How I make AI agents reliable
Better AI agents are not mainly about better prompts. They are about better sources of truth, clear procedural instructions, validation, provenance, and correction loops. I treat accuracy as an operating problem, not a model problem.
Read the full note →