What is a Intent Detection?

Older chatbot platforms treated intent detection as an explicit build step. You defined a list of intents such as book_appointment or ask_pricing, supplied a set of example phrasings for each, and trained a small classifier to sort incoming messages into one of those buckets. Anything that fell outside the list landed in a catch-all fallback, which is why those systems felt brittle the moment a visitor phrased something unexpectedly.

Modern AI chat widgets built on large language models mostly dissolve the step. The model reads the message directly and responds to what it means, without an intermediate label being assigned. Intent is still being understood, it just is not being reduced to a fixed category first. That is why an AI widget can handle a message containing two intents at once, such as a pricing question with a booking request attached, where a classifier would have had to pick one.

Clerkzo does not ask you to define intents or supply training phrases. The widget interprets each message and retrieves relevant passages from your own crawled pages and uploaded files to answer it. Where a question genuinely needs a person rather than an answer, it hands off instead of forcing the message into a category that does not fit.