Hello! Am new to AI. I am working on an agent that gets vector search results based on the user prompt, and then uses those results chunks to respond to the user. It's a RAG app.
I have noticed the agent tends to overexplain more than what was asked.
On the agent side and after it made a response (but hasnt returned it yet, at least in the UI), i want a way for it to check it's response along with what the user prompt was and rewrite if it overexplained.
Any ideas? (besides handling in the system message)
#Ask GPT to check if response covers user prompt
inner vessel