How conversation flows work
In a scripted chatbot, the flow is literally a diagram: each node is a message or question, and each branch is a button or keyword. Designers map every path. That gives control but breaks when customers answer in ways the diagram didn’t anticipate.
With an AI assistant, the flow becomes goals and rules instead of a diagram: answer questions first, collect only the details a booking needs, confirm before saving, and hand off on certain topics. The assistant chooses the order naturally, so a customer who gives everything in one message isn’t walked through five questions anyway.
Example: an HVAC tune-up booking
An HVAC company’s booking flow for a spring tune-up has four goals: confirm the service area, explain the tune-up price, find a time, and save contact details. A caller who opens with “I’m in Maplewood, can you do my AC tune-up Tuesday?” covers two goals at once, so the assistant quotes the price and offers Tuesday times right away. This scenario is illustrative.
Flows in InsertChat
InsertChat doesn’t use a flow builder. You set the rules: which scheduling actions it may take and whether to confirm appointment details before saving; when to collect contact details; topics that always need a person; promises it must not make; and whether to ask a clarifying question when a request is unclear.
The assistant follows those rules on every channel and asks one question at a time on calls. Preview and test changes before saving. See the HVAC setup.
