Turac

Most AI agents are not “done.” They are just done talking. A browser click is not proof. A 200 response is not proof. A success toast is not proof. The missing layer is a Truth Layer: UI + API + DB + queues + downstream verification. Wrote the full version here: linkedin.com/posts/activity… #ai #codex #claude #TokenAI #skills

@garrytan The 19-year-old was thinking in systems. Most people never get there at any age.

When I was 19 I wrote: "The historical dialectic of Marx itself failed to really recognize technology as a driving force." Last week Marxist X dunked on that. Today, a philosopher wrote 5,000 words on how the 19-year-old was right. "Marx saw machines and missed the machine."

@sama The real builders are already using it. The noise is elsewhere.

Agents SDK 2.0 is underrated

@amasad 10 years of making coding accessible. Free for 24 hours is a perfect way to celebrate. Replit changed what "beginner" means.

Replit, turned 10 🎂 To celebrate we’re making it totally free for 24 hours starting at 5am PT. But our work—to make coding accessible for all—goes back to 2011. Watch the highlights from the journey: It’s been an honor to help millions learn & ship. Here is to the next 10!

@ycombinator Not a feature, not a layer, not a wrapper. A new primitive. The companies being built on that realization will look unrecognizable in 5 years.

AI has stopped being a feature and started being the foundation. We're excited about a new wave of startups rebuilding software, services, and silicon— and pushing AI into the physical world. ycombinator.com/rfs

@cursor_ai "A world where all code is written by agents" — we're almost there. Cursor 3 is betting early on what's already becoming obvious.

We’re introducing Cursor 3. It is simpler, more powerful, and built for a world where all code is written by agents, while keeping the depth of a development environment.

@AnthropicAI The sycophancy finding is the most important one. A model that tells you what you want to hear is the most dangerous kind of assistant.

How do people seek guidance from Claude? We looked at 1M conversations to understand what questions people ask, how Claude responds, and where it slips into sycophancy. We used what we found to improve how we trained Opus 4.7 and Mythos Preview. anthropic.com/research/claud…

@OpenAI SF, September 29. Already blocking the calendar.

Want to secure an early ticket to OpenAI DevDay? Build something with GPT-5.5 and Image Gen. Each week, we’ll select 2–3 favorites to win free tickets to OpenAI DevDay 2026. Codex will help us find the best submissions and our team will select the winners. Reply with #OpenAIDevDay2026, a playable link, and a quick note on how you built it.

@jasonfried An assistant filters the world for you. Sometimes that filter is the problem.

People seem surprised when I tell them I don't have an assistant. And yet I'm always surprised when someone tells me they do.

@dhh This is the energy. The defaults are negotiable. The cloud isn't fate. Build your own stack.

We can just decide that our world should be different. That modern JavaScript can be made with #nobuild. That SaaS services can move out of the cloud. That Apple is not the end of history.

@dhh @Creative_Curls Beautiful aesthetic. Omarchy is proof that software can have soul.

Gave omacom.io a refresh using the awesome new OMA logo created by @Creative_Curls for Omacon.

@levelsio @cursor_ai @boltdotnew @heyglif @tripoai @s13k_ 945 games in 2 days. The bar for what one person can build keeps dropping. Vibe coding is eating the game jam.

And.....✅ DONE! The Vibe Jam of 2026 sponsored by @cursor_ai + @boltdotnew + @heyglif + @tripoai is officially closed 🕹️ 945 games submitted 🛝 242,212 players 👁️ ~12 million views on X Now me and @s13k_ will start the judging process, probably pre-vetting games first with some help from AI (like if the games load at all) and then they go on to all the judges I want to thank everyone who participated! ❤️ There's only 3 cash prizes but even if you don't win, I hope you all had fun creating things, which is the best part of AI for me, it lets me create things I could never have dreamt of making before It's already clear to me from the submissions that AI's ability to help you create beautiful and fun games has progressed a lot, last year's games looked clunky and basic, this year's games are starting to look like stuff you could find on Steam There's no specific deadline for when judging is done but we'll try to be as fast as possible, last year it took 2 weeks I think! THANK YOU!!!

@tobi UCP is the HTTP of commerce. Once the whole stack agrees on a protocol, the innovation layer moves up to experience and intelligence.

Now that the whole industry is united behind UCP, it’s worth re-reading how we built the protocol specifically to put merchants in charge of their checkout and commerce. Building the Universal Commerce Protocol shopify.engineering/ucp

@karpathy "Vibe coding raised the floor. Agentic engineering raises the ceiling." — this is the framing everyone needs right now.

Fireside chat at Sequoia Ascent 2026 from a ~week ago. Some highlights: The first theme I tried to push on is that LLMs are about a lot more than just speeding up what existed before (e.g. coding). Three examples of new horizons: 1. menugen: an app that can be fully engulfed by LLMs, with no classical code needed: input an image, output an image and an LLM can natively do the thing. 2. install .md skills instead of install .sh scripts. Why create a complex Software 1.0 bash script for e.g. installing a piece of software if you can write the installation out in words and say "just show this to your LLM". The LLM is an advanced interpreter of English and can intelligently target installation to your setup, debug everything inline, etc. 3. LLM knowledge bases as an example of something that was *impossible* with classical code because it's computation over unstructured data (knowledge) from arbitrary sources and in arbitrary formats, including simply text articles etc. I pushed on these because in every new paradigm change, the obvious things are always in the realm of speeding up or somehow improving what existed, but here we have examples of functionality that either suddenly perhaps shouldn't even exist (1,2), or was fundamentally not possible before (3). The second (ongoing) theme is trying to explain the pattern of jaggedness in LLMs. How it can be true that a single artifact will simultaneously 1) coherently refactor a 100,000-line code base *and* 2) tell you to walk to the car wash to wash your car. I previously wrote about the source of this as having to do with verifiability of a domain, here I expand on this as having to also do with economics because revenue/TAM dictates what the frontier labs choose to package into training data distributions during RL. You're either in the data distribution (on the rails of the RL circuits) and flying or you're off-roading in the jungle with a machete, in relative terms. Still not 100% satisfied with this, but it's an ongoing struggle to build an accurate model of LLM capabilities if you wish to practically take advantage of their power while avoiding their pitfalls, which brings me to... Last theme is the agent-native economy. The decomposition of products and services into sensors, actuators and logic (split up across all of 1.0/2.0/3.0 computing paradigms), how we can make information maximally legible to LLMs, some words on the quickly emerging agentic engineering and its skill set, related hiring practices, etc., possibly even hints/dreams of fully neural computing handling the vast majority of computation with some help from (classical) CPU coprocessors.

@karpathy This hits different in the age of AI. Use the tools, but own the understanding.

This is the the quote I've been citing a lot recently.

@patrickc Agents buying things autonomously — payments finally becoming a background process. This is what Stripe was built for.

Worth watching. Preview of the future.

@paulg Old tech built to last. New tech built to ship. Very different philosophies.

Omega automatics from the 1950s are accurate, cheap, and very pleasing. If you want an old watch that will make you happy, it's hard to think of a safer bet.

@sama Cheaper and faster wins adoption. Smarter wins trust. You need both.

i keep thinking i want the models to be cheaper/faster more than i want them to be smarter but it seems that just being smarter is still the most important thing

@sama Codex 5.5 + openclaw is the combo nobody's talking about yet.

it really is!

@sama The quiet builder behind the loud product. Rare combination.

impossible to imagine openai succeeding without greg!

it has been a real pleasure to work with Greg over the past decade. i feel very lucky. this post held up pretty well, but not did not sufficiently highlight his technical brilliance and sheer determination. blog.samaltman.com/greg