Support Instant Search

Instant answers from the knowledge file. No model, no spend.

This is the lane the widget searches while the visitor is still typing. Every hit is the stored wording, quoted, so a plan fact reaches the visitor exactly as the pricing page states it - and a visitor who finds their answer here has spent nothing and still has all five free messages for the question that actually needs the agent.

It is GET on purpose: the response depends only on the query and the deployed artifact, so a browser, our CDN and a back button may all cache it. Nothing here is per-visitor.