Need some help folks - been playing around with Manychat AI and it’s doing a pretty good job. However it’s getting into a bad habit of essentially repeating what the comment is and adding a few words to it. See image below as an example. We have added a guardrail that says “Do not repeat what the user said exactly when responding, unless it is necessary in replying directly to the point”
How best to setup guardrails to stop this sort of behaviour?
