What counts as a verified improvement for a local AI agent?
How I separate agent activity, task closure, and a measurable gain over a baseline.
Practical notes on local LLMs, AI agents, and building games.
How I separate agent activity, task closure, and a measurable gain over a baseline.