Writing, talks, and working notes.
What we're learning while building runtime control for agents, starting with Richard's essays.
May 6, 2026
The Architecture of Agent Behavior: Why the System Prompt Is Not Enough
Governing AI agents at runtime requires more than instructions. It requires infrastructure.
Read the essay →Apr 20, 2026
The Vercel breach is not a hacking story. It is an AI governance story.
On April 19, Vercel disclosed a security incident.
Read the essay →Apr 18, 2026
We Built AI That Can Act. We Forgot to Control It.
The Control Problem Nobody in AI Is Actually Solving
Read the essay →