What is llms.txt, and does my site need one?
llms.txt is a plain-text file at the root of your website that summarises what your business does, who it serves and where your key pages live, written for language models rather than browsers. It is an emerging convention rather than a ratified standard, and it takes under an hour to write — low cost, low risk, and not a substitute for having good content.
What it is
A Markdown-formatted text file served at yourdomain.com/llms.txt. It states plainly what you do, who you serve, what you sell, and which pages matter. The idea is that a model retrieving information about your company gets an unambiguous summary instead of having to infer everything from navigation, marketing copy and a footer.
The mental model is robots.txt: a small, conventional file at a predictable location, read by machines, ignored by humans. The difference is that robots.txt grants or denies access, while llms.txt supplies meaning.
Honest status
It is a proposed convention with growing adoption, not a standard any engine has committed to honouring. Nobody should promise you that adding one produces citations. What can be said accurately: it costs very little, it carries no downside, it forces a useful internal exercise in stating plainly what you do, and if adoption continues you are already positioned.
Treat it as cheap insurance, not as a lever.
What to put in it
- A one-paragraph summary of the business, in the first few lines
- Who you serve — industry, size band, geography
- What you make or sell, with real specifics: materials, size ranges, tolerances, certifications, lead times
- What you do not do, which prevents a model confidently miscategorising you
- Commercial basics: how to buy, minimum order, whether you quote custom work
- Links to your most important pages, each with a short description
- Contact details
If your company name is close to another firm's, or your category is adjacent to one you are not in, say so explicitly. One sentence like "we manufacture industrial gaskets; we are not affiliated with [similarly named company]" resolves an ambiguity that could otherwise put someone else's reputation on your name.
What it will not do
- It will not get you cited if you have no substantive content to cite.
- It will not override what other sites say about you.
- It does not replace schema markup — schema is machine-readable data attached to specific pages; llms.txt is a site-level summary. Do both.
- It is not a ranking factor in any documented sense.
Higher-value work in the same hour
If you have limited time, do these first — all are better established than llms.txt:
- Check you are not blocking AI crawlers. Look in
robots.txtforGoogle-ExtendedandApplebot-Extended. Sites block these by accident more often than you would expect. - Mark up your FAQ with FAQPage schema. If you have a visible FAQ and no markup on it, that is free structured data left on the table.
- Publish one real answer page for the question your buyers ask most.
- Get your spec tables out of PDFs and into HTML.
Then write the llms.txt. Ours is at foundvale.com/llms.txt if you want to see the shape of one.
Find out who gets named in your niche
The $500 audit runs 15 real buyer queries across ChatGPT, Perplexity, Gemini and Google AI Overviews, logs who gets cited query by query, and comes back with a 90-day content map and two ready-to-publish articles. Fixed scope, two weeks. The fee is credited to your first retainer month if you continue.
Start the $500 auditAsk a question