Let us examine why the AI responded incorrectly and what factors contributed to that response. The Agent Log provides limited information; it only indicates whether the AI retrieved data from the knowledge base.

It's frustrating to constantly wait for the Support Team to investigate issues when we need immediate responses for our clients. All we can tell them is, "We've logged a case," but the Support Team doesn't respond quickly either.

Often, when they do get back to us, their feedback is vague, simply stating that the prompt needs to be amended. However, we lack insight into which prompts will work and which will not. The log is not very helpful if we cannot conduct our own investigations.

What would be helpful is a log that shows what the AI did and how it decided to do it. Something like the attached example.