Home
Courses
Services
Resources
About
Aug 20, 2026
•
3 min read
A model 1,000x smaller than the frontier just undercut GPT-5.6 by 11x. There is a catch, and it is not the one you think.
Aug 5, 2026
AND: what happened last time mathematicians called their work useless.
Jul 26, 2026
4 min read
PLUS: YC's new list of what to build, and why the bottleneck stopped being intelligence.
Jun 18, 2026
2 min read
What really happened.
Jun 10, 2026
It found bugs I signed off as clean. Benchmarks, pricing, what to test before June 22.
May 27, 2026
you can build pretty much anything - the new skill is not syntax. It is asking clearly.
Apr 30, 2026
I expected the bottlenecks to be data and integrations. The real constraint is more human than that.
Apr 27, 2026
Frontier AI got commoditized this week. The question is no longer “how good are the models?” It’s “what now?”
Apr 16, 2026
A 17-year-old FreeBSD RCE. A 27-year-old OpenBSD bug. Found autonomously. Plus Opus 4.7 (RELEASED TODAY!) GPT-5.5, and what to do this week.
Apr 7, 2026
The practitioner guide I wish existed on day one.
Apr 6, 2026
OpenClaw is the first open-source agent stack I've seen that feels less like a demo and more like infrastructure.
Mar 30, 2026
The same signal, plus the playbooks and tools that actually help you succeed.
