Silicon Valley's congressman wants to tax the AI his own constituents build, while his household trades $34M in tech stocks.
CA-17 deserves a representative who understands that punishing American AI gives the advantage to China. Funny how the Ranking Member of the China Committee keeps doing things that benefit Beijing.
Vote @ethanagarwal and primary Ro Khanna
Ro Khanna: We need to tax agentic AI more than workers.
Our tax code provides bonus depreciation for AI but is stacked against human ingenuity.
It makes no sense to provide tax breaks for capital in a capital biased world.
Kane 謝凱堯
Today San Francisco public school district (53% reading, 42% math proficiency) told me they are unable to find any invoices or documents about the “adult supremacy” seminar they hosted.
They seem to have lost them all.
Kane 謝凱堯: San Francisco public schools @SFUnified told me that they need a month to find documents about the "Adult Supremacy" seminar they hosted:
Why?
The Wall Street Journal: Most green-card applicants will need to go abroad to apply for permanent residency at an American consulate, rather than filing from within the U.S. as they do now, the Trump administration announced Friday. https://on.wsj.com/4v2Fqkr
Peter Yang
Lately, I've been obsessed with how the best solo founders and engineers use agents to 10x their output. I ask them questions like:
1. What's your AI stack?
2. Show me how you build end to end.
3. How do you manage multiple agents?
📌 Subscribe to get all these episodes soon (episode with Ryan dropping this Sunday): https://www.youtube.com/@PeterYangYT?sub_confirmation=1
Lately, I've been obsessed with how the best solo founders and engineers use agents to 10x their output. I ask them questions like:
1. What's your AI stack?
2. Show me how you build end to end.
3. How do you manage multiple agents?
📌 Subscribe to get all these episodes soon (episode with Ryan dropping this Sunday): https://www.youtube.com/@PeterYangYT?sub_confirmation=1
thibauld
During his second term, Trump will have cut 2 of the US most powerful innovation and wealth creation engines:
1. skilled legal immigration
2. (non-defense) research budgets
We won't see the effect of this immediately, but in 10 years it will be obvious... and it will take at least another decade to rebuild 😢
A sad day for the US.
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.
James Blunt
The irony in all of this is incredible.
Just a few years ago, the Department of State was literally testing programs to allow certain visa renewals to happen INSIDE the United States so legal immigrants wouldn’t have to fly across the world just to get a stamp in their passport.
Everyone understood how pointless, expensive, disruptive, and inefficient the process was.
Imagine flying 20 hours to another country…
Taking weeks off work…
Pulling kids out of school…
Risking visa delays…
All so someone can place an ink stamp in a passport for a visa category you already LEGALLY hold.
Even the government recognized how absurd that system was. Now suddenly USCIS is pushing the philosophy that people should leave America and consular process abroad because Adjustment of Status is supposedly some “EXTRAORDINARY ACT OF GRACE.”
One arm of the government was trying to MODERNIZE legal immigration and reduce pointless travel burdens.
The other arm is now trying to revive 1950s immigration logic in a world where people wait DECADES for green cards.
At some point the system becomes so inefficient it almost turns comedic.
Geoffrey Moore says startups die in the chasm because pragmatist buyers demand a "whole product." These folks won't tolerate gaps. They need references. They need the complete solution. The chasm is lethal because because the buyers won't buy without perfection.
But Moore's model assumes there's an EXISTING solution the buyer is comparing you to. The whole framework assumes the buyer has a status quo they're comfortable with.
When the *bar is zero*, when the alternative is literally "we die" or "we do this entirely by hand with 2,000 people" (Block's compliance team) or "we just don't have this capability at all"?
The chasm doesn't exist for those.
Buyers start acting like visionaries instead of skeptics, because they have to buy. The alternative doesn't exist. They'll tolerate a 60% solution, missing features, no references, because 60% of something beats 100% of nothing.
The companies I get most excited about aren't disrupting incumbents. They're filling voids. 9 Mothers in the YC Spring 2026 batch is a counter-drone defense co for whom bar is zero, there is no viable close quarters defense otherwise! There's no chasm to cross for that.
The practical implication for founders: if you're in a market where the bar is zero, stop worrying about whole product, stop worrying about crossing the chasm, stop worrying about pragmatist references. Ship the 60% solution. They're begging for it.
If you're NOT in a bar-is-zero market (if there's an incumbent, a status quo, a "good enough") then Moore applies in full and you need the whole playbook (beachhead, bowling alley, whole product, the works).
The question every founder should ask: is my customer's current alternative literally nothing? If yes, you're in a different game than the textbooks describe. Ship it in whatever form you have. You'll know. And it's a great place to be.
Pawan
Reading this thread is a good reminder: without conversations, you don’t have a startup, you have a hobby.
GPT-5.5 is a very good model
DHH: For complicated agent work, it's amazing how much GPT5.5 has improved. I found 5.2 to be very far behind Opus. Now using Opus 4.7 after 5.5 feels like a big step backwards. Gotta love this level of competion! Strong comeback for OpenAI.
Tibo
A little secret. About 5% of our production traffic is on the Pi harness, about another 5% is on OpenCode. Reminder you can use your ChatGPT account in a flourishing set of other tools.
We’ll continue to make Codex awesome, but you have options.
co-sign. a very handy mental framework for what kinds of learning transformers do well today, and why it runs into limitations. when @ankit2119 and i wrote about the need for adversarial world models earlier this year, we were describing a couple of the functions of these rungs of thinking that bring us ever closer to the kolmogorov-limit generator of reality. throwing more params, more power, more everything at a demonstrably inefficient paradigm will be outclassed by the simple solution that can hypothesize and seek truth rather than backfit a house of cards - although the bitter lesson is it is simpler to scale and we may hit agi anyway because human intelligence just isn’t that smart nor plentiful
Rishabh Agarwal: Very well written blog. I think of RL as learning from interventions, and it kinda explains why it's more powerful as a paradigm than supervised learning.
Now learning from counterfactuals is something we haven't been historically good at but maybe world modelling+ RL can get
Richard Hillier
Some great advice for the AI age.
Highly recommend reading through and then applying these to really develop.
Peter Yang: 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
Shann³
What´s gBrain and how does it work?
I've been using gStack for a while when ideating, validating new projects, and some coding
now I'm experimenting with gBrain as the memory layer for my agents, starting with my Hermes Agent company
gBrain is an open-source persistent memory layer for AI agents (by @garrytan). it turns your emails, meetings, tweets, voice memos, and docs into a typed knowledge graph. essentially markdown in, graph out.
how it works:
> 1. ingest signals from your daily life
> 2. extract entities + create typed links (works_at, invested_in, attended)
> 3. store as Markdown + Postgres + pgvector
> 4. retrieve via hybrid search (keyword + vector + graph)
> 5. agents read brain first, write insights back, graph builds itself
an overnight dream cycle dedupes entities, repairs links, and updates the compiled truth
Clash Report
Francis Fukuyama on the U.S. as a declining power:
American decline is a direct product of Trump's rise since 2016.
It is as if Trump had decided to do everything in his power to weaken the United States vis-à-vis China.
He has polarized an already polarized country, cut funding for basic scientific research, and attacked American universities which are the best in the world.
He and his colleagues have openly stated that their domestic opponents — the Democrats — are a far greater threat to the future of the United States than either China or Russia.
There is agreement among America's friends and rivals that the United States has become something of a rogue state that is contributing to global instability and disorder — as well as something of a laughingstock.
Demis Hassabis
Re @garrytan great to hear that Gemini Live is working so well for you! Awesome to see what you are doing with it! cc @joshwoodward
Garry Tan: GBrain just shipped v0.40.0 gives your OpenClaw/Hermes Agent + GBrain a voice agent.
It's based on Gemini Live. (Thanks @demishassabis it's amazing) Large context, great tool use, full brain access.
Mars is a friend, Venus is your EA.
My open source gift to you.
Everyone thinks "do things that don't scale" is about building relationships with early users.
Yes AND it's about generating mistakes at maximum density.
When you're doing everything manually (onboarding, support, delivery) you hit errors every hour. Each error teaches you something the dashboard never will.
The manual work IS the learning. Automate too early and you freeze your ignorance in code (and now markdown).
A 6-person team is building task-specific AI models that are 4-8x faster than anything from OpenAI or Anthropic. 500K downloads on HuggingFace. No hype. Just better engineering winning on the merits.
This is what "make something people want" looks like in the model layer.
https://zeroentropy.dev
Bob McGrew has a framework I keep thinking about: in the AI future there are only two jobs. The Lone Genius and the Manager.
That's it. Everything else gets absorbed.
The Lone Genius is the person sitting alone at a computer, amplified 1000x by AI. One person with taste, vision, and relentless focus who can now do what used to take a team of 50.
The Manager is the person who becomes CEO of their own "firm" where most of the employees are AI agents. They define the goals. They decide what matters. They coordinate. The AI does the execution.
The Marxists will hear "two jobs" and panic. "What about everyone else?!" But here's what they're missing: AI doesn't shrink these two categories. It explodes them open. More people get to be geniuses. More people get to be managers. The barrier to entry for both just collapsed.
What actually gets eliminated? David Graeber called them "bullshit jobs." Graeber was no libertarian! He inspired Occupy Wall Street.
His words: "Huge swaths of people spend their entire working lives performing tasks they secretly believe don't really need to be performed. The moral and spiritual damage that comes from this situation is profound. It is a scar across our collective soul."
Graeber said bullshit jobs are "a form of spiritual violence directed at the essence of what it means to be a human being." They induce "hopelessness, depression, and self-loathing."
This is who the left should be fighting for. Not to preserve those jobs. To liberate people from them and give them better ones.
The dirty secret of the modern economy: millions of people sit in roles so pointless that even they can't justify their existence. Compliance layers. Reporting layers. Coordination layers. Meeting-about-the-meeting layers. They know it's meaningless. It eats them alive.
AI eats those layers. Good. That's a jailbreak.
What I love about Bob's framework is where it points. The Lone Genius used to require a PhD, a lab, institutional backing. Now a 19-year-old with taste and Codex can ship what took a research team a year. The genius bottleneck was never talent. It was access.
The Manager used to mean you needed to hire 50 people, raise money, build an org chart. Now you can orchestrate a fleet of AI agents from your laptop. The management bottleneck was never skill. It was capital.
AI doesn't concentrate genius and management into fewer hands. It distributes them into more hands. The working class kid in West Virginia. The single mom in Ohio. The 55-year-old who got laid off and now builds software for the first time. Those are some of Bob's future geniuses and managers.
The best founders I see at YC are already living this. They toggle between both modes in the same day. Morning: lone genius, creative insight, the thing nobody else sees. Afternoon: manager, spinning up agents, steering, shipping.
The cycle time between genius and manager IS the new productivity metric.
So when someone tells you AI means "only two jobs and everyone else starves," quote Graeber to them, they’ll get it.
Graeber knew the real violence was making people do meaningless work and pretending it was dignity. AI ends that. More genius. More agency. Fewer spiritual prisons.
Dr. Julie Gurner
Re @garrytan What this really emphasizes that perhaps is not as clearly said, is that "tasks" are no longer the most substantial element in work. Cognitive direction, clarity, taste, creativity, etc. become more of the work.
What defines work pulls into a different category.
Ghita
Thank you @garrytan 🙏
None of this happens without @ycombinator believing in us early
Insanely proud of this team!
6 people shipping like 60 @ZeroEntropy_AI
Garry Tan: A 6-person team is building task-specific AI models that are 4-8x faster than anything from OpenAI or Anthropic. 500K downloads on HuggingFace. No hype. Just better engineering winning on the merits.
This is what "make something people want" looks like in the model layer.
Quick Thoughts
This is misleading. The issue affects 4 families, there is other construction in the area, and it has nothing to do with data centers.
Headquarters: Rep. @AOC: I have a jar right here. This is the current drinking water in Morgan County, Georgia, right after a Meta data center was constructed. The only difference between the clean water and this was that data center. This is what the drinking water now looks like next to that
Melanie D'Arrigo
“If we erase January 6th, it didn’t happen”
“If we don’t release certain inflation data, there’s no inflation”
“If we don’t measure food insecurity, no one’s hungry”
“If we delete research on right-wing violence, there’s no right-wing violence”
“If we stop Covid testing, we’ll have fewer cases”
The Trump administration wants you to believe their propaganda is fact, and control the majority of mainstream media and social platforms to reinforce their lies.
This is what authoritarianism looks like.
ABC News: The Department of Justice is acknowledging it has removed from its website news releases about criminal cases related to the Jan. 6, 2021, riot, calling the information about the prosecutions “partisan propaganda.” https://abcnews.link/A8W2vOT
Michael
ngl Latent Space is quietly the best podcast in the tech space right now by far.
The quality of conversation is insane. I just learned:
-Turbopuffer CEO ran the co off his personal credit cards because he messed up on the napkin math and had negative margins with Cursor
-Turbopuffer ceo LITERALLY bought a fiberoptic line from
-Will Bryk didn't learn to drive until 23 because he worked at SpaceX/Zoox and believed Elon's self-driving car timeline
-People are genuinely using Exa to find partners
-Companies are basically using Modal as a second backend
good location to build
Thomas Ricouard: Codex & Pool
Morgan J. Freeman
And just like that, it’s completely VANISHED from the media.
A sitting congressman, Ted Lieu, said on the record the Epstein files are being blocked because they show Trump raped and threatened to kill children.
Lets make this viral again 👇
Codex for building and debugging an iPhone simulator end to end:
Justin: Codex computer use entirely driving iphone simulator to bug bash a feature it just built
Noah Smith 🐇🇺🇸🇺🇦🇹🇼
If Trump kicking the Tech Right in the nuts doesn't put an end to the Tech Right, I don't know what would
Beff (e/acc): Feeling robbed of my path to citizenship right now after grinding a PhD and contributing to foundational AI + computing technologies for the United States for the past ~ 10 years.
Feels like robbing top and technologists like me of the opportunity to achieve the American Dream.
David Senra
This conversation with @RickRubin is special. Here's a 2 minute preview that I’ve watched 15 times.
The episode will be out tomorrow:
gm 🏃♂️
Show me the thing you’ve built with AI you’re most proud of. Reply with a working product URL and what model / agent you primarily used.
Jared Friedman
I did the math, and $2M in OpenAI tokens is equivalent to a 450 pre-AI human engineering team for 2 years, in terms of raw LOC output.
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
Replit ⠕
Replit Agent builds your app. Squidler tests it like a real user. Replit Agent fixes what's broken.
That's the full AI QA loop, and it's now live in Replit's MCP library. You describe what your app should do in plain English. Squidler navigates it the way a real person would. Issues flow back automatically and get fixed. No test-writing skills required.
Build with Replit. Test with Squidler. Ship with confidence.
Squidler: Official today: Squidler is in @Replit MCP library. Build with #Replit. Test with Squidler. Replit Agent builds, Squidler runs real user flows against the live URL, results loop back to the agent. No selectors, no scripts. https://replit.com/partners/squidler
Rodney Brooks
He has directly betrayed all his followers who trusted him, all of the rest of us who never trusted him, and his country.
Bojan Tunguz
Here is one big reason why this matters. Time spent on non-LLM inference tasks is only going to increase. However, tools that these AI system use are *very* inefficient and have been built from the ground up for CPU and human use. There is a huge untapped opportunity there to significantly improve those processes with AI agents in mind from the ground up.
SemiAnalysis: FACT ALERT 🚨 : In modern agentic coding, 42% of the time is spent on CPU doing tool use such as editing files, running Bash scripts, running lints, etc. The economy of traditional cloud computing charges at $ per cpu core. In the economy of agents, the business model is $ per
Alec Stapp
Nearly half of the founders of billion-dollar tech startups are immigrants
Brad Gessler
Zap the rocks and tighten the loops.
Erik Brynjolfsson
The Journal of the @americanacad just published a new issue of Daedalus on AI and Science, edited by James Manyika.
It has terrific line-up of contributors, including @demishassabis, @ylecun, Josh Tenenbaum, @AnimaAnandkumar, @EricTopol, @alondra and many others.
It’s going to be epic!
Replit ⠕: Oscar-winning filmmaker behind Her, Being John Malkovich, and iconic music videos for Beastie Boys and Björk, Spike Jonze is a boundary dissolving storyteller making machines feel human and humans feel surreal.
See him take the stage with @amasad on day one of Vibecon.
NYC,
Erik Brynjolfsson
Historically, the US has been the greatest magnet for talent on earth.
Whether that continues to be the case is not inevitable.
Alec Stapp: Nearly half of the founders of billion-dollar tech startups are immigrants
AI Engineer
Most agents die after a few seconds. @AnthropicAI's workshop shows how to build agents that run for hours. full 75-min session with Ash Prabaker & Andrew Wilson.
https://www.youtube.com/watch?v=mR-WAvEPRwE
Finn Mallery
talked to a YC company that scaled from $0 → $2m ARR in their first 6 months with their ENTIRE GTM built off going to conferences.
Here's the playbook they cracked (step by step):
~4 weeks before:
> Post abt the conference and tell attendees exactly how to reach you
> Send personal DMs to the right ppl on LinkedIn and X
> Reply within the hour & lock in 10 top targets to close.
> Send everyone else to your drip email campaign.
Then, set a meeting block of 1-3 days during the conference:
> make shared booking link for the team
> Reserve a quiet café / private dining room
> Pack in 12 meetings per day, 30 min each, with buffer time built in
While you're there:
>Hand every prospect a thoughtful small gift and a personal card
>Single out 5 standout customers whose pain ur product actually solves
>Pull them aside for a casual on-camera Q&A in a solid film spot
>Don't pitch hard.
>Let the conversation breathe and weave your product in naturally.
The 4 weeks after
>Hand the raw footage to a freelance editor + ask for ~15-20 punchy clips with captions.
>Drop a new clip every couple of days on LI / X
> use these clips when you post online about the next conference to keep the momentum
This is the formula, costs less than a few thousand dollars to execute.
They’re on track to end the Y1 at ~$6m ARR (B2B, targeting large enterprises) + STILL not using any other channels for customer acquisition
Matthew Yglesias
I feel like if people move here from abroad and get high-paid jobs and pay taxes that's good?
Mickey Kaus: "All told, a remarkable two-thirds of the Valley’s nearly 400,000 tech jobs are now held by those born abroad, according to a 2025 report from the think tank Joint Venture Silicon Valley." https://www.realclearinvestigations.com/articles/2026/05/20/the_replacements_how_us_policy_helps_foreign_workers_take_american_jobs_1183978.html
