Set the chatbot's introduction, fallback, schedule, traffic protection, and response rules before enabling it.
Open AI Chatbot from the main navigation. Keep the chatbot paused while you prepare its configuration, then test the visitor experience before enabling autonomous replies for a production workspace.
Write an optional introduction and a fallback that makes clear when the AI could not provide a reliable answer. Configure the AI response schedule when autonomous replies should run only during specific hours. Leave traffic protection on Standard unless testing shows that the workspace needs stricter controls.
Choose Add Rule and give the rule a clear name. The editor separates the decision into two steps: choose when the rule should run, then choose what it should send or collect. Conditions can match keywords, a regular expression, or every message. Responses can send text, offer quick replies, or collect information. Rules are evaluated in priority order and can be reordered from the list.
Use a narrow keyword list or a carefully tested regular expression. The editor rejects an invalid regular expression before saving. Write the complete visitor-facing message and fill the fields required by the response type. A rule can contain up to five quick replies or eight collection fields. Collection keys must be unique, start with a letter, and contain only letters, numbers, dots, underscores, or hyphens.
Avoid overlapping rules that can answer the same question differently.
Save the configuration while the chatbot is paused. Test a known question, an unknown question, and a case with no operator available through a staging widget. Confirm that the fallback is honest and that the conversation remains visible to the support team before enabling the chatbot.
Last updated: Jul 23, 2026
Was this page helpful?