A plain guide to which questions an assistant should answer, and which ones it should hand to a person.
Owners get pitched chatbots that will handle everything. The useful ones do the opposite. They answer the questions you have already written down, and they stop when the next sentence would be a promise.
A good handoff is not a please-call-us line with the conversation deleted. The person who picks it up should see the question in the guest's words, what the assistant already said, a name and a way to reply, and whether the request is urgent.
Read the last answer your assistant would send. If it commits you to a price, a time, or an outcome you have not approved, it should not send that answer. It should collect the request and stop.
The restaurant chat rep on this site follows that split: hours and menu questions get an answer, and catering and large parties are saved for the team. Ben Guard Pediatrics is the other kind of limit. That site publishes documents the team approves. It does not practice medicine in a chat box.
Pick the one promise you are not willing to let software make. Write it down before you turn anything on. The Responsible AI page on this site is the longer version of that rule: humans stay in the loop, answers stay grounded in your business, and a missing fact becomes a handoff instead of a guess.
Let's discuss how we can help you implement these strategies for your business.
Book a Free Consultation