Problem AI Agent with Post-Tool Responses

Questions
1

Good afternoon.

I am trying to control exactly what response my AI agent gives after a tool is executed. Let me explain: when the customer writes “I want to speak with an advisor,” within the prompt I have the “AndresCera_Hablar_Humano” execute its function, but I don’t want the AI to respond with “You have been transferred to an advisor” when it returns to the AI flow. I don’t want it to say anything at all.

I have tried setting an empty JSON {{}} and nothing worked. I indicated in the general prompt that it should not respond if it receives an empty JSON, but it always generates a response anyway.

I modified the prompt in different ways so that if it receives certain variables it shouldn’t respond, but that didn’t work either.

I placed a node with return ; and it wasn’t possible either - it still gives a response.

image
image537×640 25.7 KB

image
image622×714 49.7 KB

If you could guide me on how to implement this, I would appreciate it.

Information on your n8n setup

  • **n8n version:1.90.2 in VPS
  • Database (default: SQLite):
2

Here’s how the chat works:

The user sends a message, and then the agent responds.

Because of this, the send button is locked until the agent replies.

From the users’s point of view, it’s very confusing if no response is received.

The agent can simply respond by asking what assistance is needed.
Or that the request is forwarded.