Sep 14, 2026
•
2 min read
Stolen AI logins, AI costs that finally match, sketching instead of prompting, and what Google's release pace should cost you.
Sep 8, 2026
A cheaper model, two commands worth knowing, and a benchmark claim that didn't hold up.
Sep 1, 2026
What that actually looks like when it works, and when it doesn't.
newsletter
+1
Aug 24, 2026
Telling people one by one doesn't work. It has to happen to the group.
+2
Jul 31, 2026
1 min read
The businesses seeing real results fixed the work before they ever pointed AI at it.
Jul 28, 2026
Anthropic just admitted the rulebooks that worked for last year's Claude models is actively hurting this year's models.
Jul 22, 2026
A government ban that took a frontier model offline overnight, and the pricing war that followed two weeks later.
+3
Jul 13, 2026
3 min read
Welcome to the first issue. This month: why 95% of enterprise AI deployments show zero return, and where to start fixing it.
