@Juste Thanks Juste for link. This is good to kick off the interaction with the human e-chat system on intent-action mapping, but then what to do on 2nd message? I.e. if this code hard cods the intent to action mapping:
intents:
- request_human: {"triggers": "action_human_handoff"}
Then how do we keep sending subsequent messages to this action without triggering the predictive next step? As at that moment it’s more of human chat conduit vs predictive engine.
Bonus feature: ok to have the NLU still evaluate what the user set to actually assist agent with user’s intent. Thanks