AI-to-Human Handoff
Handoffs matter most in the situations that tend to be more complex, so it's worth understanding exactly how they work.
How It Works
When the AI agent doesn't have the right answer, it sends the fallback message and flags the conversation for a human. An automatic label, "AI Human Required," appears on the conversation, letting your support team know they're needed.
A human can take control of any conversation, at any point and the moment they do, the AI pauses.
When the AI Steps Back In
On the Guidelines page, you decide when the AI resumes replying after a human has taken over. If a team member responds to a customer, the AI goes quiet; you choose how many hours of no further reply from your team should pass before the AI automatically steps back in.
What happens after a Fallback Message
Below the fallback message setting, there's an option that controls what happens after it's sent:
- Left unticked: the conversation is flagged for a human, but the AI keeps responding normally.
- Ticked: the AI stops responding entirely once it reaches the fallback message in that conversation.
Powered by