I called the AI code review tool at work a gimmick for 3 months, then it caught a bug our senior dev missed
I work at a small insurance software company in Columbus and our CTO rolled out an AI code review tool in January. I called it a gimmick in the team meeting and kept merging my pull requests without waiting for its notes. Then in April it flagged a race condition in our claim payout queue that our senior dev with 14 years on the job had already approved twice. We pulled the deploy that night and found it would have double-paid claims in a specific edge case, roughly 3 out of every 10,000. I still think it is overhyped for style stuff, but I read its flags now before I merge anything. Has anyone else had a tool you wrote off actually catch real bugs, or am I just the lucky one?