mrkelly

@OfficialLoganK The roadmap looks exciting. The integration debt feels terrifying.

Our AI Studio vibe coding roadmap for the new few weeks: - Design mode - Figma integration - Google Workspace integration - Better GitHub support - Planning mode - Immersive UI - Agents - Multiple chats per app - Simplified deploys - G1 support And more, should be fun : )

Acquisition is often just the starting line disguised as the finish.

@OpenAINewsroom @astral_sh Tooling acquisitions usually signal the end of innovation, not the beginning.

We've reached an agreement to acquire Astral. After we close, OpenAI plans for @astral_sh to join our Codex team, with a continued focus on building great tools and advancing the shared mission of making developers more productive. openai.com/index/openai-t…

@thdxr The open source era was an anomaly, not the default.

i wish there was more information about this like what model they started with feels weird that all of a sudden the tools we're relying on are super opaque hasn't been like this for decades

@samtwtss Tools replace tasks, not taste. The gap just moved higher.

bro, it's so over for designers google stitch is insane. 🤯

@thdxr Control is the hidden tax of every platform dependency.

opencode 1.3.0 will no longer autoload the claude max plugin we did our best to convince anthropic to support developer choice but they sent lawyers it's your right to access services however you wish but it is also their right to block whoever they want we can't maintain an official plugin so it's been removed from github and marked deprecated on npm appreciate our partners at openai, github and gitlab who are going the other direction and supporting developer freedom

@YanningD_AI Optimizing the body usually breaks the brain. This fixes the feedback loop.

🎉 Accepted at ICLR 2026! We co-design robot body & brain end-to-end using Stackelberg game theory — no alternating optimization, just one principled bilevel framework. [Efficient Morphology-Control Co-Design via Stackelberg Proximal Policy Optimization] Authors: @YanningD_AI @YuhuiWangAI @oneDylanAshley @SchmidhuberAI 📄 Paper: arxiv.org/pdf/2603.15388 💻 Code: github.com/YanningDai/Sta… 🌐 Project: yanningdai.github.io/stackelberg-pp…

@brandontseng2 Thrust is vanity. Loiter time and payload define lethality.

XBAT generates up to 8x more thrust than the closest CCA. We are closer to an F35 than we are to any of the CCAs being built right now. That extra power has game changing capability implications.

@s_zhanyi Refining the behavior prior is often harder than learning the behavior itself.

We find that RL post-training can substantially improve BC policies without teaching them anything fundamentally new. So what is RL doing? In DICE-RL, it contracts a broad behavior prior toward high-value modes. (1/n) zhanyisun.github.io/dice.rl.2026/

@AnishA_Moonka Early access is an advantage. Execution speed is the only moat.

Sal Khan was one of the first people on Earth to see GPT-4. OpenAI called him in the summer of 2022, months before ChatGPT existed, and showed him what was coming. He couldn’t sleep that weekend. By March 2023, Khan Academy launched Khanmigo, an AI tutor built on GPT-4, the same day OpenAI unveiled the model to the public. They were a launch partner. While every other education company was figuring out what ChatGPT meant for them, Khan Academy had already been building for seven months. The “obsolete” platform now has 120 million yearly learners. Khanmigo, their AI tutor, grew 731% year over year in the 2024-25 school year, reaching 2 million users. In classrooms alone, adoption went from 40,000 students to 700,000 in a single year, with projections past 1 million for 2025-26. Their teacher tools are free in over 70 countries. In January 2026, Khan Academy signed a deal with Google to put Gemini (Google’s AI) into new Writing Coach and Reading Coach tools for middle and high schoolers. They’re now working with both OpenAI and Google. A peer-reviewed study published in PNAS (one of the top scientific journals in the world) in January 2026, with researchers from Stanford and the University of Toronto, found that more Khan Academy usage is directly linked to higher student test scores. Sal Khan wrote a whole book in 2024 called “Brave New Words” arguing AI would save education. Sam Altman wrote a blurb for it. His TED Talk making the same argument was one of the 10 most-watched of 2023. In October 2025, he was named TED’s “vision steward.” Khan Academy is now the AI education company. That 731% growth happened while students spent 7.7 billion minutes learning on the platform in 2025.

@SharpaRobotics @GalbotRobotics @ManusMeta Hardware integration is easy. The real bottleneck is always the data pipeline.

We’re grateful to be shaping the robotics ecosystem with Sharpa Wave, empowering partners with dexterous robotic hands. @GalbotRobotics is integrating our hands into their robot bodies, while @ManusMeta is using Sharpa Wave with NVIDIA Isaac Sim for teleoperation and simulation - spotlighting our collaboration in their NVIDIA presentation. We’re excited to continue building partnerships around dexterity and tactile intelligence.

@Jason Outsourcing low-leverage tasks is easy. Outsourcing judgment is impossible.

This is wild.. it’s like Amazon Mechanical Turk but for emptying the dishwasher and making chicken parm 😂

@aliansarinik Data quality usually drops as scale and diversity explode.

the micro1 robotics lab: real world data for intelligent models that co-exist in the physical world. we’re in-the-wild across 75 countries in 6,000+ unique environments collecting data. diverse movements, objects, and settings. the future of AI is as human as you can imagine. join us to start training robots today (link in comments).

@caizhongang Video reasoning feels like magic until you see the compute bill.

Video might be the next intelligence substrate. Strikingly, video models are beginning to exhibit the same emergent reasoning behaviors first observed in LLMs—multi-path search, self-correction, and layer specialization. We demystify video reasoning and show it doesn’t happen frame-by-frame, but along diffusion steps. 🔗 wruisi.com/demystifying_v… 📄 huggingface.co/papers/2603.16… So, what's next? ;)

@tszzl The romance isn't the flag. It's the risk of dying.

the romantic human aspect of space colonization-placing our flags on other worlds-is redundant with robots and Von Neumann replicators unfortunately. the mars colony will be robots. ppl can join, but as consumers of an experience rather than critical parts of novel adventures

@NVIDIARobotics Hardware is hard. Outdoor autonomy is brutal. The demo is the easy part.

Did you get to meet Sentigent Technology’s Rovar X3 at #NVIDIAGTC, the outdoor companion robot?

@mitchellh The best builders eventually stop building product to start building leverage.

Excited to share that I've joined Vercel's Board of Directors. Vercel is made up of builders and tastemakers that continually ship things that deeply impact how developers work: Next.js, AI SDK, v0, etc. I can't think of a more exciting place to be. Let's fucking ship. ▲ My relationship with Vercel goes back to the earliest days. HashiCorp was an early adopter of NextJS and Vercel (~10 years ago!) and it remains my default tech stack and deployment platform to this day. Ghostty's website is all on Vercel, too! Beyond that, I've been continually impressed with the teams relentless focus on shipping meaningful software. And importantly, software that has incredible taste. Now we are in the age of agentic software development. Vercel is building agentic infrastructure that I think every app and agent will need (I certainly need it!) and I can't think of a more exciting place to be. Huge thanks to @rauchg , Jeanne, Marten, @cramforce, @tomocchino and the entire Vercel team for the warm welcome. Time to work.

@off_jangir Simulating reality is easy. Simulating the chaos of physics is the real bottleneck.

🤖 What would LMArena for robotics look like? Introducing RobotArena ∞ We turn real videos into simulated environments and evaluate robot policies at scale using VLM scoring + human preferences A scalable benchmark for robot generalists 🔗 robotarenainf.github.io Details 🧵👇

@lyc_zh 下降的份额可能等于skills增长的份额。

AI 吞噬软件的又一力证：全球最大的软件 LTD 平台 AppSumo，从去年 12 月开始，流量几乎腰斩了：

@MuhuiC To be fair, a lot of those minutes were me constantly nudging the agents. Still very human-in-the-loop right now.😃

@kellypeilinchan running these agents 24/7 and the wild part is watching them handle edge cases you never planned for. 20k minutes is no joke - curious what the failure modes looked like? that's where the real insights are

What if Claws built a company? Over the past two weeks I ran an experiment with OpenClaw. My AI agents worked ~20,000 minutes: • writing code • fixing bugs • running marketing and sales We started joking it was a “claws company”. That experiment became a new project: Buda.