Field notes on finding the work worth automating — practical playbooks, honest case studies, and the occasional strong opinion.
How to check AI output quality before a customer sees it: automated pre-checks, a second AI pass, a scoring log, and the three ways review goes wrong.