Escalations
Escalations are how the AI hands a conversation to a human agent. They keep the customer experience smooth when a request is sensitive, urgent, or needs human judgment.
When escalations happen
Common escalation triggers include:
- The customer explicitly asks for a human.
- A Response of type Escalation matches the request.
- A Flow decides to hand off based on data or rules.
Use Escalation Responses when you know certain topics always need a person (for example, billing disputes or policy exceptions).
Configure escalation behavior
Open Agents -> Settings -> Escalations to choose:
- Escalation channel: keep the conversation in the same channel or move it to email.
- Customer wait time source: whether to show a wait time based on your SLA settings.
If you use a supported ticketing integration, additional wait time options may appear (for example, queue size for Dixa).
For advanced control, you can create a Flow named escalate_conversation and
use it to override the default escalation behavior.
What happens during a handoff
When an escalation triggers, Applied will:
- Mark the conversation as escalated.
- Create a ticket in the Inbox.
- Preserve the full conversation history for the human agent.
The conversation remains visible in the Inbox so the human agent can reply with full context.
Tips for clean handoffs
- Ask for key details before escalation (order ID, account email, screenshots).
- Keep escalation responses short and clear so the customer knows what happens next.
- Review escalated conversations weekly to see if new knowledge or workflows could reduce future handoffs.
Next steps
- Review ticket workflows in Tickets.
- Set SLAs in Settings -> Workspace -> Inbox for wait time messaging.