Escalation
What happens when the agent hands a conversation to a human — the yellow note, the silent agent, and how to pick it up.
When a customer is clearly unhappy or explicitly asks for a human, the conversation is escalated: a yellow note appears and the agent stops replying to that conversation. It is waiting for you.
Why a conversation escalates
The agent hands over in three situations:
- The request goes beyond what it can do. A ticket in the AI Limitation category is created.
- The customer explicitly asks to speak to a human.
- The situation matches one of your own escalation rules, set in the AI Center.
Pick up an escalated conversation
- Open the conversation from the inbox — the yellow note tells you what went wrong.
- Reply yourself. The agent stays silent on this conversation, so there is no risk of it talking over you.
- Resolve the matching ticket in Support once the customer is taken care of.
Review escalations weekly
Filter tickets on the AI Limitation category and read the recent ones. Each is telling you something your agent is missing — a policy, a piece of product information, an instruction. Fill the gap in the AI Center and that class of escalation stops recurring.
FAQ and troubleshooting
The agent is silent and I do not know why
Look for a yellow escalation note at the top of the conversation. If there is none, check AI Autopilot on the conversation and the global Active switch in the AI Center.
Does the agent resume after I reply?
The conversation stays with you once escalated. Turn AI Autopilot back on for that conversation when the issue is settled and you want the agent to carry on.
Can I escalate a conversation myself?
Turn AI Autopilot off on it. The effect is the same in practice: the agent stops and the conversation is yours.
What's next?
Escalations arrive as tickets. Here is how that queue works.