Nikunj's no-moat manifesto (1483 likes, top of the day): 'The models have no moat (OpenAI, Anthropic, XAI). The IDEs have no moat (Cursor, Windsurf). The harnesses have no moat (Cognition, Factory, LangChain). The app builders have no moat (Replit, Lovable, Bolt). The wrappers have no moat (Harvey, Abridge, OpenEvidence). The inference providers have no moat (Together, Fireworks, Groq). The voice layer ha[s no moat]...' Plus his hot take (789 likes): 'Brand marketing is going to be THE major differentiator and one of the most prized assets for a company going forward. You still need retention before attention.'

Rauchg ships Cursor Origin × Vercel deployment (3600 likes): 'You can now host your repos in Cursor Origin and deploy to Vercel via Cursor Origin which is itself hosted on Vercel. And unlike GitHub, it's online 😁' Sottiaux's midnight Codex song (971 likes): 'Gimme, gimme, gimme Codex after midnight / Won't somebody make these failing tests all go away? / Gimme, gimme, gimme Codex after midnight / Ship it through the darkness by the start of the day' And his community ask (3726 likes): 'What is an obvious thing that we should do with Codex, API or our models that we should just do but haven't yet? What is 100% within reach, but we just seem to be missing?' Boris Cherny's quality-of-life improvements (678/739 likes).

Thariq's double: the CC /design command (880 likes): 'Go into CC and type /design <something you want to design>. Do it rn.' And his update on creative coding (470 likes): 'All of the recent proc gen art, video editing and 3d game demos recently have made me update towards LLM coding models being better at a lot of creative work than diffusion models.' Garry Tan open-sources his Personal AGI kit (11-20 likes): 'A private github repo with 70 of my proven skills and the beginnings of your Karpathy-style knowledge wiki. All MIT-licensed and free. Works with your existing Claude Code or Codex subscription right now.'

Madhu Guru's evals playbook (198 likes): 'The best way to get good at evals is to take a workflow you know really well and figure out how to make its quality measurable. Study the actual traces — the sequence of prompts typical users have, what good responses would look like at each step and for the end to end outcome.' Levie on data as the new oil (304 likes): 'AI has such a thirst for data that we're entering an era where it's valuable almost in any form. In a world of AI, information actually belongs as an asset on the balance sheet.' Amjad's stealth AI company observation (156 likes): 'This team doesn't have AI anywhere in their pitch but has AI growth rates and would have 10x the headcount if they weren't so AI-pilled.' Plus pen-testing over scanning (267 likes). Josh Woodward's Gemini roadmap update (467 likes): revamped Workspace tools, 3.7 Flash tool calling, new Projects design, 49 connectors and counting.

No Priors hosts Sarah Guo and Elad Gil talk risk management, RSI, trillion-dollar companies, token budgets, and regulatory capture. Key threads: three companies went from ~zero to a trillion in five years (Anthropic, OpenAI, SpaceX) — an unprecedented inflection, since these arcs usually take 15-20 years; people now assume a bunch more trillion-dollar companies will form in 3-5 years, but punctuated equilibrium says waves consolidate — 'there's still a lot of $100B companies to be built, but multi-trillion is hard'; trillion requires $50-100B revenue with good margin — very few markets support that; coding was a much bigger market than anyone thought, and the evidence is now public; Sarah's pushback: investors still under-imagine market size, evaluating Harvey on per-lawyer TAM instead of pricing outcomes; compute constraints enforce an oligopoly by capping any single lab's rate of progress; the '18 months to RSI' belief has weird second-order effects — researchers asking whether to get married, burnout risk, 'psychologically similar to thinking you're going to die'; the power law of researchers — a few dozen drive 80% of results at any lab, and compute is now differentially allocated to them, making hiring slower; return on invested tokens — who gets the token budget and why; why the death of SaaS is overstated (tokens are better spent on core product than on SaaS subscriptions); Minecraft as the original multi-billion one-person company; displaced engineers permeating to GE/P&G/Hershey's; the California billionaire tax and possible exit tax — 'the fastest way to chase the entire ecosystem out'; Texas energy and hardware corridors rising purely from regulatory arbitrage; transformer alternatives getting copied or lapsing to the labs with compute; and the closing call to arms against regulatory capture — France at 70% nuclear with nothing happening, the US at 18% having built no reactor in 40 years, 'a safety lobby in the seventies basically killed abundant clean energy'; risk-only regulation ignores benefit, and tech succeeded because it was lightly regulated.