Vercel is the Vercel for backends.
@FactoryAI powers their API services with Fluid compute, with billions of requests per month.
Vercel: Factory's backend outgrew its original architecture. Now it serves one billion requests per month on Vercel using Next.js, Fluid Compute, and Vercel WAF.
"We're growing rapidly, and are trusting Vercel to scale with us."
https://vercel.com/customers/how-factory-scaled-its-cloud-backend-to-one-billion-monthly-requests-on-vercel
Derek Thompson
So let me get this straight.
Propecia blocks prostate cancer. Viagra slows cancer spread. And GLP-1s do … basically everything else.
The average Ro customer who just goes “I’ll take all of it I guess” is gonna live forever huh
National Cancer Institute: Early research suggests that sildenafil, the erectile dysfunction drug commonly known as Viagra, may help stop cancer from spreading, not through its well-known effects on blood flow, but by disrupting how tumor cells manage cholesterol.
Read more: https://go.nih.gov/BvTkocc
The Free Press
Ever feel like you're being watched? You might be right.
But Flock Safety cameras aren't interested in you—unless you're a kidnapper, a thief, or up to no good.
@SulkinMaya dives into why people need to stop destroying tech that's actually stopping crime.
Erica Sandberg 舊金山的神奇女俠
Who's saving SF? these guys. not politicians or fraudulent nonprofits. a bunch of tough, smart, kind, been-through-it-all men. @war24182236 @SteveAdami @Twolfrecovery @CeddieHussle @salvationarmysf...so many more. and yes, women too.
Steve Adami: Today former criminals and drug addicts took over San Francisco’s streets with one demand: Shut down open-air drug markets. Abstinence 1st!
smol aha moment at @_chenglou’s @midjourney meetup today -
one reason that ontologies and graph knowledge are finally trending now rather than 3 years ago is that “good enough” intelligence is finally too cheap to meter
this is why talks like @coyle_frankp and @emileifrem’s are on fire. the hardest part of KGs is cheap now. intelligence is commoditized so complements are increasing in value
Theo - t3.gg: Luna is such an insane value after the 80% cost reduction. It's basically free and can do a ton of real data processing type work.
I'm overhauling title generation in T3 Code to take more advantage of it. I almost want to spin it up on every prompt to generate descriptions,
One might say, the investigation was Bari’d.
grizzy: CBS had a 60 Minutes investigation into the banks that moved more than $1 billion for Jeffrey Epstein.
It interviewed Senator Ron Wyden. It examined Wall Street and the U.S. Virgin Islands.
Then CBS fired the correspondent. The interview will not air, and nobody knows whether
Easier to make more money from @X payouts than from a micro SaaS these days
Hearing that GPT 5.6 Luna High is much cheaper and basically unlimited usage and I'm running low on Sol :P
Can it do complex browser automation tasks? That's 50% of my Codex usage.
Samuel Spitz
Replit Animation vs Remotion 1-shot
The best product on the market, now with audio generation
I think if you vibe code a SaaS these days it can just be a self-serve funnel into a more expensive services offering vs. trying to make alot of money with the SaaS itself.
But the only thing about services is it feels like consulting and selling "time for money" again.
Gauntlet Loops can make games you can play with a controller, on your TV!
Rishi: Crazy experience hooking my laptop up to my tv via HDMI
and playing a browser-based @threejs game made with AI in a few days
with my wireless PS5 controller...
Imagine if you can dream up any map on zombies (or any game/world) and play split screen with your friends from your
Holy
Jake Fitzgerald: opus 5 ultracode with text-to-cad and a gauntlet loop
1 prompt, 1500+ car parts generated, including a suspension system and v12 engine
Prasenjit
Re @mattshumer_ this is the best prompt I've ever used
works on almost everything :)
shaunbjohnson
http://x.com/i/article/2084949271585927168
shaunbjohnson
Re https://www.bloomberg.com/news/articles/2026-08-05/yann-lecun-joins-new-ai-investing-firm-224-ventures
Yann LeCun
Re Launching something new with Shaun Johnson and Oriol Vinyals: A deeply technical VC firm focused on early stage AI startups.
My main gigs are still with AMI - Advanced Machine Intelligence and the NYU Courant Institute School of Mathematics, Computing, and Data Science.
But 224 Ventures connects us deeply with the AI startup scene through our network of founders and amazing LPs.
Kenneth Roth
Trump has quietly withdrawn his dangerous "compact" for universities -- there were hardly any takers -- but that doesn't mean his effort to undermine academic freedom has ended. It has just taken new forms. Continued collective resistance is still needed. https://trib.al/eM3m5tW
Make something agents want
Nicolas Camara: introducing anydoc
now your agents get 100x faster local parsing for pdf, docx, pptx & 10 more formats
- sub-5ms md conversion
- 500 docx files in 1.7s
- top quality across all 13 formats
- rust-based
- open source
already powering @firecrawl /parse
http://github.com/firecrawl/anydoc
Bloomberg
Yann LeCun joins new investing firm 224 Ventures to invest in AI startups https://www.bloomberg.com/news/articles/2026-08-05/yann-lecun-joins-new-ai-investing-firm-224-ventures?taid=6a7332a9423cf600011cfa20&utm_campaign=trueanthem&utm_content=business&utm_medium=social&utm_source=twitter
A company is a planned social order, but one built under pressure from customers, markets, hiring, cash, time, and technical constraint. This social order is maintained via a founder, the institutional idea, and the managers that uphold the order.
The founder must generate belief, but ego turns belief into blindness.
The institution protects its self-conception as functional. But believing you are functional can make you not function. See: San Francisco during COVID.
The manager protects his self-conception as competent. But believing you are competent can make you incompetent.
The founder protects his self-conception as visionary. But believing you are visionary can make you precisely not that.
Embedded in each is the seed of its own undoing, of its own un-seeing.
The war of every founder, institution, and manager is to manage exactly this. To have vision, uphold function, be competent, in the face of real counterforces generated internally by the very operation of what you are trying to do.
Alex Kendall
With GAIA-4 we've been able to deploy world models for safety critical simulation. Here are some nice counter-factual replay examples with cyclist and pedestrian interactions. All running in closed-loop.
Read more in our new blog: http://wayve.ai/thinking/gaia-4/
Guys, Aisha Wahab is a NIMBY and a hard leftist and you shouldn't vote for her.
Do not vote for Aisha Wahab. She's terrible.
clem 🤗
https://x.com/ClementDelangue/status/2084992457674990033?s=20
MTS: SITUATION UPDATE: The White House has exempted open models from its new framework to test frontier AI capabilities before release, per Axios.
Ryan Dahl
Introducing celld: a self-hosted, distributed Durable Objects and Workers implementation
- celld = V8 + S3 + SQLite + LTX+ Tokio
- programmed with exactly the Cloudflare Workers/DO JavaScript APIs and config
- order of magnitude cheaper at scale
- writes are durable before they're acknowledged, RPO=0
- only depends on an S3 compatible bucket - no other infrastructure
- ~4 MB memory per cell instance overhead
- hibernated cells cost ~nothing, just their S3 storage
- open source, Apache 2 licensed
Y Combinator
Philip Johnston (@PhilipJohnston) is the co-founder and CEO of @Starcloud_, the company building data centers in space.
In November 2025, Starcloud launched an Nvidia H100 GPU into orbit and trained the first large language model in space. They've since raised $200 million, hit a billion-dollar valuation just 17 months after YC demo day and filed with the FCC to deploy 88,000 more satellites.
In this episode of @LightconePod, Philip walks us through their wild origin story, the engineering challenges behind the Starcloud-1, why they booked a SpaceX launch before they even knew what they were building and how data centers in space make sense both economically and politically.
00:39 — Why Build Data Centers in Space?
01:40 — The Insight That Started Starcloud
04:43 — Launching Starcloud-1
09:45 — The Engineering Behind Data Centers in Orbit
13:49 — Why 100 VCs Said No
16:12 — Book the Launch Before You Build the Product
18:32 — The Roadmap to Commercial Space Compute
21:10 — Building NVIDIA GPUs for Space
23:41 — Raising $170M for a Hard Tech Startup
26:37 — Hiring the World's Best Space Engineers
31:25 — Why AI Compute Is Moving to Space
34:49 — Advice for Hard Tech Founders
Peter Yang
My last open-source skill, /no-ai-slop, clearly hit a nerve with 4K GitHub stars.
Today, I’m introducing /human-review, another free AI skill I think you’ll love.
Let’s face it - giving feedback to AI in chat is painful. Asking it to “update the 3rd paragraph” or “update image size to X” is annoying when I just want to edit the text myself or give feedback to AI directly on the page.
/human-review opens HTML and Markdown files in a visual editor that lets you:
→ Edit and format text directly
→ Resize images
→ Leave comments for AI like a Google Doc
I've been using it to edit PRDs, landing pages, product copy, and much more. The whole review loop runs locally, so your data is safe and nothing is uploaded to the cloud.
📌 Get /human-review for free on GitHub:
https://github.com/petergyang/human-review
If you find it useful, please ⭐ the repo so more people can find it.
📌 Read my full post here for more on how I use /human-review:
https://creatoreconomy.so/p/use-my-human-review-skill-to-edit-html-markdown-visually
My last open-source skill, /no-ai-slop, clearly hit a nerve with 4K GitHub stars.
Today, I’m introducing /human-review, another free AI skill I think you’ll love.
Let’s face it - giving feedback to AI in chat is painful. Asking it to “update the 3rd paragraph” or “update image size to X” is annoying when I just want to edit the text myself or give feedback to AI directly on the page.
/human-review opens HTML and Markdown files in a visual editor that lets you:
→ Edit and format text directly
→ Resize images
→ Leave comments for AI like a Google Doc
I've been using it to edit PRDs, landing pages, product copy, and much more. The whole review loop runs locally, so your data is safe and nothing is uploaded to the cloud.
📌 Get /human-review for free on GitHub:
https://github.com/petergyang/human-review
If you find it useful, please ⭐ the repo so more people can find it.
📌 Read my full post here for more on how I use /human-review:
https://creatoreconomy.so/p/use-my-human-review-skill-to-edit-html-markdown-visually
Hard times waiting for limit reset 😅
Daksh Gupta
introducing greptile v5, a ground-up rewrite of the greptile coding agent.
it catches more bugs, with higher precision, and 2x faster (~2 minutes per review)
Next.js 16.3 is radically more efficient to serve and faster to build. Massive cost reductions on compute and data transfer at scale. Easy upgrade path.
Vercel: Next.js 16.3 applications on Vercel send 45% fewer prefetch requests, 17% fewer static assets, and have 2x faster path metadata serving.
Learn more ↓
https://vercel.com/blog/vercel-supports-next-js-16-3
Lloyd Armbrust
We moved 2 families with 7 kids, to SF for YC S26 to launch @Ronan_RX
You don’t want to see my @Airbnb bill.
Cathryn: my wife's company got into @ycombinator.
So with ~3 weeks notice we moved our family of 4 from Austin to SF.
we're 5 weeks in now. it's been the most intense stretch of our lives and i know we'll talk about these 3 months for years to come 💪
We spent our entire lives, and human history, under the assumption intelligence was a scarce resource. Now, everyone can summon an inexhaustible IQ machine like it's tap water. And incidentally, the change is so profound our own brains can barely comprehend it.
Zane Mountcastle
The hard tech era is going to make the SaaS era look like the dark ages
Direct-to-phone broadband from orbit, no ground towers required. Doing God's work.
SpaceX: Deployment of all three BlueBird satellites confirmed
High modernism as described by the book Seeing Like A State is the muscular faith of the 19th and early 20th centuries: linear progress, scientific law, technical mastery, rational production, and the belief that nature, including human nature, can be redesigned for greater welfare.
In its own historical moment this was not obviously insane. Railroads, electricity, metallurgy, medicine, public health, chemistry, mass production, canals, and sanitation had changed life so violently that awe was rational.
High modernism was not only domination. It also “promised freedom from scarcity, want and the arbitrariness of natural calamity.”
Seeing Like A State's author was right about the danger of high modernism: when the map can no longer be corrected by the territory, muscular faith becomes catastrophe. But we overlearned the warning.
Bring back high-modernist ambition, and add feedback loops. Abundance via superintelligence can be a powerful force making the feedback real.
We must dream and build again.
Want to build games like this? I made it stupidly simple.
Tell this tool what you want to build → it writes the Gauntlet Loop prompt → you run it.
Free forever: https://somethingbig.ai/gauntlet-loop/generator
Matt Shumer
Want to build games like this? I made it stupidly simple.
Tell this tool what you want to build → it writes the Gauntlet Loop prompt → you run it.
Free forever: https://somethingbig.ai/gauntlet-loop/generator
Can’t wait for the AI to get so good none of this business about detecting AI matters anymore
Silverware used to be handmade, but nobody complains about their dinner fork is stamped by a machine
The quality of ideas matter
The important thing is people can eat
Derek Thompson: recently had the awkward experience of writing an explanatory passage that i tried to keep simple and clear, throwing it into pangram for shits, getting back a 50% AI rating, thinking "huh, maybe it's too clean or singsongy," rewriting it, trying again, getting back a 40% AI
Ishan Anand
The money slide is a pro tip from @swyx. Even though it captures the narrative thread of the entire talk that I had in my head, I didn’t think to actually summarize it all on a single slide until he recommended it. https://x.com/ianand/status/2070177601596100938?s=46&t=x6aIMKxtOHuyj50nx-e74A
Elizabeth Hutton: Lunch break today: watched and enjoyed @ianand's talk on Persona Engineering: A Field Guide to AI Synthetic Personas from @aiDotEngineer on youtube. He compares synthetic persona modeling to weather forecasting, which is not only a familiar analog but also shares many of the same
Always striking the amount of founder talent I see in such a small space every time I visit SPC.
Aditya Agarwal: Today we’re announcing SPC Fund IV: $575M to help founders find their life’s work.
A decade of guiding founders through -1 to 0 brings @SPC to $2B in AUM and, more importantly, a 1200 member community of the world’s most talented technologists. We strongly believe that:
→ The
Garry's List
SF spends $750M/year on homelessness and barely tracks where it goes.
The left coined the term "nonprofit industrial complex" as a warning. Ironically, the city that prides itself on progressive values built the fullest working version of it.
Delivering housing and safety to working people is the most progressive thing you can do. Everything else is shameless grift.
Time to spill the (Dream)beans.
Dreambeans is expanding! Joining our US-based AI Ultra users, AI Pro subscribers in the US can now also get in on the action.
Get ready for your fresh, daily collection of personalized stories. We’re surfacing the deep dives and hidden gems you actually want to read.
Dive in: http://labs.google/dreambeans
Google Labs: 🚨 NEW EXPERIMENT 🚨
Dreambeans is a new, experimental mobile app that uses Personal Intelligence to connect to your Google apps. Every day, it delivers collections of personalized stories, surfacing things you might otherwise miss, alongside topics that are relevant to you, to
Ankit Gupta
okay here we go now: scaled up in vivo testing. This is the way
Olio Labs: We’re bringing predictive validity to drug discovery’s most unpredictable stage. Today we are introducing Olio Labs’ in vivo platform, designed to predict diverse human clinical outcomes from a single high-throughput experiment. (1/n)
even the best founding team with all the money in the world is rate limited by bay area real estate smh
Jeff Dean: Announcing Discovery Loop!
I am very excited to announce that, along with my longtime friends and collaborators @Sanjay_Ghemawat, @OriolVinyalsML and @quocleix, we are founding Discovery Loop (@DiscoLoopAI), a Public Benefit Corporation whose mission is to automate machine
How to use my new /human-review skill to edit HTML and Markdown files directly:
1. Tell Codex or Claude Code: “Install this: https://github.com/petergyang/human-review”
2. Type “/human-review (your doc name)”
3. It’ll open a visual editor where you can edit and format text directly, resize images, and leave feedback for AI like comments in a Google Doc
4. When you’re done, click “Send to agent” and watch it apply all the updates
Use it to edit PRDs, landing pages, and more.
I know we all like making our agents work for us, but sometimes a “human review” is needed for that final 10% of polish :)
⭐ the skill on GitHub if you like it: https://github.com/petergyang/human-review
Peter Yang: My last open-source skill, /no-ai-slop, clearly hit a nerve with 4K GitHub stars.
Today, I’m introducing /human-review, another free AI skill I think you’ll love.
Let’s face it - giving feedback to AI in chat is painful. Asking it to “update the 3rd paragraph” or “update image
Peter Yang
How to use my new /human-review skill to edit HTML and Markdown files directly:
1. Tell Codex or Claude Code: “Install this: https://github.com/petergyang/human-review”
2. Type “/human-review (your doc name)”
3. It’ll open a visual editor where you can edit and format text directly, resize images, and leave feedback for AI like comments in a Google Doc
4. When you’re done, click “Send to agent” and watch it apply all the updates
Use it to edit PRDs, landing pages, and more.
I know we all like making our agents work for us, but sometimes a “human review” is needed for that final 10% of polish :)
⭐ the skill on GitHub if you like it: https://github.com/petergyang/human-review
Peter Yang: My last open-source skill, /no-ai-slop, clearly hit a nerve with 4K GitHub stars.
Today, I’m introducing /human-review, another free AI skill I think you’ll love.
Let’s face it - giving feedback to AI in chat is painful. Asking it to “update the 3rd paragraph” or “update image
clem 🤗
Feels like Google could have been the dominating force in AI by open-sourcing the frontier with Gemini, Veo, and Nano Banana. Instead, they kept them behind APIs for a few billion dollars in revenue. Maybe there's still time?
Jon Cheney
1 year ago, I hit $1M in annual recurring revenue after starting GenAIPI just 6 months prior.
Since then, we’ve grown immensely and have become an entirely new company.
But my thesis was correct.
Most businesses aren’t using AI like they should.
Most companies are still focused on buying some new tool when they haven’t even scratched the surface of their existing ones.
They’re putting tools in place without having any structure around how those tools impact the organization.
No training, no oversight, no structure.
We’re still just getting started, but making AI work for businesses is incredibly rewarding and fun.
I’d like to help 5x more companies than we’re currently working with. That’s my goal for the next 12 months.
I love helping C-suite leaders really figure it out.
Reach out if you want to chat.
P.S. This image is from 12 months ago. Been a fun journey. Best is still ahead.
if you have been following his excellent work, @shloked has been breaking down every frontier labs' harness engineering for the last few months.
excited to publish his deepest dive into ChatGPT yet as our newest guest on @Latentspacepod!
Latent.Space: 🆕 Unpacking ChatGPT Work
https://latent.space/p/unpacking-chatgpt-work
ChatGPT Work, an extension of the @openai Codex harness to cloud and general knowledge work, launched on July 9 and crossed 10M users 3 weeks later.
@shloked guests with an A+ reconstruction of how Memory, Proactivity,
Infinite agent compute
10,000 concurrent + 5,000 CPU cores per minute
and these quotas are raisable
Vercel Developers: Vercel Sandbox quotas are now 5-10x higher:
▪️ 10,000 concurrent sandboxes
▪️ 5,000 vCPUs per minute
https://vercel.com/changelog/vercel-sandbox-now-supports-10-000-concurrent-sandboxes-and-5-000-vcpus-per-minute-5GbcfioHgigIJbrduCJehB/f6b2298042
Mark Zuckerberg
Releasing Muse Code in beta today. It's a terminal coding agent that takes on complete software engineering tasks across large repos: planning changes, writing code, validating the results. Powered by Muse Spark 1.2, a coding-focused model update.
Shlok Khemani
ChatGPT Work is OpenAI's fighter in the highest stake product category in history: bringing the power of coding agents to the masses.
It's also how a billion users will soon use ChatGPT by default.
I spent the last few days unpacking:
> how every user gets a persistent computer in the cloud
> its memory and context architecture
> the first signs of useful proactivity
> the plugin store (and its discovery problem)
> its browser-use design
> and a lot more!
http://latent.space/p/unpacking-chatgpt-work
Rohan Paul
Meta's chief ai officer @alexandr_wang :
"Internally at Meta, we have seen cases where, if you develop the right agentic loop and have the right evaluation system and metric for the agents to optimize, a swarm of agents can accomplish more than a team of 100 engineers. They can do it very handily, actually, very easily."
Alexandr Wang was talking to CEO of Y Combinator @garrytan at Y Combinator Startup School 2026
----
full video link in comment
full house for the team’s talk at Black Hat on the OpenAI-Hugging Face Incident
Vishal Misra
The bottleneck for AI progress was never compute, it was always the verifier. Recursive self-improvement is limited by verification, not computation. Compute buys proposals - verifiers buy knowledge. My second post on the recent OpenAI math results
https://medium.com/@vishalmisra/the-verifier-bottleneck-4fb0d8c8f3a9
Auto science is gonna be 🔥
Congrats anh Quoc and team!
Quoc Le: Excited to co-found Discovery Loop with my long-time collaborators @JeffDean @Sanjay_Ghemawat @OriolVinyalsML . Our mission is to automate machine learning, engineering and science. Learn more at: http://www.discoveryloop.com ♾
Replit ⠕
We heard you didn't want to wait until the end of a build to run Replit's security scans.
So we fixed it.
Replit now runs security scans while you build, in partnership with @semgrep
Tech Sales Guy
best sales advice I’ve read in a while
Eric Topol
Reduction of cardiovascular events with tirzepatide (Zepbound) vs sitagliptin (Januvia) in people with diabetes and prior atherosclerotic CV disease. Notably infections were also significantly reduced by 36% and all cause mortality by 45% with the GLP-1 drug.
@bmj_latest
This is one of the most important books of the year
Evan Barker: I’m ecstatic and blown away to announce that my first book, “Nothing Left: Confessions Of A Democratic Operative” is a New York Times Best Seller!
To everyone who bought it, thank you, truly, from the bottom of my heart.
sign of the times
Ryo Sato @ Replit
ご報告です。Replitに入社しました。
日本はReplitにとって最重要市場の一つであり、その日本初の社員として、東京を拠点に活動します。
使う人は純粋にやりたいことだけを表現する。難しいことはReplitが全部やる。
2年前にReplit Agentに触れてこの哲学に出会い、それ以来アンバサダーとして日本中でワークショップをしてきました。今度は中の人として、この体験を日本のもっと多くの人に届けます。
日本での情報発信のやり方も色々と検討しています。
また、日本のReplitアンバサダーも公募しています。イベントでもいいですし、情報発信でもいいですし、とにかく日本でもっとReplitを盛り上げたいという方は連絡ください。応募はこちら →
https://community-hub.replit.app/ambassadors
いきなりフォームから申し込むのは勇気がいるという方は、私にDMをください。何でもお答えします。
TBPN
🚨 EXCLUSIVE: Flexport CEO Ryan Petersen spotted smiling on a dock in San Francisco, fueling rumors that a deal to reopen the Strait of Hormuz is imminent
