I measure whether AI actually moved the numbers in real codebases (PR throughput, lead time, change-failure rate, review quality) and publish the result, wins and losses alike.

https://lenikirilov.com