·
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: Software development gets a culture reset
Claude Opus 4.6 drops with 1M token context, better planning, and it actually catches its own mistakes.
Changing user behavior means you have to overcorrect for a while, especially with long tenured users.
OpenAI's new GPT-5.3-Codex just dropped, and you can just build things now, apparently.
Anthropic gave Opus 4.6 two weeks to build a C compiler that runs on Linux, mostly hands off.
Two flagship coding models dropped within minutes of each other, and the timing feels deliberate.
Vibe coding is here, and vibe research is coming next, whatever that means.
OpenAI's new Frontier platform helps enterprises build AI coworkers that do real work, not just chat.
GPT-5 connected to an autonomous lab cut protein production costs by 40% by running its own experiments.
Benchmarks undersell it, but this model feels like a real step forward when you actually build with it.
Greg Brockman says agents are now the first tool at OpenAI, not editors or terminals. Software dev is getting retooled.
v0 switched to Opus 4.6 because they're giving the people what they want.
levelsio dropped another $237K into Tesla, betting everything merges into X Corp and wants in early.
Coding looks like a chat app now, not an IDE, and that feels like the future.
Ethan Mollick asks AI labs to slow down for half a decade so the world can catch up with testing.
Infrastructure noise can swing coding benchmarks by several points, sometimes more than the gap between top models.
Someone used Opus 4.6 to script videos, build React components, and never looked at the code once.
Claude Code's new agent teams feature lets you spin up four agents to hunt and fix security vulnerabilities automatically.
Marc Lou moved his startup to Cyprus for 12.5% corporate tax and zero capital gains, plus good vibes.
Opus 4.6 hit 93% on ARC-AGI-1, the new state of the art benchmark score.
Ethan Mollick wants a moratorium on floating blue holographic brains and 1990s AI stock imagery.
Lovable added Opus 4.6 and claims it makes their app building 21% better on their benchmark.
Basecamp turns 22, and DHH says agents are using human affordances in their projects already, not special tools.
Turn hours of scrolling into a five minute read.