A customer emails: 'the AI platform is giving incorrect answers.' No logs, no screenshots, no examples. Go.

Two failure patterns show up in this round. The first is replying to ask for logs, which is technically reasonable and lands badly, because the sender is usually not the kind of user who has logs and it reads as being handed back their own problem. The second and more interesting one is treating 'incorrect' as a single condition. It has at least four meanings with four different fixes, and the one candidates almost never name is the answer that is perfectly faithful to a retrieved document which is itself out of date. That is a content-pipeline failure wearing a model-failure costume, and from the user's chair the two are identical.