Why it matters
AI tools do not read a site the way a person does: they do not see the menu, the design or the scripts. llms.txt hands them a short map — here is the site, here are the main sections, here are the key facts. It is the mirror image of robots.txt, which states what not to read. Expectations should stay modest: Google has said its search systems do not use the file, so it is cheap to provide but not a ranking lever.
How it works and how it is measured
- Location: yourdomain.com/llms.txt — a public address anyone can fetch.
- Contents: a one-paragraph description, the main sections with links, services, offices and contact details; optionally the core facts about the company.
- Freshness: the file has to change as the site changes — a stale llms.txt produces stale answers.
- Multiple markets: on a multilingual site, keeping a file per language makes it clear which section serves which market.
- Limits: some AI tools read it, Google search does not. Treat it as a low-cost addition rather than a budget line.
Frequently asked questions
If we publish llms.txt, will ChatGPT cite us?
No. The file makes a site easier to interpret but guarantees nothing. AI systems still weigh the pages themselves and the external sources that mention you.
How does it differ from robots.txt?
robots.txt is a list of permissions — which crawler may access what. llms.txt is a content map — what the site is about and which page answers which question. The two complement each other.
Should the file be written by hand?
On a small site, yes. On a large one it is safer to generate it from the site itself so that new pages appear in the file automatically rather than being forgotten.