How to Improve AI Agent's Recognition of User Answers to Questions in n8n?

Questions
1

Hi everyone, I have a rather complex flow in n8n. The flow receives a message from WhatsApp, and the starting point is determining whether the user has answered the question they were asked. In the AI agent node, I pass in the question I asked and the user’s response. The agent should return one of three response options: answer, general, or var_3.

general is for cases when the user writes a general message or asks a question themselves.

The problem is that sometimes the agent doesn’t understand the response, even though the user clearly answered the question. How can I make sure that the agent definitely understands that the user’s message is an answer to the question asked? Are there any tricks or hacks?

2

There is no definitive way. AIs are about probabilities to begin with. That on top of users sometimes not being able to put even a simple sentence together (in their own language) can be a tough obstacle.

3

This topic was automatically closed 90 days after the last reply. New replies are no longer allowed.