AI & Search
llms.txt Explained: Should Your Site Have One?
A plain-language guide to llms.txt — what the file is, who actually reads it, and whether it is worth twenty minutes of your afternoon.
· 6 min read · Flowmatix
The short answer
llms.txt is a plain-text file at the root of your site that summarises what your business does and links to your most important pages. No major AI company has committed to reading it, so it is not a ranking factor. It costs twenty minutes, and the discipline of writing one usually improves the pages it points at.
If you have read anything about AI search in the last year, you have probably been told your site needs an llms.txt file. The advice is usually delivered with more confidence than the evidence supports, so here is the honest version.
What llms.txt actually is
It is a markdown file served at yoursite.com/llms.txt. It describes what your organisation does in plain language and links to the pages a machine should read first — pricing, services, contact, key articles. Think of it as a reception desk for software: here is who we are, here is where everything lives.
The format was proposed by analogy with robots.txt and sitemap.xml, both of which are honoured by the crawlers they were written for. That analogy is doing a lot of persuasive work, and it is where most of the confusion starts.
Who reads it
Honestly: nobody has committed to. As of writing, no major AI company has published a statement saying its crawler consumes llms.txt as a ranking or retrieval signal. Some developer tools and smaller agents do read it. Plenty of sites publish one and see nothing measurable.
Anyone telling you it is required is describing a hope, not a mechanism. Treat that as a useful filter on the rest of their SEO advice.
Why we still publish one
Three reasons, none of them magic.
- 1The cost is twenty minutes and one file. The downside risk is zero — it cannot hurt rankings, confuse a crawler, or slow a page.
- 2Writing it forces you to state, in one page, what you sell and what it costs. Most businesses discover that their site never says this plainly anywhere. Fixing that helps human buyers immediately, which is a return that does not depend on any AI company adopting anything.
- 3If adoption does arrive, the file is already there and already accurate. Optionality is cheap here.
What to put in it
Keep it short and specific. A model summarising your business does better with sentences than with adjectives.
- One paragraph on what the business does, for whom, and where.
- Real prices, or at least real starting prices. Bespoke tells a machine nothing.
- Links to pricing, services, contact and your best three or four articles, each with a one-line description.
- Anything commonly misunderstood about you — what you do not do is as useful as what you do.
What to do before you bother
None of this matters if the crawlers cannot reach you. Check your robots.txt allows GPTBot, OAI-SearchBot, PerplexityBot, ClaudeBot and Google-Extended. A surprising number of sites block them by accident, usually through an over-broad rule or a security plugin's defaults. That single check is worth more than every llms.txt on the internet.
The verdict
Publish one. Spend twenty minutes, not twenty hours. Then go and do the things that demonstrably work: answer the question in the first forty words of every page, publish real numbers, write headings the way people ask questions, and earn mentions on sites the models already trust.
Frequently asked questions
Is llms.txt a Google ranking factor?
No. Google has not stated that it reads llms.txt, and no major AI company has committed to using it as a retrieval or ranking signal. It is a proposed convention with partial adoption, not a confirmed mechanism.
Where does llms.txt go?
At the root of your domain, served as plain text or markdown: yoursite.com/llms.txt. It sits alongside robots.txt and sitemap.xml and needs no configuration beyond being reachable without authentication.
What is the difference between llms.txt and robots.txt?
robots.txt tells crawlers what they may access and is universally honoured. llms.txt tries to tell AI systems what your site means and is honoured by almost nobody yet. One is a rule, the other is a suggestion.