The most daunting screen in a DBA’s life isn’t a red dashboard; it’s the blank cursor blinking inside a query editor. You’ve identified the bottleneck. You know the query is performing a massive table scan where it should be seeking. Now you face the tedious, manual reality of remediation: rewriting the SQL, checking syntax, verifying logic, and ensuring you haven’t introduced a regression. In the past year, the temptation to paste a slow query into a public large language model (LLM) has grown. For a seasoned DBA, that’s an obvious nonstarter.