chathq.ai logo
chathq.ai

No fake actions

Visitors do not only ask questions. They ask the assistant to do things: "show me today's picks", "book me in for Tuesday", "cancel my order". A ChatHQ assistant answers from your content; it does not take actions on your systems. The important part is what it does instead.

It never claims to have done something it has not done.

What happens instead

  • It shares the right link — If the thing the visitor wants lives on a page, the assistant points to that page. "Today's picks are published here: …"
  • It gives the steps — If there is a process, the assistant explains it: where to click, what to have ready, how long it takes.
  • It helps when the link does not work — If the visitor comes back saying the page would not open, the assistant offers troubleshooting steps or another way to get there, rather than repeating the same link.
  • It falls back honestly — If nothing in your content covers the request, the assistant says so and points to a person. See Evidence verifier.

What it does not do is say "Done, I've booked that for you" when nothing happened.

Why it matters

A promise the assistant cannot keep becomes a support ticket, a refund request or a lost customer. Visitors who are told clearly what the assistant can and cannot do trust the answers it does give.

Prepare your assistant

  • Put the links in your content — Add the pages visitors most often need (booking, pricing, account, status) to the knowledge base, or list them in your instructions, so the assistant can point to them precisely.
  • Describe the processes — A short "how to cancel an order" document lets the assistant walk a visitor through it step by step.
  • Name the handoff — Tell the assistant in its instructions who handles requests it cannot complete. See Instructions & behavior.

Related

Was this article helpful?

Build AI assistants you can actually rely on

Turn your company's knowledge into assistants your team and customers can rely on.