Matt Luttrell's AI experimentation page
This is my personal AI experimentation page: the half-finished ideas, side projects and "what would happen if I tried this" prototypes, living in public. Some of it ships. Some of it gets abandoned in a branch. All of it is here because I learned something building it.
What I am working on
Custom agentic orchestration: multi-agent pipelines, tool palettes, restricted loops, and the plumbing that lets agents hand off work to each other without losing context.
Custom-trained local models: small, task-specific models running on hardware I own, for the workloads where latency, privacy or cost make a hosted API the wrong tool.
Advanced cowork and AI tools: the day-to-day surfaces where a human and an agent share a workspace, from editors and shells to review queues and knowledge bases that survive sessions.
Getting back into music: the songbook is the start of it, with chord charts, a transposer, a tuner and tempo-driven auto-scroll.
The arcade and the rest
I like simple, ad-free games, and it has been easier to just generate my own. Poker, chess, Othello, sudoku and the rest live at the arcade.
Also here: the restoration of a classic Land Rover Defender, and a growing collection of plants.