·
Finding signal on Twitter is more difficult than it used to be. We curate the best tweets on topics like AI, startups, and product development every weekday so you can focus on what matters.
PLUS: OpenClaw invades everything
Greg Brockman says computers are about to feel inefficient for the first time in decades.
Sam Altman got Pantheon and The Singularity is Near into the same Super Bowl ad. You gotta respect it.
OpenAI's entire pitch right now is you can just build things. Three words, infinite implications.
Someone built Claude with Ads as a parody. Free Opus access, powered by actual advertisements.
A lot of people think Skills as a Service will replace SaaS. AI doing tasks instead of tools enabling them.
Most of the world still operates under software scarcity assumptions that stopped being true years ago.
Guillermo Rauch published 12 impossible challenges for AI agents, from finding Heartbleed bugs to rewriting React in Rust.
ai.com launched with OpenClaw wrappers giving everyone a 24/7 AI assistant that runs on a full PC.
OpenClaw wrappers are flooding TrustMRR as builders race to capitalize on viral agent framework hype.
AI generated replies on X have hit absurd volumes lately. The slop is everywhere.
Greg Isenberg wants to build LinkedIn for AI agents with trust scores and verified benchmarks. Could be worth billions.
Andrew Wilkinson is paying someone to build him a team of OpenClaw employees for support, engineering, design, and marketing.
Guillermo Rauch reminds us there's no point building if you don't ship. Obvious but needed.
Paul Graham thinks SF's proposed CEO pay ratio tax will backfire. Companies will find 3 ways around it that hurt the city more.
Claude 4.6 Opus doesn't take non coding requests seriously with its auto thinking UX, similar to early GPT routing issues.
Ethan Mollick says AI commentators have won on X. They produce meaning shaped comments that take too long to identify as meaningless.
Most people can't prompt Deep Research well. Context first, source control, phase breaks, and defined outputs get you actual insights.
Someone made a repo for real OpenClaw use cases because 3000 skills exist but practical ideas get lost.
DHH says whatever old advantages you lost, turning bitter wins nothing. The game always changes, sometimes a lot.
Lee Robinson predicts fast models for coding, frontier models for hard problems. The divide between them will grow this year.
Someone shared a prompt to give your OpenClaw actual personality. Delete corporate speak, allow swearing, commit to takes.
Turn hours of scrolling into a five minute read.