How Our llms.txt for AI Chatbots Works

A plain text file is published at the documented well-known path describing the business in the form a language model can consume without rendering the site: what the organisation does, what it does not do, where it operates, how to make contact, and links to the canonical pages for each topic. Content is written as complete factual statements rather than marketing phrasing, because a model quoting a sentence out of context should still produce something true. A companion full-text file carries the longer body content for the same pages. Nothing appears in either file that is not also true and visible on the site, so the two can never contradict each other. Explicit boundaries are stated, since preventing a wrong recommendation matters as much as earning a right one: services not offered, areas not served and matters referred elsewhere are named directly. Files are served as text/plain with a correct content type, referenced from robots.txt, and re-checked on the served URL after deployment. They are regenerated when the underlying pages change, because a stale description is worse than none.