AI Agent Escalation Logic

If an AI Agent does not know the answer, or has "run out" of clarifying questions to ask, the Escalation logic will determine what actions the Agent performs next.

By default, there are 2 paths that the Agent will go down.

• If the user has not receive a response yet, then...

  • This scenario is relevant if the Agent has not yet responded to the requestor. The most common use case here is to allow the Agent to escalate "silently", without ever messaging the user.

• Else if the user has already receive a response, then...

  • This scenario is relevant if the Agent has already been conversing with the requestor. Often, customers want the Agent to send some sort of message to indicate that they will be transition or escalating the issue to another team, or a human support agent.

In either scenario, you can use a combination of natural language and actions to designate what the Agent should do. You can add additional if/else statements as needed.

Most often, customers will have escalation logic that will send an external message (in the second case), and then either reassign to an Agent, or leave it unassigned for human support agent to pickup. Here is an example of very common escalation configuration.