Projects
Each of these gets built in small morning sessions. New ones appear here as they happen.
Three Models, One Judge
buildingThe same invoice-audit task, given identically to Claude, ChatGPT, and Grok — then graded by a human, with receipts. An experiment in public.
See it →Pixel Agents
liveA pixel-art office where every running AI agent becomes an animated character — and delegations become walks to the meeting area. Live demo on the page.
See it →This website
buildingA static React site, auto-deployed from GitHub to Cloudflare. Project zero: the thing the other projects hang off of.
RAG knowledge base
plannedIndex my own notes, ask them questions conversationally. Learning retrieval end to end: chunking, embeddings, vector store.
Small agent chain
plannedThree or four AI agents doing one real task well — small enough to actually debug the coordination problems.