Hand Off to a Human
Open in CMSThis is the last step of Stage 2. You set up escalation: the agent hands a conversation to a human supervisor when it detects a situation it should not handle alone, such as an upset customer, a request for an exception, a critical incident, or a question outside policy.
In the CMS, escalation settings live under Conversation → Escalation in your agent's sidebar.
If the organization has Privacy Mode on, a supervisor handling an escalation cannot see the conversation content or get a real AI summary of it. See Receive and Manage Escalations.
Before you start
- Your agent is published and reachable by users: see Preview and Publish Changes and at least one channel, such as Embed Chat on Your Website.
- You have Owner or Contributor access to the agent.
- At least one person who can act as supervisor is a member of your team. See Manage Who Can Use and Build.
Pages in this step
| Page | What you set up | Order |
|---|---|---|
| Configure Escalation | Turning escalation on, trigger conditions, the escalation response message (with summary language and phone country code), and supervisors | 1. Start here. |
| Receive and Manage Escalations | The supervisor workflow: live updates, claim, review, resolve, or release | 2. Verify that people can act on escalations. |
| Send Escalation Events to a Webhook | Notifying external tools (ticketing, CRM, chat) automatically | 3. Optional, once the first two work. |
This order builds each part on the previous one: rule quality first, human handling second, outbound integration last.
Why escalation matters
Without escalation, an agent can keep answering in loops, miss emotional urgency, or fail to route sensitive requests to the right owner. Escalation gives you:
- Customer protection: upset users reach a real person quickly.
- Operational control: supervisors claim, resolve, or release cases in a clear workflow.
- Risk reduction: sensitive situations go to humans instead of forced AI answers.
- System integration: webhook events can notify external tools.
How escalation works
- The agent detects a trigger condition from your escalation rules.
- An escalation record is created with urgency, reason, topic, and area.
- The supervisor workspace updates in real time.
- A supervisor claims the escalation and reviews the conversation context.
- The supervisor resolves or releases it, and the audit history is updated.
- An optional webhook event is sent to external systems.
The tutorials in this step use one running example: an e-commerce support agent and a customer who is angry because their order was charged twice and demands an urgent refund. The expected result is that the agent escalates to the supervisor queue in real time.
Prepare before you configure
A short checklist makes your first test meaningful:
- Trigger phrases and boundaries: list examples that should escalate (anger, legal risk, refund disputes) and examples that should stay with the agent.
- Urgency levels: decide what counts as low, medium, and high urgency, and the expected response time for each.
- Supervisor coverage: make sure at least one supervisor can claim cases during your operating hours.
- Labels: standardise reason, topic, and area labels so filtering and reporting stay consistent.
- Test scripts: prepare 2 to 3 conversations that should escalate and 2 that should not, and decide who plays supervisor during testing.
- Webhook endpoint (if you plan to use one): have a test endpoint ready to capture payloads.
Next steps
- Start with Configure Escalation.
- When escalation works, Stage 3 begins with Search and Read the Web.