Assistance Requests
Artificial Intelligence is great at repetitive tasks, but there are situations that require empathy, judgment, or human authorization.
The Assistance module lets you define the "escalation rules." In other words, you teach your agent to recognize when it should step back and ask your team for help.
How does it work?
When the agent detects that the conversation matches the criteria you defined:
- It sends a Waiting Message to the customer (e.g., "I'm transferring you to a specialist").
- It stops responding automatically (it pauses).
- The conversation appears in the inbox with the Assistance status.
- It sends a notification to your team so a human can take over.
Step-by-Step Configuration
Enable the Option
Inside the Agent, go to the Assistance tab. Turn on the "Assistance requests" switch.
Define the Criteria (The "When")
In the main text box, describe in natural language the situations that require a human. Be specific.
Configure the Waiting Message (The "What")
Write the message the customer will see right before being transferred. This sets expectations about the wait.
Example:
I understand this needs special attention. I'm connecting you with a human advisor — please hold on a moment while they review your case.
By default, the agent will always request assistance in two situations: when the customer explicitly asks to talk to a person, and when the agent runs into doubts or topics outside its knowledge base. This guarantees that customers never feel trapped with a bot and that complex cases always reach your team.
Taking Back Control
Once the human resolves the issue in the Inbox, they can hand control back to the Agent. The agent will pick the conversation up right where it left off, ready to keep helping.