What is llms.txt and does my website need one?

llms.txt is a proposed convention: a Markdown file served at yoursite.com/llms.txt that tells AI systems, in plain text, what your site is, what you offer and where your key pages are. Think of it as robots.txt’s constructive sibling. Robots.txt says what crawlers may not do; llms.txt says “here is what this site is actually about, read this first”.

A minimal one is short: a heading with your name, a one-line summary in a blockquote, a few sections listing your key pages with one-line descriptions. Ours is here if you want a live example.

Does your site need one?

Honest answer: it is one useful part of a bigger job, not a switch that makes AI recommend you. Adoption by the major AI providers is uneven and mostly undocumented, and nobody serious should promise you that the file alone changes your visibility. What it definitely does:

• Gives machines a canonical summary. Extraction from real web pages is messy; navigation, cookie banners and marketing copy blur what you actually do. llms.txt is the clean version, in a format models parse trivially.
• Costs nearly nothing. It is a text file. Writing a good one forces the useful discipline of stating plainly what you do, which most sites have never done anywhere.
• Signals current facts. For a company whose description online is stale or confused, a dated, factual llms.txt is one more piece of fresh evidence pointing the right way.

If your site already answers buyers’ questions in plain, extractable text with structured data, llms.txt is a sensible finishing touch. If your site does none of that, write the llms.txt anyway, because the exercise produces the sentences your homepage was missing.

Writing one that helps

State facts, not slogans. Name the company, say what it does in one sentence a stranger could repeat, list who it serves, link the handful of pages that matter with a line each. Keep it current; a stale llms.txt is worse than none, because it is stale evidence you authored yourself. Skip keywords and superlatives; the reader is a machine assembling a description, and it has seen every trick.

Place it at the site root as /llms.txt, plain text, and check it actually serves. That is the whole deployment.

If you would rather not write it by hand, Sourcive generates one from your actual site content as part of its audit fix pack, alongside the structured data and the buyer-question pages that do the heavier lifting for AI visibility.