Glossary

AI chatbot

Older chatbots followed fixed decision trees: click a button, get a canned answer. AI chatbots use large language models, so visitors type questions in their own words and get answers drawn from the business's own information.

On a business website a good AI chatbot does more than answer. It asks the qualifying questions a salesperson would, books meetings into a calendar and passes the lead, with the whole conversation, to the team.

Quality depends on what it is given to work with. A chatbot trained on clear, current information about your services and prices gives clear answers; one left to guess will make things up.

A worked example

An online furniture shop gets the same questions all day: delivery times, sizes, whether a sofa comes in grey. Say 60 chats a day, and a person needs three minutes for each — three hours of someone's day. An AI chatbot that answers the routine ones instantly, and passes the rest to a person with the conversation attached, gives most of those three hours back and answers the 11 PM shopper too.

When it is worth it, and when it isn't

Worth it: visitors ask the same questions repeatedly, enquiries arrive outside office hours, or leads need qualifying before a call. Less useful: a site with a handful of visitors a day, or a business where every enquiry is a long custom conversation.

Common questions

Is an AI chatbot the same as live chat?

No. Live chat is a person typing replies; an AI chatbot answers automatically. Many businesses combine them: the bot answers first and hands over to a person when needed.

Can the same chatbot work on WhatsApp and Instagram?

Usually, yes. The same assistant and the same information can answer on your website, WhatsApp and Instagram messages, so customers get the same answers everywhere.

How long does setup take?

For a typical small business, days rather than months. Most of the time goes into gathering the answers it should give, not into the technology.