Writing about the continual learning loop

Notes on agent harness improvement, intent discovery, and frustration root cause attribution from the team building the continual learning layer for AI agents.