The asset seizure tax drafters say a lot of things but they’re all basically uninformed lies by academics who have no experience in business reality
That they keep being quoted by gullible reporters is a miscarriage of journalism
Jared Walczak: The California wealth tax could be levied on founders' voting interests in their companies, which often far exceed their economic stakes.
The initiative's drafters say that won't happen. Here's why it might:
https://taxfoundation.org/blog/california-wealth-tax-voting/
Amazing to see almost 2000 people at ClawCon Michigan!!
Weird lobster cult 🦞🥳
ClawCon: ClawCon Michigan - Live #ClawConMichigan https://x.com/i/broadcasts/1oKMvRLVRqkGQ
AI Engineer
🆕 Harness Engineering: How to Build Software When Humans Steer, Agents Execute
https://www.youtube.com/watch?v=am_oeAoUhew
@_lopopolo is one of the emerging class of token billionaires at @OpenAI, and is unapologetically exploring how you can get to 5 -> 50 -> 5000 agents working for you 24/7. Special Q&A moderated by @vibhuuuus.
I was trying out my voice agent re smart scheduling and the STARLUX flight attendant in Taipei immediately asked what is that and can I have it yet?
She is a fan of the movie Her.
The future becomes obvious with voice. We are so close to what we saw in the movies.
I will say the hard left contingent of SF is going to downplay violence as simple vandalism and if extremist Chesa Boudin were still DA there likely wouldn’t even be serious charges being pursued
💥Susan Dyer Reynolds🗞️: The defense attorney says throwing a Molotov cocktail at Altman’s house is a case of “simple vandalism.” 🤦🏻♀️ @BrookeJenkinsSF points out there was a manifesto, and he said at the OpenAi HQ he wanted to burn down the building. This terrorist drove all the way from Texas. Don’t give
Jeremy Howard
Wow I can already say after just 5 hours using @AnthropicAI Opus 4.7 that this is the first model that "gets" what I'm doing when I'm working. It feels aligned with me in a way no previous model did.
(4.6 actively worked against me. I hated it. So this is *very* exciting!)
Hi, I might give it away as open source (it's called GBrain)
https://github.com/garrytan/gbrain
Alex Lieberman: Someone is going to build a worldclass “Brain” for enterprises & make a stupid amount of money.
Why? As @da_fant said, “coding w ai is solved bc all context is in the git repo. knowledge work is difficult bc context is spread out. an ai system that creates a git repo w all
Ryan Lopopolo
Hey that’s me!
AI Engineer: 🆕 Harness Engineering: How to Build Software When Humans Steer, Agents Execute
https://www.youtube.com/watch?v=am_oeAoUhew
@_lopopolo is one of the emerging class of token billionaires at @OpenAI, and is unapologetically exploring how you can get to 5 -> 50 -> 5000 agents working for you 24/7.
Anthropic or some other AI native tool will likely disrupt design tools and Google Workspace / Microsoft Office.
Most of these tools were designed for humans to hand craft slides, docs, and spreadsheets one by one.
Getting an agent who understands you to generate these artifacts via some brain dump is much faster.
Although the last 10% should still be hand-crafted based on my experience.
You can turn your web app into an iOS app for less than $10
Rick Delashmit: This is just wild. @Replit Agent 4 worked for over an hour completely autonomously, refactored my web app into a native React iOS app, then tested the app, ran 69 tests, and optimized the code … for $7 dollars.
We need a return to judges who uphold the law instead of exercise their personal virtue signal hobby horses from the bench
Vote Phoebe Maffei For SF Superior Court Judge
Erica Sandberg 舊金山的神奇女俠: Cops arrest the bad guys; judges decide what happens to them. assistant DA Phoebe Maffei is running to become SF's next Superior Court Judge. Her opponent is from the public defenders office. read up! https://thevoicesf.org/phoebe-maffei-races-to-court/
Why do any real work when you can just migrate from openclaw -> hermes -> perplexity computer -> openclaw again
You hate AI slop but do you hate corpo-media slop enough?
Ana Mostarac: The meta for journalism needs to shift away from writing hysterical, incendiary, alarmist, engagement optimized slop.
tmctmt
Spying on everybody's Discord attachments with HTTP desync
https://tmctmt.com/posts/http-desync-in-discord/
imagegen in codex is easy to underestimate, but it's quite powerful:
Won Park: Image generation is now live in Codex!
You can now generate visuals, edit existing images, and create GIFs from a single image directly inside Codex.
I spent a lot of time testing different use cases while working on this feature, and it was genuinely impressive to see how
When I build skills now I always ask AI to spin up a separate eval agent to do yes/no checks to grade the first agent's output.
And if the output isn't "yes" across the board it asks the first agent to keep working.
Building one for YouTube thumbnails and titles now.
it's the little details
Ed: this flow
My claw and I searched high and low for proper e2e Gemini Live tests and in the end we decided to do it ourselves
Coming to GBrain Voice, open source release soon.
Claude and Codex apps should build an easy way to view and edit markdown files from web and mobile if I want to hand craft the last 10% of instructions
Kane 謝凱堯
Today I learned you can murder a toddler in San Francisco and @sfgov still won’t keep you detained so you can just skip court.
KRON4 News: A San Francisco mother and father failed to show up in court to face murder charges today for the death of their 2-year-old daughter.
The toddler was fatally poisoned by fentanyl inside a home riddled with drugs, prosecutors said, and several hours passed before anyone called
Re Lots of GBrain security fixes just dropped
Now sending out orange tickets
eve: we started sending invites to @ycombinator Startup School 2026 and i thought it'd be fun to give people a ticket when they get accepted
The future is MCP/CLI
Marc Benioff: Welcome Salesforce Headless 360: No Browser Required! Our API is the UI. Entire Salesforce & Agentforce & Slack platforms are now exposed as APIs, MCP, & CLI. All AI agents can access data, workflows, and tasks directly in Slack, Voice, or anywhere else with Salesforce Headless
Daniel
1/
Old Nvidia chips are getting MORE valuable with age, not less.
That has never happened in computing history.
Marc Andreessen (@pmarca) on @latentspacepod with Alessio Fanelli (@FanaHOVA) and @swyx
https://www.youtube.com/watch?v=knx2wrILP1M
George Conway ⚖️🇺🇸
"I don't think people are sufficiently prepared notwithstanding what happened in 2021, for the possibility that he will try to fuck with this election. And he will. He's already basically telling us that's what he's going to do, just the same way he told us he would do that in 2020."
Jared Friedman
Tomorrow. YC Startup School India.
Brian Krassenstein
Trump is reportedly negotiating a deal which would stop Iran from producing a nuclear weapon in exchange for $20 Billion in Iranian assets being unfrozen.
Note that The Iranian Nuclear agreement (JCPOA) that Obama signed, that Trump tore up, did the same thing. It halted their Uranium enrichment while turning over frozen assets.
In other words Trump Tore up the agreement, spent $55 Billion on a war, got hundreds of Americans injured, killed 150+ kids and exploded oil prices to try and negotiate the same basic deal that Obama did.
"If you're a PM and you think your job is to make documents and slide decks for upwards review and alignment, you're going to love this new world because you get to make things too.
People need to see that leaders in their company are making things. That's what actually inspires and creates the inflection."
- @zoink
📌 Watch our full interview here: https://youtu.be/eqPljh_9C9Y
Peter Yang: "AI gets you to average quickly. Your job is to push past that."
Here's my new episode with @zoink (Figma CEO) where I asked him some tough questions, including:
→ Can you teach AI design taste?
→ Do design systems hurt creativity?
→ What's Figma's role when code is free?
Peter Yang
"If you're a PM and you think your job is to make documents and slide decks for upwards review and alignment, you're going to love this new world because you get to make things too.
People need to see that leaders in their company are making things. That's what actually inspires and creates the inflection."
- @zoink
📌 Watch our full interview here: https://youtu.be/eqPljh_9C9Y
Peter Yang: "AI gets you to average quickly. Your job is to push past that."
Here's my new episode with @zoink (Figma CEO) where I asked him some tough questions, including:
→ Can you teach AI design taste?
→ Do design systems hurt creativity?
→ What's Figma's role when code is free?
Daniel Jeffries
I don't understand why we automatically give people credit for "sincere views."
Like who gives a fuck?
If I have a sincere view that a transdimensional vampire attack is imminent it doesn't make it sane just because I'm sincere in my fucking delusion.
The guy who tossed a firebomb at Sam Altman's house is probably sincere in his delusion too. And he is trying to solve a fake future problem by creating a real problem here and now in the real fucking world.
Lots of folks are "sincere" in their delusions. Charley Manson was sincere. So where the Marx and Goebbels and everyone else who was wrong in history. It's delusion and Maya all around. History of the human race.
Sincerity means exactly zero in judging the rightness of someone's claim or prediction.
And also we give way too much credit for "someone works in field X so they must be able to predict the future about X."
Why?
If I'm a bridge engineer what the fuck do I know about the future of how bridges will effect society?
Answer, I don't. I fix fucking bridges.
The fact that I'm predicting something about bridges is irrelevant because the skills do not overlap.
I may have the skill to predict the future well but that skill is completely and totally independent of whether I work in the field.
The truth is most of us can't predict shit.
We are generally terrible at long term predictions.
If you're living in Germany in 1440 and making predictions about the future of German society you're going to be wrong because you can't predict the printing press and how it alters the structure and trajectory of society. If you could have predicted the printing press you'd be Gutenberg and you would have invented the printing press.
If you're Paul Ehrlich you're a failure because you're not actually smart, you're a smug, smart sounding asshole who couldn't predict the green revolution.
We give way too much credit by default in way too many ways to people and their ability to predict anything accurately.
A few people are good at it and most suck at it.
Matthew Yglesias: “Our product will generate mass unemployment and possible mass extinction” isn’t a bad messaging choice, it’s a wildly held sincere view among the people building AI.
https://www.slowboring.com/p/its-not-bad-marketing-from-ai-companies
china discount is real, damn
A lot of bugs that folks may have hit yesterday when first trying Opus 4.7 are now fixed. Thanks for bearing with us🙏
Ethan Mollick: I'll give Anthropic credit for moving quickly. Opus 4.7 Adaptive Thinking now triggers thinking much more often, including for the tasks it failed at yesterday. That also means it is doing a lot more web search.
So far, a large improvement in output quality on non-coding tasks.
Thanks for trying GStack and for your honest review 🫡
GBrain is designed for you to take what you want of my OpenClaw setup and it’s all open source & optional what you implement. GBrain tries to figure out what you have and only give you what would be helpful
No Supabase instance if you have less than 1000 md files for instance
VibeTokens: @garrytan Compiled truth + timeline. Signal detection. Tier system. Back-linking.
Four things built, three repos pushed, one morning.
Full build log: http://vibetokens.io/blog/i-stole-garry-tans-brain-architecture
Anthropic has no chill 🥲
Claude: Introducing Claude Design by Anthropic Labs: make prototypes, slides, and one-pagers by talking to Claude.
Powered by Claude Opus 4.7, our most capable vision model. Available in research preview on the Pro, Max, Team, and Enterprise plans, rolling out throughout the day.
Everyone with a vision can produce very high-quality designs now (with a lil help from Claude)
Claude: Introducing Claude Design by Anthropic Labs: make prototypes, slides, and one-pagers by talking to Claude.
Powered by Claude Opus 4.7, our most capable vision model. Available in research preview on the Pro, Max, Team, and Enterprise plans, rolling out throughout the day.
same
Patrick Collison: I'm lucky enough to have a great doctor and access to excellent Bay Area medical care. I've taken lots of standard screening tests over the years and have tried lots of "health tech" devices and tools.
With all this said, by far the most useful preventative medical advice that
Jack Cheng
My new one for @every on “living software” vs tool-like software: https://every.to/p/living-software
AI Engineer
In @steipete's latest State of the Claw, he gives an update on 5 months of @OpenClaw and some behind the scenes on what it's like maintaining the fastest growing open source of all time:
https://www.youtube.com/watch?v=zgNvts_2TUE
eg:
- 60x more security reports than curl
- a "Bullshit Taxonomy" of illegitimate reports
- Nation State attacks
- 12%-20% of skills contributions malicious
- contributors burning multiple Codex Pro per day
- academic FUD
Agents are both the product AND the attack vector. @simonw's Lethal Trifecta is not solved. Come for Pete's recommendations, what OpenClaw is doing on security, OpenClaw Foundation roadmap, and then subsequence audience Q&A with @swyx on taste, dreaming, and OpenAI.
Made this 30 second video of Claude Design just by pasting in the Claude Design blog post and some tweets from @AnthropicAI employees
Kinda speechless.
sarah li
just dropped my first video in a 5 part series where we build a full travel planner app on Replit and i am so excited 🌍✈️
as a software engineer, design has always felt a little out of reach for me. like i had all these ideas in my head but turning them into something that actually looks good was a whole other skill set. but the Replit Canvas genuinely changed that for me.
my favourite way to use the Canvas is to bring in reference images and websites to help me to create designs!
in this first video we jump into the Canvas, explore design directions for our travel app, and apply a final look to a real working app. it is so fun i promise.
go watch it and let me know what you think 🎨👇
grateful to @steipete for prompting me to start the AMA with “what’s the future of ClosedClaw”? 😈
AI Engineer: In @steipete's latest State of the Claw, he gives an update on 5 months of @OpenClaw and some behind the scenes on what it's like maintaining the fastest growing open source of all time:
https://www.youtube.com/watch?v=zgNvts_2TUE
eg:
- 60x more security reports than curl
- a "Bullshit Taxonomy"
Amit Bhatia
if you missed @every compound engineering camp, here's some gems to help you with becoming a better builder:
the AI sandwich: you think at the start and end. agents do the middle.
ideate before code, polish after the PR. save plans in the repo, so they can be used as context for agents and postmortems later.
creates a product workflow where the review loop and the captured learning are as important as the generation.
that's the part I want to build better habits around.
big thanks to @kieranklaassen, @trevin, @danshipper for creating compound engineering, contributing, and sharing to help builders.
everyone who joins @every reads + discussed the last chapter of The Writing Life
we’re hiring a lot, so…
Sheel Mohnot
A down-ballot race that matters!
TL;DR - I did my homework and urge you to vote Patrick Wolff for Insurance Commissioner. He's the only one who has any clue what he's doing.
California's insurance commissioner race doesn't get much attention, but it's actually one of the most important races on your June 2 ballot.
It took me over a month to buy a car because insurers had paused new auto policies in the state and I needed insurance before being able to sign the lease papers. Many homeowners can't get coverage at all right now, and a lot of homes in the state are losing value because they can't get insured! We've had a series of fires over the last many years and folks who lost everything got screwed by the companies they'd been paying for decades.
TBH It's crazy that we vote for insurance commissioner at all. This should clearly be an appointed role. Most states appoint someone with actual expertise. But we live in California and have this silly form of direct democracy. For this position, we've ended up with a conveyor belt of bad or termed-out politicians with zero relevant background trying to get into political office somewhere.
That's how we got Ricardo Lara, who ran for the job on a platform of (I kid you not!) - being openly gay and "standing up to fight our bullying President, Donald Trump." This is the insurance commissioner. We elected a journalism major who had never worked anywhere near insurance because he's gay and promised to fight Donald Trump. WTF!
The job is rate filings and claims oversight. Lara fleeced us. He took money from companies he regulated, and gave them preferable access, and spent tons of taxpayer money traveling around the world, while at the same time being too busy to accept rate filings from insurers who then pulled out of the state.
Rate filings supposed to take 60 days routinely took over a year. Auto insurance rate approvals were frozen for two years. Seven major carriers paused or stopped writing policies in CA. The FAIR Plan doubled. Meanwhile, we've been in crisis mode with some of the worst crises in history during his term...
The field to replace him is mostly more of the same. Has-beens who have no insurance background. Jane Kim wants to insurance run by the state, and is endorsed by Bernie Sanders. We have a bunch of other career politicians in the race, hoping to become insurance commissioner because they failed at whatever else they were actually aspiring for.
I'm voting for Patrick Wolff. He's a CFA who built an insurance brokerage, analyzed insurers for 20 years, and passed the property and casualty license exam during his campaign, the first candidate ever to do that. He's self-funded his campaign and won't run for another office.
His plan is straightforward: allow real competition so no single insurer can gain enough market power to blackmail the regulator on pricing. Require subsidiaries operating in CA to have financial backstops from their parent companies. Publish claims performance report cards for every insurer so consumers can reward good actors and punish bad ones. Streamline rate filings so they actually get processed.
The crisis in our state is the result of bad regulation... Good regulation can fix it.
Vote Patrick Wolff, and enjoy this awesome video!
Patrick Wolff:
I’ve been really enjoying this feature. It’s specially good at anticipating minor but important improvements you can make to your app.
Replit ⠕: Replit Agent got even better at keeping you in your creative flow!
It now suggests follow-up tasks using full context of your project to build on your ideas:
• new features to build
• performance improvements
• user experience enhancements
Review the plan, accept what you
Animate your way to virality.
Samuel Spitz: I've gotten more than 10M organic impressions with vibecoded videos.
How? Replit Animation.
Today, me and the team share how you can do this too
🧵
codex is for everyone. learn how to get the most out of it:
Derrick Choi: As you may have seen, we announced some major updates to the Codex app yesterday.
If Codex still feels intimidating, or like something you think only engineers use, this session is for you.
I walk through how Codex can help with everyday work:
- organizing files
- combining
Liu Liu
You cannot believe how much intrinsic knowledge GPT-5.4 has about building GPU compute kernels. I honestly think this will materially impact NVIDIA's lead in software.
Jason ✨👾SaaStr.Ai✨ Lemkin
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 SaaStr AI Annual 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.
SaaStr AI Annual 2026. May 12-14 in SF Bay!!
Codex is open source, enabling anyone to build awesome applications on top of it:
Developing Adventures: Want to use Codex computer control from your phone?
OpenAssist got you covered.
Let the agent work… you go touch grass 🌱😂
Built on codex app server thanks for making it open source.
@OpenAIDevs @thsottiaux
FWIW i have just updated chrome and do not see this.
google has this big issue of ultra slow incremental rollouts. it really kills the vibe - i see stuff they launch, i am excited to try out, "oh its still rolling out". 10 days later its still rolling out.
i'm going back to my @kagihq Orion.
Google: Too many @GoogleChrome tabs open? Try vertical tabs, rolling out now.
Just right-click any Chrome window and select “Show Tabs Vertically” to move your tabs to the side of the browser window, making it easier to read page titles and manage tab groups.
Muzzammil Zaveri (MZ)
missed a few crucial ones:
• Ben Silberman - Act I and Act II - AudioBeta (W06) and MightyQuiz (W08) // Act III - Pinterest
• Anton Osika - Act I - Depict (YC S20) // Act II Lovable - Valued at $6.6B
• Karri Saarinen - Act I - Kippt (YC S12) // Act II Linear - Valued at $1.25B
Muzzammil Zaveri (MZ): Repeat @ycombinator founders hit different. Early success + the YC learnings = massively higher odds of building a category-defining company. Eg:
1. Sam Altman
• Act I: Loopt (YC S05) — location-based social networking app. Sold to Green Dot for $43.4M
• Act II: OpenAI —
Claude for Word is now available on Pro and Max plans to use alongside Opus 4.7: https://claude.com/claude-for-word
codex for proactively suggesting what it can do for you:
Anthony Kroeger: every couple of hours i just get mind blown by the Codex app, it's actually insane
i opened a new chat and noticed that there was a list of suggested tasks
this might sound normal, but i hovered one and it was suggested based off some bug reports in Slack threads that it read
Matt Ronge
http://x.com/i/article/2045233659612172288
Kieran Klaassen
Apparently @trevin and I created an AI sandwich?
Introducing: the AI sandwich... lol
Kieran Klaassen: The mistake isn't automating too much; it's not knowing *when* to think.
Building Cora, I found two moments where humans belong in the loop: **Brainstorm** (what to build) and **Polish** (is it actually good?). Everything else – plan, code, review, test, PR – is automated.
This is what founders are for
Would YC be funding 800+ brand new sets of almost all first time founders if that weren’t the case?
The people on the other side of this bet are the commercialization folks at the frontier labs, awkwardly!
Adel Bucetta: @garrytan you're missing the point: even with those tools, someone still has to own the judgment of what to build
here is our full Opus 4.7 vibe check
we broke down how it performs on coding, writing, spreadsheets and more: https://every.to/vibe-check/opus-4-7
Shayan
Opus is now way more expensive, much slower, and its only remaining edge over GPT models are design and slightly better prose.
I'm using GPT-5.4 for almost everything now.
Noah Smith 🐇🇺🇸🇺🇦🇹🇼
I’ve sat here for years and watched the Republicans embrace their worst extremists. I don’t want to see the Democrats do the same.
https://www.noahpinion.blog/p/hasan-piker-is-bad-for-the-democrats
Mike Taylor
PowerPoint king
Dan Shipper 📧: here is our full Opus 4.7 vibe check
we broke down how it performs on coding, writing, spreadsheets and more: https://every.to/vibe-check/opus-4-7
Every @claudeai model has a different idea of how much to help.
@alexalbert__ from Anthropic explains why—and what to expect from Opus 4.7.
Re watch the full Opus 4.7 Vibe Check: https://www.youtube.com/watch?v=W--hvgRLmJM&t=13s
SFMTA promised 465 affordable units to the Mission. Community members spent 8 years planning them. Then the agency admitted the housing was never funded—and cut 365 units. No accountability. No consequences.
That's not a bug. It's a feature.
https://gli.st/wsq0vkdx
The Claude Code hackathon is back for Opus 4.7.
Join builders from around the world for a week with the Claude Code team in the room, with a prize pool of $100K in API credits.
Apply by Sunday: https://cerebralvalley.ai/e/built-with-4-7-hackathon
Sometimes when people critique AI it really is a skill issue
LegalAI: @garrytan This dude's critique is so, "Tesla's are crap cars, I couldn't get mine to make toast."
Deva Hazarika
One of the worst communication gaps is between people who care about precision in language and those who don’t
Spotted in NYC.
I need 10 of these shirts.
Trung Phan
SCOOP: Leaked photo of the first AI compute data centre that Allbirds built with its recent $50m investment. Insiders say that the rack has enough capacity to provide tokens for 7 AI chatbot users.
Katie Parrott
Anthropic models seem to be a pendulum that's swinging back and forth between creativity and literalness. Opus 4.7 is a swing in the literal direction.
Full @every Vibe Check: https://every.to/vibe-check/opus-4-7
Vancouver, it’s been a blast! 🇨🇦
TED Talks: “The lobster is loose, and it’s not going back into the tank,” says @openclaw founder @steipete. In this brand new talk from #TED2026 he shares why AI agents — built by you — are the future: http://t.ted.com/DPASxmF
Xbotter
In most cases, indexing or routing is the best way to improve data retrieval.
This same principle is now proving true in AI agents.
Garry Tan’s Resolvers show exactly why: a lightweight routing layer beats cramming everything into prompts. Smart retrieval > brute-force memory.
Garry Tan: http://x.com/i/article/2017109679416233987
水滴石穿
Must watch. OpenClaw is a revelation. https://www.ted.com/talks/peter_steinberger_how_i_created_openclaw_the_breakthrough_ai_agent
Stargate is a step towards meeting the demand of the compute-powered economy
Epoch AI: In 2025, OpenAI announced Stargate, a $500 billion data center initiative. We surveyed all 7 US sites and found visible development at each.
There's a long road ahead, but the project appears on track to reach 9+ GW by 2029—comparable to New York City's peak power demand. 🧵
Parker Conrad
Rory's takes are always very sharp.
Rory O'Driscoll: The weird thing right now is the public markets don't have access to the growth side of software. Right now the trade is to sell SaaS and buy semis (the raw material of AI). What you don’t have yet in the public markets are the AI native software companies and therefore, you’re
The ultimate "they don't know" alpha leak here 👇
Winston B.: @garrytan Yes, and I think the skill gap is way larger than most people admit. For heavy users, we're finding that we can encode entire workflows to run night and day on plain english via markdown. Engineering context = engineering code in today's era.
OpenClaw is straight magic
Anish Acharya: openclaw wins on ergonomics. texting with models via imessage/whatsapp/telegram subtly sets expectations that the reply could take 5 minutes or 2 hours - the same expectation we have of humans
meanwhile we expect apps and websites to be instant, which is why chatGPT has to make
