Founders overestimate how hard it is to become a top 100 expert on most subjects.
For most things you need to solve, there’s usually a very clear path: just go do the work, obsessively, and become friends with everyone else doing it.

Re @garrytan My block went from most car break ins in the country to very safe / no car break ins at all because a Flock camera was installed.

Re @garrytan this logic for Flock Safety is basically the civic version of how cloudflare changed the internet by making ddos protection the default

44% drop in major crimes.
54% drop in car thefts.
In just 2 years.
It’s a choice.
Garry Tan: San Francisco is safer because of Flock Safety. Every city can be safer. We don't have to choose a world where people are unsafe. It is a choice, though.

Dial Moments: Audio Guestbook is ranked first on https://www.unicor.ne/?t=2026-05-22 (got alerted by @marclou @trust_mrr ) 🙌🏽
Built with @Replit over a weekend 🔥 have focused on marketing with TikTok and lemon8 ✨ @raymmar @MannyBernabe

Yesterday @sama just offered to invest $2M of OAI credits into any current YC company as an uncapped SAFE.
If I was in YC today, here is why I would take the deal:
1. You will spend that much $ on tokens building your product quicker than you think.
The best AI-leveraged engineers are spending ~$10k+ per month on tokens. Thats $1.2M/year for a 10 person engineering team. The OAI deal means you don’t have to think twice about accelerating your engineering.
2. OAI tokens are worth 2X Anthropic tokens.
Our frontier models are ~50% more token efficient than Ant’s. This means our $2M in GPT tokens is worth $4M in work accomplished with Opus.
3. You can use the tokens on the API.
This means you can offer agentic products to customers without worrying about price or charging while you find PMF. Any product you offer that is useful with agents will use a non-trivial amount of tokens. Worrying about those costs while finding PMF doesn’t seem worth it.
4. The dilution will be minimal if you find PMF as an AI-native product and raise a Series A.
A lot of series A’s I see these days are in the $100-$200M valuation range. $2M at that valuation is 1-2% dilution, which is very worth it if it allowed you to defer raising as much in your seed in order to pay for tokens.
Would be curious how folks are thinking about it! Definitely an interesting offer to consider 👀

A side effect of this I'm very excited about: it solves the "tokens are too expensive" problem for consumer AI ideas.
by the time you eat the $2M in the credits, say over 1-2 years, token costs at the current level of intelligence will probably fall 10-20x.
so all you need to do is make something people want with the current intelligence levels, and you won't touch your cash runway now and token costs will go to 5% of what they are now by the time you run out of credits.
Y Combinator: OpenAI is offering $2M in tokens to every YC company in the spring and summer batches.
We extended the summer deadline to May 25 so more founders can get in on it.
http://ycombinator.com/apply

Game changer Codex automation 🙂

Karaokebench (TM @swyx) is insanely fun and gpt advanced voice does hilarious things beyond singing - which makes it actually an awesome bench. So much randomness.
Also, everyone should watch @swyx's karaokebench (video below) - our guy has a velvety voice.
I have a working theory that better singers are better primers, so the bench goes both ways. (swyx's gpt is much more focused than mine)
You know what they say, "When you sing long into the abyss [well/poorly], the abyss also sings back-" 😉
Anyway, I gave it a go. You can watch a very late night minu after a full day and a "why-are-you-making-me-sing" gpt.
A few of my favorites:
> An alarmed me at gpt's very, very off-pitch mid-duet: "oh my god, I think that's wrong. Where are you getting that from"
> excited gpt cutting off my turn to belt "STAND BY ME - DARLING DARLING" and then immediately losing enthusiasm in the same instance
> me: "Choose another song"
gpt: "(straigtup refusing) I'd love to, but I think we've had our moment of musical magic."
> me: "do you know the national Korean anthem?"
gpt: "I do know of it, but out of respect, I wouldn't sing a national anthem like that."
me: "Please. I think it would make me really happy. It's a song that I grew up singing."
gpt: "I completely understand it holds meaning for you. Let's pick something special that we can both enjoy."
Just delightful - singing is so unchartered that I get sonnet 3 days excitement where you are delightfully surprised by every right pitch, and entertained by every wrong one. I might normalize late night kakaobench routines.
swyx🛬 SFO: @minu_who i do karaoke bench also
https://x.com/swyx/status/1928270740517724262?s=46

This is HOW my Hermes and OpenClaw Agents share a single brain.
Hermes Agent is the orchestrator with OpenClaw agents as the squad.

Closed and gated surveillance economy SaaS systems and models will get replaced by open source in two tiers:
- The models themselves
- The everything-ultra-app harness
If an American open source champion does not emerge, the world will run on whoever produces the best open models and open software that is strong, stable, affordable, customizable, expandable and adaptable to both individual and business needs.
We don't need another WinTel dynasty in the agent era, or an everything is "free" so we can monitor you, censor you and spy on you era.
As AI becomes our interface to the world it needs to be open and people powered.
Hedgie: 🦔Microsoft canceled its internal Claude Code licenses this week after token-based billing made the cost untenable, even for a company with effectively infinite cloud resources. Uber's CTO sent an internal memo warning the company burned through its entire 2026 AI budget in just

It’s a rough time to be a scientist.
You spend decades learning how to measure things properly.
How to separate signal from noise.
How to distinguish hazard from risk.
And then you open your phone and watch fear beat data - again and again and again.
In the lab, we argue about controls, concentrations, confidence intervals.
Online, people argue about screenshots, podcasters and blogs.
This little molecule in my hand?
The dose matters.
Exposure matters.
Evidence matters.
Being a scientist today means doing more than running experiments.
It means explaining, clarifying, correcting - over and over and over again.
So yes - it’s a rough time to be a scientist.
But it’s also the most important time to speak up.

Open weight models running on open source harnesses solve this problem.
One of these days, American companies will wake up to the same solution we pioneered decades ago that Chinese teams are now running against the closed source new wave East India companies of American AI.
Ethan Mollick: We are quite short of compute, and that is going to result in compute becoming very expensive for complex agentic workflows even as single-turn chatbots get cheaper. So the richest companies & most pressing use cases will use AI agents & everyone else will be stuck with chatbots?

My new article "Toward a science of intelligence: unifying physics, neuroscience and AI" https://www.amacad.org/publication/daedalus/toward-science-of-intelligence-unifying-physics-neuroscience-ai
published in the Daedelus journal of @americanacad
Its part of a special issue on AI+Science with many amazing contributors lead by James Manyika https://www.amacad.org/daedalus/ai-science-what-is-the-future-of-discovery

Eight key points from the most recent essay in the “AI as Normal Technology” series by @sayashk and me.
Do AI Risks Require Extraordinary Government Intervention?
1. There is general consensus that AI is so far a “normal” general purpose technology when it comes to its economic and labor market impacts, but there is debate about whether its safety risks are so abnormal as to warrant extraordinary government responses.
2. What are these “extraordinary” government interventions that are problematic in a liberal democracy? They tend to (1) be based on anticipated harms rather than realized or demonstrated ones (2) impose burdens on actors not directly responsible for the harms (3) enacted with unilateral authority, bypassing the normal process of governance.
3. Voluntary commitments and export controls are relatively modest interventions. But we must recognize that the most they can accomplish is buy us a few months of time. Unlike nuclear nonproliferation, AI lacks a physical bottleneck like enriched uranium.
4. So AI nonproliferation risks creating a slippery slope as AI capabilities continue to advance. We might quickly enter a state where governments exercise control over what AI research and products can be shared publicly. Advocates for nonproliferation must state what their bright lines are — otherwise it’s reasonable for skeptics to assume that there will be escalating calls for more authoritarian interventions down the line.
5. Nonproliferation is brittle because it relies on a single chokepoint. The dam will break — it’s a matter of when, not if. Our preferred approach is resilience, which distributes defenses across society.
6. While LLM-aided cyber-vulnerability detection is powerful, it is not as if we have superhuman vulnerability detection for the first time! This stylized spectrum of vulnerability detection capability (see image) illustrates that we crossed that point long ago. And we managed to navigate the transition without imposing any restrictions on the tools. Today we use them effectively for defensive purposes. Of course, the transition wasn’t smooth or painless.
7. A resilience approach to AI cyberrisk would emphasize things like AI-assisted red-teaming not just for tech companies, but for schools, hospitals, power grids, small businesses, and government systems that currently lack the capacity for defense.
8. But if resilience is so helpful, why haven't we prioritized it already? The problem is we are not great at normal policymaking. It requires polycentric governance in which many decision-makers work harmoniously together. This is a tough sell given that state capacity in the United States has been hobbled by decades of accumulating veto points and creeping proceduralism. As a result, unilateral actions by the executive branch are often seen as the way out for developing and enforcing AI policy.
So we understand why extraordinary government interventions are tempting. But AI is not the last digital technology that will pose major risks, nor is this the last round of AI capability improvements. Getting our policy act together is hard, but important—not just to address the current challenges, but for all future responses to technology-enabled harms, and for the democratic process to work more generally.
Full essay published on the @knightcolumbia website: https://knightcolumbia.org/blog/do-ai-risks-require-extraordinary-government-intervention

I hate seeing all the mass layoffs. Here are 6 things you can do as an employee to take back control:
1. Read the signals
If business growth is stagnant and leadership is suddenly obsessed with "flatter orgs" or "restructuring for the agentic era," you already know what's coming.
2. Learn Codex or Claude Code
These apps are the best training grounds for working with AI agents. Tell them about your job and ask them what you can build. If your company doesn’t let you use these tools, use them for your personal projects.
3. Build side projects
Ship a small product or make an open-source contribution. If you work at a large company, your builder skills have probably atrophied. It’s time to bring them back.
4. Develop a GitHub history
If you do step 3 consistently, you’ll develop a body of work on GitHub over time. My friend @zarazhangrui is a great example — she built a frontend-slides skill that now has 16K GitHub stars and has been shipping non-stop since.
5. Become top 10% at your craft
AI gets people to average really fast. But that just means customers are willing to pay more for human craft and taste. Pick one skill you genuinely enjoy working on and put in the reps until you’re in the top 10% at it.
6. Let the market determine your value
Build in public, solve real user problems, and craft great products. That’s how you’ll get noticed by great employers, not by submitting 100s of resumes. And if all else fails, consider becoming a founder. I think entrepreneurship is the safest job in the AI era anyway.

“Models follow culture. Humans move it first.”
Amjad Masad, founder and CEO of Replit, says current LLMs are still downstream of existing culture.
They’re great at scaling what already works. But they struggle with the edge cases where new demand actually appears trends, taste, naming, internet-native behavior. A model doesn’t invent “looksmaxxing.” It learns it after humans make it real.

Most vibe-coded apps forget every user who opens them.
With one prompt to Replit Agent fixes it! 💻
Try adding authentication to your app today!

Couple days back @swyx posted a challenge: code a ~10M transformer in JAX/Flax/Optax, run it in free Colab, and train it on addition w/ your agent!
I gave Codex the screenshot + /goal.
It controlled Colab through Chrome, used my signed-in session, handled runtime/editing weirdness, ran the T4 job, then used subagents to audit the result.
End state: 10,652,557 params, ~19 min train, 99/100 exact random checks 🤯
Still needs cleaner evals, but autonomously babysiting the training run over chrome is pretty wild!
Vaibhav (VB) Srivastav: http://x.com/i/article/2057864011719344128

🚨 🚨 🚨 NEW: A shocking @USCIS memo seems to declare that hundreds of thousands of immigrants living in this country and applying for green cards must instead apply for visas abroad; which could MASSIVELY disrupt lives.
🧵 on what we know, and what we don’t.
USCIS: USCIS is applying long-standing law and prior court decisions to require certain aliens with temporary visas who decide they want to permanently reside in the U.S. to return to their home countries to apply for permanent visas through the @StateDept.
We're returning to the

New connectors available
Build sites with Vercel, create anything in Canva, design decks in Gamma, and tap into market data from S&P Global.

Excited about this one, and @Replit, thanks for the tools to make it happen.
Aaron Horne: DemoGuards is live.
It is a Chrome extension and web dashboard that automatically blurs sensitive data during demos, including emails, phone numbers, API keys, JWTs, names, and customer data.
Protect every demo.
https://demoguards.com
#DemoGuards #SaaS #SalesEngineering

Howard Lutnick's FIRST political donation since becoming Commerce Secretary was a $5 MILLION donation to House Republicans 1 month before they interviewed him about his ties to Jeffrey Epstein.
 
Let that sink in.
https://www.nytimes.com/2026/05/22/us/politics/howard-lutnick-donation-house-republicans.html

The new White House policy requiring green card applicants to apply from outside the US is a capricious attack on legal immigration. It will hurt families, leave us with fewer doctors, teachers and scientists, and hurt American competitiveness in AI.

an enormous gift to China
USCIS: USCIS is applying long-standing law and prior court decisions to require certain aliens with temporary visas who decide they want to permanently reside in the U.S. to return to their home countries to apply for permanent visas through the @StateDept.
We're returning to the

Mark Twain, the time nears

This is the author who thrust AI water hysteria into the mainstream by overestimating data center water use by 100,000% in her book Empire of AI by mixing up units.
She is a source of wild misinformation. Imagine writing an Econ book on the premise that minimum wage is $7,250/hr
Karen Hao: On the one-year anniversary of EMPIRE OF AI, I am so, so excited to announce The AI Resist List, a new project that documents examples of resistance to the AI empires around the world.
http://airesistlist.org

This claim is a LIE. Congress created adjustment of status in 1960. It's not a loophole. In fact, @USCIS's OWN WEBSITE says that getting a green card from inside the United States was expressly intended by Congress!
Just wait; give it 24 hours and they'll erase that history.
Homeland Security: An alien who is in the U.S. temporarily and wants a Green Card must return to their home country to apply.
This policy allows our immigration system to function as the law intended instead of incentivizing loopholes.
The era of abusing our nation’s immigration system is over.

I have known @garrytan a long time. It was great to sit down with him at YC and explore what the next generation of founders need to do to protect themselves from these forces.
YC touches so many founders and, through them, millions of other people. When they help those founders stay incorruptible, we all benefit.
Y Combinator: Eric Ries (@ericries) is the author of The Lean Startup, the NYT bestseller that became a playbook for a generation of founders. He's also the founder of the Long-Term Stock Exchange and has a new book out called Incorruptible: Why Good Companies Go Bad and How Great Companies

The legendary art tech program that sparked the world’s first art NFT comes to Vibecon, pairing leading artists and technologists to make something new. Featuring @gucaslelfond, @frogspitsimulat, @delibeat, and @poolnoodle93.
See @rhizome’s 7x7 take the stage on day one of Vibecon.
NYC, June 17–18.
Get your tickets at http://vibecon.ai

a framework for agent memory: Remember, Cite, Forget.
here's one way to do it. mine looks like this:
→ Remember: hot session takes lossless-claw's pattern (raw in SQLite, grep originals when needed), cross-session lives in gbrain with provenance.
→ Cite: authority order written into AGENTS.md. higher tier always wins.
→ Forget: timestamped facts in gbrain plus Mem0-style soft decay on retrieval.
short version: an agent pulls memory from many places at once. miss any of the three and it confidently uses stale facts, unauthorized sources, or both.
Vox: http://x.com/i/article/2057843767126700032

I understand why we don’t want people to come to the US to be criminals, mooch on welfare, open learing centers and otherwise undermine the country.
But I don’t understand why we make it harder for motivated, ambitious, hardworking people to come to the land of opportunity.
Homeland Security: An alien who is in the U.S. temporarily and wants a Green Card must return to their home country to apply.
This policy allows our immigration system to function as the law intended instead of incentivizing loopholes.
The era of abusing our nation’s immigration system is over.

In a new statement, U.S. Citizenship and Immigration Services spokesman Zach Kahler suggests that green card applicants in the U.S. who will provide an "economic benefit" or serve the "national interest" will be allowed to complete their processing here, without having to leave.
Camilo Montoya-Galvez: Major policy change: The Trump administration is making it much harder for immigrants already in the U.S. to get permanent residency, or a green card, without having to leave the country.
More on @CBSNews:
https://www.cbsnews.com/news/trump-green-cards-leave-us/

This is rotten to the core.
Howard Lutnick, the Commerce Secretary, cut a $5 million check to the Congressional Leadership Fund on April 1, just weeks after agreeing to testify before the House Oversight Committee about his past ties to Jeffrey Epstein.
That testimony happened on May 6. The super PAC he wrote that check to is the main vehicle backing House Republicans and Speaker Mike Johnson, the very people deciding how hard to press him.
Lawmakers were reportedly weighing a subpoena before Lutnick agreed to show up voluntarily.
His name appears in more than 250 Epstein-related records.
He long claimed he hadn’t seen Epstein after 2005, but records show he visited Epstein’s private island in 2012 with his family. He called the encounters “meaningless and inconsequential.”
A $5 million donation from a sitting cabinet secretary to the party running the investigation into him is not a coincidence. It is not normal. It is not how a healthy democracy works.
If a Democratic cabinet official did this, every Republican in Washington would be screaming for hearings, indictments, and resignations.
The silence tells you everything.
Release the Epstein files. All of them.
The American people deserve to know who protected this monster, who profited from his network, and who is paying to make the questions go away. https://www.mediaite.com/media/news/howard-lutnick-cut-5-million-check-to-house-republicans-after-scheduling-his-epstein-testimony/