How to configure an AI agent in the chatbot
Link an AI agent to the chatbot flow, collect user attributes, and set function exit conditions.
Add the agent to the canvas
- In the chatbot builder, drag the Agent box onto the canvas.
- Rename the agent to make it easier to identify in the project.
- In the Linked agent field, select the agent created under Modules.
Collect user data (attributes)
- Add custom attributes (ID number, email, address) or include a direct instruction in the agent, for example: "Collect user email before starting the conversation."
Configure memory and interaction
- Enable memory so the agent remembers previous attendances. Define how many previous attendances it should consider.
- Also configure: maximum wait time between interactions, maximum number of interactions, and token limit per attendance.
These parameters control operating costs since each model charges per input and output token.
Add a function exit
- Click Add function exit.
- Fill in the fields:
- Exit name (e.g., "Confirm ID") - Operator type: AND (all conditions) or OR (one of several) - Validate by: typically "Last message" - Message field: Text or Type - Validation type: equal, different, empty, and others - Expected value (e.g., "Yes", an ID number)
- Click Save.