Escalation to Human Agent

When the AI can't confidently answer a question, it escalates the conversation to a human agent rather than guessing.

When escalation happens

• AI confidence falls below your threshold (default: 70%)
• Visitor explicitly asks for a human ("speak to a person", "talk to someone")
• Visitor is frustrated (negative sentiment detected)
• Conversation matches an escalation trigger you've set up in Automations

What happens during escalation

1. The AI sends a handoff message: "Let me connect you with a team member."
2. The conversation is flagged as needs agent
3. Available agents are notified (push, email, or both, based on their preferences)
4. The conversation appears in the agents' queue with a summary of what was discussed

Escalation message

Customise the message visitors see when the AI escalates:

In Settings → AI → Escalation message:

I wasn't able to fully answer your question. I'm connecting you with a team member now. They'll be with you shortly.

Outside business hours

If no agents are online when escalation happens:

1. The AI sends your offline message ("Our team is offline, but we'll reply within a few hours")
2. The visitor can leave their email
3. The org owner receives an email notification immediately

Configure offline behavior in Settings → Business Hours.

Escalation analytics

Track escalation rates in Analytics → AI to see:

• Which questions the AI struggles with most (add these to your knowledge base)
• Average time from escalation to first human response
• Escalation rate trend over time