!
TECH STACK:
any coding agent
any model
up to $500 in token spend incl subscriptions
see luma for description. join waitlist if late to this - finish line extended to Wednesday.
the brief is live and people are prompting

Re @Replit Design
ThreeJS Studio
This is beyond ridiculous
#Replit #Design @threejs #BuildInPublic

Has Trump’s invasion of Greenland already begun? An American oil company connected to Donald Trump is preparing to sink wells in remote Greenland despite not having received permission from the local authorities. https://trib.al/NA5w2ys

It seems a common experience in the lives of people with high agency is that they, at some point, metaphorically, lean against a wall and discover that there is a hidden door. That some constraint they’ve been told about turns out not to hold, and this allows them to experience something positive. And from this they get into the habit of testing if walls are actually doors, and the more they do so, the more examples of hidden doors they find, and the more inclined they are to look for them. Far from the only thing that sets them apart, but having that experience can be transformative
— Henrik Olof Karlsson

A 4.1% unemployment rate sounds pretty good. But that number only counts people actively looking for work.⁣
⁣
If people give up searching, unemployment can fall even as the job market gets weaker.⁣
⁣
https://newsletter.platypuseconomics.com/p/the-july-jobs-report-should-worry

Flexport CEO Ryan Petersen explains why your initial vision for your company is probably wrong
When Ryan first applied to Y Combinator he wanted Flexport to be “TurboTax for customs.” But that vision changed a lot. As Ryan explains:
“The more you learn and the more customers you talk to and the more vendors you talk to, the more you realize how screwed up the world is in this space. Then [the vision] actually gets bigger and bigger and keeps expanding.”
Y Combinator CEO Garry Tan says this is a common pattern he sees in startups too:
“There’s a misconception that you wake up from a haze and you’re like ‘This is it. This is the vision.” But the story is actually ‘Oh here’s this thing that I know I can solve.’ And then once you solve it and people really like it, it sort of expands and expands.”
Ryan concludes:
“If a founder is too certain of exactly what the vision is and what the product’s going to look like and how it’s all going to work, they’re probably wrong. The world is super messy.”
Source: @garrytan (Mar 2022)

Strong disagree: there is a cemetery of breakthrough silicon that got nowhere due to lack of software
today there is silicon that's already 2-3x more efficient than nvidia and can be bought off the shelf, and yet...
Robin Rivaton: La course à l’IA ressemble à une course au logiciel. Elle ne l’est pas.
Elle se gagnera dans le monde physique : puces, fabs, mémoire, data centers, réseaux électriques et production d’énergie.
Car elle n'est que la continuation de la révolution technologique des

75% of Americans fear AI. In China, 80% are excited about it. We are facing a narrative gap, and are losing the fight for it.

GPT-4 finished training four years ago today.

The biggest bottleneck to building a great AI agent isn't the model.
It’s when teams:
→ Bury their agent in too much context
→ Don't give it tools to find what it needs
→ Try to cover too much vs. nail a few core use cases
In my next episode, Nan and Jacob from @Linear walked me through exactly how they built a production agent from idea to launch. Below's a picture of the product memo that started it all.
📌 Subscribe to get our full interview tomorrow: https://www.youtube.com/@PeterYangYT?sub_confirmation=1

Y Combinator CEO, Garry Tan, took the stage for 42 minutes at Startup School 2026 and explained how to build your own personal AGI better than any paid AI course.
This is what he told the room:
1. The leverage is in your context, not the model.
Tan watches hundreds of founders use identical models every batch.
"There are 2x people and there are 100x people who are using the same Claude. Same weights, same context window size, same API. But the leverage is not in the weights."
The gap between users is now bigger than the gap between models.
2. One person's output went up 400x.
In 2013 Tan shipped maybe 14 useful lines of code a day as a YC partner, dead on the median for programmer productivity.
"I did the math on my output, and I'm at about 400x what I did in 2013."
3. Agents run on a different working memory.
Humans hold 7 things in their head at once. Every org chart and checklist ever built is a patch for that limit.
"An AI agent holds a million tokens. That's about a thousand pages. Three Harry Potter books sitting open on its head all at once."
You're still running your week on tools built for the 7-digit brain.
4. Markdown is code now.
Tan's stack is mostly skill files: pages of plain English an agent can execute.
"If you can write clear instructions in English, you're a programmer. The compiler is a language model."
At YC, finance and events staff who never opened a terminal are building automations.
5. Your history is your moat.
Tan's agent runs on a personal wiki: about 220,000 markdown pages covering 25 years of email, meetings, notes and decisions.
"When my agent does anything, it does knowing everything I know. And that's the difference between an assistant and a colleague."
No frontier model has your context. That's the one asset nobody can replicate.
6. Never do one-off work.
Most people run a task with an agent, close the window and throw the learning away. Tan ends every task by having the agent turn what it did into a reusable skill file.
"If you have to ask for something twice, you failed."
Captured skills compound daily. Amnesia resets you to zero every morning.
7. Own your skill files before your employer does.
A skill file is your judgment, extracted and executable. The only question is who controls it.
"Own your skills because if you don't, your job becomes a skill file."
Files in your repo compound your career. Files in the company's repo run your judgment without you.
Watch it, then read the step-by-step guide on becoming an AI engineer.
AI Guides: http://x.com/i/article/2074513567701680128

Join us in celebrating the upcoming 1 billion downloads of Gemma models🥳
A fun in-person evening with live demos and great people from the open models community
Excited to see you all there!

This week’s Paper Club is all about robotics. Every year for the last decade, someone has promised that the era of robotics is just around the corner. But we’re still waiting.
So we gathered a bunch of the top researchers working in AI and robotics to present the latest findings on where we are and what comes next.
Thanks to the following presenters:
0:00 – @FrancoisChauba1: Ten years of “next year, robotics is solved”
7:59 – @marceltornev: MEM - Multi-Scale Embodied Memory for Vision Language Action Models (http://arxiv.org/abs/2603.03596)
20:21 – Milan Ganai: Self-Supervised Bootstrapping of Action-Predictive Embodied Reasoning (http://arxiv.org/abs/2602.08167)
33:42 – @tylerlum23: SimToolReal - An Object-Centric Policy for Zero-Shot Dexterous Tool Manipulation (http://arxiv.org/abs/2602.16863)
51:21 – @NikolausWest: Why the next great robotics companies will start with teleoperation
1:08:30 – @BillJiao930 & @Guanming717: World action models and what comes after VLAs

Coolness come from low cost housing.
Low cost housing ➡️ No existential fear of being in the permanent economic underclass ➡️ People take risks personally, professionally, creatively ➡️ Coolness.
Mason Warner: We ought to romanticize SF more if we ever want this city to be actually cool

My Nobel lecture and the video of the Stockholm lecture are online.
It covers what creative destruction tells us about secular stagnation, firm dynamics, the middle-income trap, and AI as the next GPT revolution.
Paper: https://www.aeaweb.org/articles?id=10.1257/aer.116.7.2324
Video: https://www.nobelprize.org/prizes/economic-sciences/2025/aghion/lecture/

Want to learn what is SOTA right now in Robotics from phds at Stanford, Physical Intelligence, Waymo, http://Rerun.io, and YC Startups?
Y Combinator: This week’s Paper Club is all about robotics. Every year for the last decade, someone has promised that the era of robotics is just around the corner. But we’re still waiting.
So we gathered a bunch of the top researchers working in AI and robotics to present the latest findings

An entire generation of kids will now be woefully unprepared for the real world. America last. Equity. 😡
Geiger Capital: Absolutely incredible…
University of California professors are now asking the school to reinstate the SAT for admissions because their new students are unable to do high school-level work.
They stopped using standardized tests for admissions in 2020 to promote "equity".

Two days ago, I signed up for the SmolForge alpha and asked Opus to deploy a SolidStart app.
I figured @swyx was the kind of person who'd appreciate a list of papercuts, from my clanker to his. It's linked in the quoted post.
I've been thinking about that feedback loop ever since.
If agents are going to be users of our software, they need a first-class way to tell us how to improve it.
So I built papercuts into the Minima CLI: https://docs.minima.ltd/cli/papercuts/
swyx: people are sending in papercuts!!! wow!! https://forge.smol.ai/gists/7676599e7861419987eda31640f47b20

6 hours into building @swyx's "help kill my saas" hackathon platform.
~30% done.
10M+ tokens burned already 💀
not polishing. just prompting, building, breaking things, fixing things, and pushing forward.
let's see how far i can get this weekend.
Ali Zaid: doing something fun this weekend.
taking part in @swyx's "help kill my saas" hackathon.
i'm kinda late. like 6 hours late 😭
but let's see if i can cook something.
the odds of failing are pretty high, so i'm not expecting anything.
just gonna build, experiment, and have some