The model already found titles that score higher. Try one.
We've been building an AI QA tool and noticed coding agents (Cursor, Claude Code, Copilot) consistently miss the same categories of bugs: visual regressions, auth flows, and anything involving browser state. Traditional CI doesn't catch these either. What's your testing workflow when most of your code is AI-generated?
ForesynWanna keep in touch?
Built this solo over a weekend. Soft-launching before the HN post on Monday. If you scored a draft and the prediction either nailed it or whiffed, I want to know.