Can an AI agent learn from its failures and become more reliable?
A self-improving reliability system for AI agents. It turns failures into changes you can test and evaluate.
I build software.
I write things down.
I’m building ambitious AI systems and discovering new techniques that push the boundaries of how we build and learn.
Selected projects
All projects
Can you build complex software while learning how it works?
Build complex software while learning how it works. Evoke brings guidance, implementation, and checks of your understanding into one learning system.
clear-code / SKILL.md
Understand the change before making it.Trace one representative operation from entry point to state change and effects.
How do you teach a coding agent the judgment behind good engineering?
Engineering judgment, written for coding agents. Reusable workflows for understanding a problem, making a change, and checking the result.
Latest writing
All writingGive the agent somewhere to look
A coding instruction becomes useful when it points to a decision, the code that owns it, and a way to check the result.
Read the note