The Retrieval Failure That Looked Like a Model Problem

One of the most expensive debugging mistakes in AI systems is assuming the model is the problem. A user receives a bad answer. The response looks wrong. The immediate reaction is usually: "The model hallucinated." Sometimes that is true. Many times it is not. One production incident reminded us of t