agents make for a surprisingly great product
You know what would be a good AI automation:
When I receive those 10 page weekly newsletters from my kid's school I want AI to tell me if there's early dismissal or anything I should pay attention to.
Shaun Maguire
I know this sounds insane
I read "Extraordinary Popular Delusions: And the Madness of Crowds" as a teenager
I thought AI would likely have a bubble phase 18 months ago
But now I think this is 1000x bigger than the Industrial Evolution
This is an evolution of our species
Geiger Capital: Have you considered the possibility that it’s not a bubble and the world is indeed changing at a pace humanity has never seen before, anon.
The highest and most important form of design is actually pure transmutation of human pain and suffering.
Stephen
My talk at @aiDotEngineer is now online.
I talked about our research and where @bfl_ml is heading.
Thanks @swyx for the invite
https://youtu.be/x8Yb4RidLgM?is=ad8jtmUrL5boUbTU
/goal is underrated
james yu: We have a gnarly refactor in our codebase that I test every frontier model on. I've been doing this since the release of o1. Nothing has been able to solve it.
GPT-5.5 extra high thinking with /goal just did it.
I am floored.
Rodney Brooks
What will we be like when he is gone? Can we return to mutual respect? Can we believe we are all on the same team as Obama and McCain did? Can we imagine the mutual respect of those two, competitors but compatriots? See this at 12:11 for what our country should be as stated by McCain and reported by Obama. https://www.youtube.com/watch?v=eLIlOQuzTnU
Alex Lupsasca
I recently joined @latentspacepod to talk about AI for physics.
We dug into recent work on scattering amplitudes with GPT, and what it suggests about how AI will accelerate theoretical discovery in a rapidly evolving field.
Latent.Space: 🔬Doing Vibe Physics
The full story of how GPT‑5.x derived new results in theoretical physics and quantum gravity, live on our Science pod today!
https://latent.space/p/lupsasca
our conversation with @ALupsasca, an award winning theoretical physicist on his AGI-pilling journey
Don’t ride someone else’s horseless carriage
Time to build your own Ferrari
Personal AI is here
阿绎 AYi: 我一直觉得，现在 99% 的人使用 AI 的方式都是错的。
他们还在和 ChatGPT 聊天，
还在纠结哪个模型更好，
还在网上抄各种“神奇 prompt”。
就像汽车刚发明的时候，所有人还在研究怎么把马养得更壮、跑得更快。
而 Garry Tan 已经造好了属于自己的跑车。
我们正站在一个历史性的转折点：
从 Prompt
My viral meta-meta-prompting post about personal AI is directly a consequence of the thinking about and building with metaprompting in January and February of this year 👇
Garry Tan: The one skill everyone who uses AI needs to master in 2026
Meta-prompting: using an LLM to generate, refine, and improve the very prompts you use to get work done.
https://garryslist.org/posts/metaprompting-is-a-skill-everyone-who-uses-ai-needs-to-master-in-2026
Nadia I
Part of why I love building with @Replit is because the fingerprints of a mother builder are all over it. And as a mother myself, I can tell.
@HayaOdeh, the co-founder and “mama” of @Replit, said this perfectly:
“You did not become less capable when you became a mother. You became more capable in ways that don't always get named.”
https://blog.replit.com/mothers-who-build
#WomenInTech #MothersWhoBuild #MomFounder #BuildInPublic #EdTech #Replit
Sherry Jiang
“we gonna yolo our way into running the biggest conf in town.” and somehow… we actually did it with @aiDotEngineer singapore
a year ago, @agrimsingh @unprofeshme and i joked about the idea of running our own conference (after finishing an escape room).
the idea started from frustration -
that there wasn't a quality conference built for builders, by builders in asia.
that there's so much untapped talent here yet ppl think the "scene is dead."
it’s not dead. there’s an absurd amount of untapped talent here. the right people just weren’t in the same room.
and instead of waiting around for someone else to step in, we decided to do it ourselves.
5 ppl, all part-time, running on agents and fumes.
and this week, somehow, that turned into minister vivian balakrishnan keynoting alongside many of the best ai companies and builders in the world.
thank you @swyx for trusting us to bring @aiDotEngineer to asia.
and thank you to our team, volunteers, speakers, sponsors, and everyone who believed in this before it was obvious.
let’s make a little history this week.
Erica Sandberg 舊金山的神奇女俠
I don't sugarcoat SF for international press: “I used to be enormously proud of my city, but now I sometimes turn away from people on the tram [cable car] because I’m so afraid they’ll be disappointed.”
The American freelance journalist has lived in San Francisco almost her entire life. She calls the city “my first love.”
“Now it’s like watching your child get cancer, and you want so desperately to help.”
The Mission is sick, so I gave these wonderful Norwegian reporters a tour of health. the amazing restaurants of Hayes Vally ("that's Absinthe, Chez Mama, A Mano!"), then Nob Hill (dreamy views, clean, charming), and made sure they drove down Lombard to visit North Beach.
https://www.vg.no/sport/i/Pdvjj0/san-francisco-vm-byen-som-doer
the inside story of the legendary Cog House.
i believe there have not been any public photos of this place until now (bc we were explicitly not allowed to lol)
as an advisor its been awe inspiring to see this company grow into a well oiled product and gtm machine that will be worth $100B by EOY (imo)
Colossus: Scott Wu is the co-founder of Cognition AI, one of the fastest-growing companies in history. He’s also the greatest competitive programmer the US has ever produced. You may have seen him doing impossible card tricks and mental math.
You’ve never seen him asked about weed,
openai's new $10b forward deployed company is so locked in they can't even bother to server side render properly
(jk congrats, nice win for UK AI 🇬🇧)
OpenAI: We’ve also agreed to acquire Tomoro, which will bring 150 experienced Forward Deployed Engineers and Deployment Specialists to the OpenAI Deployment Company from day one.
Consensus
Today, we're announcing $30M in new funding to build the AI OS for Research.
2.5M researchers start their work with Consensus every month. Their work is the foundation that all progress is built upon.
We could tell you our story. We'd rather they did👇
Kayvon Jafarzadeh
karpathy said “i don’t think i’ve typed a line of code since december” and everyone treated it like a meme.
garry tan treated it like a design prompt: what does it look like when one person runs like a whole software team?
gstack is the first oss repo in a while that actually feels like that answer.
not ai as autocomplete.
ai as CEO + staff eng + qa + security + design + release + browser operator + parallel execution layer, all wired through workflows.
and the number is wild: garry claims ~810x higher pace vs 2013, normalized for logical changes (not fake ai loc).
the shift isn’t “faster coding”.
it’s directing + reviewing + orchestrating a swarm without shipping garbage.
stuff that stood out:
→ /office-hours challenges your product before you build
→ /autoplan runs the CEO/design/eng pass
→ /qa literally drives a browser, finds bugs, fixes them
→ /review catches prod-tier issues before you ship
→ /pair-agent + parallel sprints across projects
we’re moving from “ai helps devs code”
to “devs operate systems of ai workers”.
Suryansh Tiwari: http://x.com/i/article/2053414177054842880
This works really well btw, at the end of your query ask your LLM to "structure your response as HTML", then view the generated file in your browser. I've also had some success asking the LLM to present its output as slideshows, etc.
More generally, imo audio is the human-preferred input to AIs but vision (images/animations/video) is the preferred output from them. Around a ~third of our brains are a massively parallel processor dedicated to vision, it is the 10-lane superhighway of information into brain. As AI improves, I think we'll see a progression that takes advantage:
1) raw text (hard/effortful to read)
2) markdown (bold, italic, headings, tables, a bit easier on the eyes) <-- current default
3) HTML (still procedural with underlying code, but a lot more flexibility on the graphics, layout, even interactivity) <-- early but forming new good default
...4,5,6,...
n) interactive neural videos/simulations
Imo the extrapolation (though the technology doesn't exist just yet) ends in some kind of interactive videos generated directly by a diffusion neural net. Many open questions as to how exact/procedural "Software 1.0" artifacts (e.g. interactive simulations) may be woven together with neural artifacts (diffusion grids), but generally something in the direction of the recently viral https://x.com/zan2434/status/2046982383430496444
There are also improvements necessary and pending at the input. Audio nor text nor video alone are not enough, e.g. I feel a need to point/gesture to things on the screen, similar to all the things you would do with a person physically next to you and your computer screen.
TLDR The input/output mind meld between humans and AIs is ongoing and there is a lot of work to do and significant progress to be made, way before jumping all the way into neuralink-esque BCIs and all that. For what's worth exploring at the current stage, hot tip try ask for HTML.
Thariq: http://x.com/i/article/2052796100608974848
Tero Tasanen
Just fired up DS4 by @antirez on my Mac Studio M3 Ultra 256GB and man, it’s seriously impressive. A clean, purpose-built engine for DeepSeek V4 Flash that actually makes frontier-level reasoning feel usable locally.
1M context, strong coherence, and solid speed on consumer hardware. This is the kind of focused, no-bullshit effort that finally brings real frontier models to regular machines instead of just giant GPU clusters. Huge respect @antirez — thank you for building this 🔥
https://github.com/antirez/ds4
My First Million
Replit's founder made $3M in 2 days shipping a half-ready product.
Amjad Masad's team wasn't ready to launch Replit Agent.
He said: "I don't care if it's semi-broken. If it works 50% of the time, it'll wow the world."
For the first time ever, an AI agent could write code, debug it, create a database, and deploy to the cloud.
End-to-end.
He shot a video on his iPhone. Posted it.
Andrej Karpathy quote-tweeted: "I feel-the-AGI moment."
OpenAI and Anthropic researchers reached out: "We didn't know our models could do that."
Day 1: $1M ARR Day 2: $2M ARR
In 48 hours, he made more than 8 years of trying.
Amjad: "Product-market fit feels like stepping on a landmine. We pivoted for years. That moment was the landmine."
If it doesn't feel like a landmine, you haven't found it yet.
Full episode: https://www.youtube.com/watch?v=ddSucXf0CuY&t=1661s
@amasad @thesamparr @ShaanVP
Greg Brockman
Introducing the OpenAI Deployment Company, which will help businesses maximally succeed with their deployments of AI.
Starting with 150 Forward Deployed Engineers and Deployment Specialists, and $4 billion of initial investment from 19 partners.
OpenAI: Today we’re launching the OpenAI Deployment Company to help businesses build and deploy AI.
It's majority-owned and controlled by OpenAI. It brings together 19 leading investment firms, consultancies, and system integrators to help organizations deploy frontier AI to production
Black Forest Labs
The next generation of models won't just generate images - they'll understand worlds, motion, interaction, and action. We've been building toward this for a while.
Visual intelligence is becoming real-time. @stephenbtl spoke about where we're headed at @aiDotEngineer:
You haven’t felt AI progress if you’ve merely use agents and haven’t experienced massively parallel agents.
Merely running multiple agents is fun, but the breakthrough is having them orchestrated correctly, merge back seamlessly, and making 10x faster progress on your projects.
Replit ⠕: Meet Replit Parallel Agents
Build faster by running up to 10 agents in parallel
Each agent gets its own copy of your app
They work on their own computer
Then merge their work agentically
Gopi Krishna
this is such an important take on so many levels from @garrytan
I've been working on GPT models creating different products for over 3 years now.
first, models are now not the bottleneck. Not at least since last Nov when Opus was out.
it is back to engineering problems now. I've built something so cool as a solo-founder running all the GTM myself via Claude Code - quite deterministically.
My posts on instagram look stunning - because I have built skills and commands around it containing python pipelines and markdown files - ad reports come from my Google Ads MCC account, which auto-improves campaigns. Emails - same thing.
Images and Videos - easy peesy. The key is meta-meta-prompting.
in simple words, learn to create the harness around the model. Model may fail, but the system will ensure the outputs don't.
Garry Tan: http://x.com/i/article/2052898104039657472
Prince Canuma
My @aiDotEngineer talk is live: "On-device Intelligence using MLX" 🎥
Huge thanks to @swyx and the team for having me — hands down the best tech event I've been to.
And a shoutout to the community shipping with our packages and pushing the ecosystem forward: @MaziyarPanahi, @adrgrondin, @nopmobiel, @lllucas, @KarnikShreyas, @beshkenadze and many more. You make this fun.
https://youtu.be/zTLJNHj0DeQ?si=cjIPK4ZDm7TUuN_o
Internet Archive
News organizations are increasingly blocking the Wayback Machine even as their reporters still depend on it 📰
In PRESERVING THE WEB IN THE AGE OF AI, Mark Graham, Director of the Wayback Machine at the Internet Archive, explains how major newsroom staff rely on archived web history because their internal archives often miss the deeper public record.
That contradiction sits at the heart of the preservation crisis unfolding online.
🎧 Listen on the Future Knowledge #podcast ⤵️
https://futureknowledge.transistor.fm/episodes/preserving-the-web-in-the-age-of-ai
📚 Read VANISHING CULTURE free from the Internet Archive ⤵️
https://archive.org/details/vanishing-culture-2026
#WaybackMachine #AI #WebArchive #FutureKnowledge @MarkGraham
Replit ⠕
20 builders. Week 3 of 8. No more dreaming, only shipping.
See who's chasing their first dollar this Wednesday.
Episode 3 of Race to Revenue ⠕
Daybreak: our umbrella effort for defensive acceleration, equipping cyber defenders with the best possible frontier AI capabilities.
OpenAI: Introducing Daybreak: frontier AI for cyber defenders.
Daybreak brings together the most capable OpenAI models, Codex, and our security partners to accelerate cyber defense and continuously secure software.
A step toward a future where security teams can move at the speed
I believe the kids call this "@thinkymachines just brutally framemogged gdm and oai".
basically everyone's definition of "realtime" just got a massive frciking upgrade
swyx 🌉: lowkey the funniest videos of the batch. thinky has some comedians!!
congrats to @thinkymachines on reviving the omnimodel dream that others could not
would you call it
a superapp?
Rex Sorgatz: After being a Claude Code devotee for a year, I finally tried Codex on a new project this weekend. Once again, in the matter of a few months, it feels like the world changed.
I can see myself doing *everything* inside of Codex this week.
speaking of things that have gotten over a threshold for me, the combo of the new ChatGPT model, personality, and personalization feels like a new thing
FBI SanFrancisco
#FBI CASE UPDATE: Three men have been indicted on robbery, kidnapping, and conspiracy charges related to a $6 million cryptocurrency robbery spree throughout the Bay Area and LA. Elijah Armstrong, Nino Chindavanh, and Jayden Rucker - all from Tennessee- were charged on Conspiracy to Commit Hobbs Act Robbery, Conspiracy to Commit Kidnapping, Attempted Hobbs Act Robbery, and Attempted Kidnapping relating to a violent robbery spree targeting cryptocurrency owners. Armstrong and Rucker were arrested in Los Angeles on December 31, 2025, and Chindavanh was arrested on December 22, 2025, in Sunnyvale. Chindavanh made his initial appearance in federal court in San Francisco on April 14, 2026, and Armstrong and Rucker made their initial appearances in federal court in San Francisco earlier today. According to the indictment filed March 31, 2025, Armstrong, Chindavanh, and Rucker are alleged to have conspired to kidnap and rob individuals in San Francisco, San Jose, Sunnyvale, and Los Angeles in efforts to steal cryptocurrency from the victims. The defendants traveled from Tennessee to commit the alleged crimes and posed as delivery persons to gain access or attempt to gain access to the victims’ residences. They then used firearms, duct tape, and zip ties to assault their victims, including by binding and restraining a victim in order to force him to divulge his account information. In one of the incidents perpetrated by the conspiracy, the victim was forced at gunpoint to sign into his cryptocurrency accounts so that a co-conspirator could transfer approximately $6.5 million from his cryptocurrency accounts to a wallet controlled by the co-conspirators. Armstrong, Chindavanh, and Rucker are currently in federal custody. Armstrong and Rucker are next scheduled to appear on May 12, 2026, for appointment of counsel before U.S. Magistrate Judge Thomas S. Hixson. Chindavanh is next scheduled to appear on June 26, 2026, for a status hearing before U.S. District Judge Trina L. Thompson.
x.com
