Escalation
The Escalation agent helps customer experience teams with detects VIP, outage, or legal-sensitive threads. It runs with EU AI Act–ready guardrails: your integrations stay connected, sensitive actions pause for human approval, and every output stays traceable for audit.
Capabilities
- Detects VIP, outage, or legal-sensitive threads
- Bundles timeline, sentiment, and assets
- Routes to on-call with severity scoring
- Suggests interim customer messaging
How it works
Listen to the customer
Tickets, chats, and feedback streams feed the agent with full thread context.
Draft grounded responses
The Escalation agent cites macros and articles while respecting tone and policy.
Close or escalate
Resolved threads auto-close; edge cases bundle evidence for specialists.
Required integrations & tools
Connects to: Zendesk, PagerDuty / Opsgenie, Slack.
What a run costs
Typical cost: ~6–12 tokens per run
You see an estimate before the run starts and the real amount when it finishes, drawn from your balance. Admins can cap what this agent — or the person using it — may spend in a month.
You stay in control
- This template is a starting point: you get your own copy and can change its instructions, model, tools and knowledge without affecting anyone else.
- Anything that changes the outside world asks first — and you can edit what it proposed before approving. Sending mail, deleting and paying always ask again.
- Run it from the chat, on a schedule, or from an external event via a webhook — the history, cost and output of every run are kept.
Related agents
More support templates you can deploy the same way.