I build AI systems that do real work.
I write about agents, automation, memory, and workflow — mostly where models meet software and messy business reality.
Recent work
Writing and projects
A mental model for Claude Code (and every other modern agent)
A mental model for how modern AI agents actually work — small enough to hold in your head. Plus the npm packages I built to make it concrete.
AI AgentsOpen SourceTypeScript
MARCO: the loop inside a harness, in code
Companion code to my article on AI agent architecture. A small TypeScript harness that makes the loop-inside-a-harness split literal — readable in an afternoon.
AI AgentsArchitectureTypeScriptOpen Source
AI agentsReference repo
HowAIAgentsWork
A small reference repo for explaining AI agents through one practical split: the inner loop does the work, and the outer loop — the harness — makes it usable.
See the project