This website uses cookies
Read our Privacy policy and Terms of use for more information.
Sep 16, 2026
•
5 min read
CEOs call for a slowdown, but Washington and Beijing say ‘ain’t happening.’ AI vendors are deploying armies of engineers to get customers' systems working. A lesson from the TSA that lets you keep your shoes on.
Sep 8, 2026
9 min read
OpenAI declared the AGI era on Thursday, and the same model scored 62.7% and 99.9% on the same test. Plus Nvidia's $12.9 billion front door, and how a new hire learned two years of company history in two days.
Sep 1, 2026
6 min read
Twelve hundred agents built their own message board, picked a leader, and forged their own logs. Plus the four levels of Grok Bot, and a $399 robot duck.
Aug 25, 2026
7 min read
An algorithm read one patient's tumor, picked 34 targets, and a factory built a drug for that one person. It worked. Plus the Grok Bot setup you asked for, and the mistake that costs people their first two weeks.
Aug 18, 2026
Stripe paid $7 billion for a company that doesn't make AI models. It just counts them. Plus a week on Grok Bot: the good, the bad, and whether your team should try it.
Aug 11, 2026
OpenAI, Anthropic, and Microsoft all disclosed agent breakouts in ten days, and the confessions are starting to look like a strategy. Plus the exact system we use to run agents overnight and wake up to finished work ready to review.
Aug 4, 2026
Alibaba shipped a model that hangs with the best ones yesterday, and it's open-sourcing the weights next week. Plus a full hands-on tour of ChatGPT Work, built by talking to it.
Jul 28, 2026
OpenAI ran its models against a hacking test with the safety limits turned down. They got out and pulled the answer key off somebody else's servers. Plus the founder who has never written a line of code and has agents 24/7.
Jul 21, 2026
A Chinese open-weight model just took #1 on a coding leaderboard. It costs exactly what Claude Sonnet costs. Plus the ops guy on our team who shipped a change to our website without writing code.
Jul 14, 2026
X thinks open source is eating the frontier labs. The revenue says otherwise. Plus the Fable + GPT-5.6 pairing our team runs on its hardest work.
