Recent Posts
View all →Ten Months of Solo AI Coding Revealed These Five Structural Failures
AI coding promises 10×. The real story: 10× tactical throughput, roughly 1× strategic judgment. A 5-layer governance framework for what actually breaks when you push AI coding past a single repo and a single sprint — and the order that matters for adopting it.
Your AI Coder Needs a Strict Father, Not Better Prompts
AI coding agents write code and say 'Done' — but Done ≠ working. The fix isn't better prompts. It's an automated QA agent that opens browsers, hits APIs, and queries databases before letting your coder finish.
Stop Trying to Fix AI Sycophancy. Design Around It.
AI coding assistants agree with everything you say. Prompt engineering won't fix it — the flaw is structural. Here's the verification-first workflow I use to ship reliable code anyway.
Side Projects
Mozi
Local-first AI assistant with Brain/Tools/Memory architecture. 20+ built-in skills, multi-interface support, 15+ LLM providers.
TypeScript · Node.js · MITMasetro
Thin orchestration layer over AI coding tools. Artifact-first verification, workspace isolation, 4-tier timeout defense.
TypeScript · 229 tests · MIT