Short answer: llms.txt is a plain Markdown file at the root of a website that says what the site is, what it offers and where its best pages live, so AI assistants can read the essentials in one pass. It was proposed in September 2024, no major AI engine has publicly committed to reading it, and Google has said its search does not use it. It also takes about twenty minutes and carries zero risk — which is why we keep one ourselves at morneto.com/llms.txt and will not sell you a package for it.
Last updated August 30, 2026. Every claim about adoption links to the source that made it, and the only site whose file we describe from the inside is our own. No invented case studies.
Few SEO topics have produced this much confident advice on this little evidence. Consequently, half the internet says llms.txt is the new robots.txt and the other half calls it useless. Both camps overstate it. This guide explains what the file actually is, who reads it today, what ours contains after a month of maintaining it, and when paying someone for it should worry you.
What is llms.txt and how does the file work?
The idea comes from Jeremy Howard of Answer.AI, who published the llms.txt proposal in September 2024. The problem it addresses is real: a language model answering a question about your business has a limited context window, and a modern web page spends most of its bytes on navigation, scripts and styling rather than facts.
The proposed fix is a single Markdown file at your root — yourdomain.com/llms.txt — with a fixed shape:
- An H1 with the site or business name. The only required part.
- A blockquote summary saying in a few lines what the site is and does.
- Sections of links to your most useful pages, each with a one-line description.
- Optionally an llms-full.txt, which inlines the full content instead of linking to it.
Because it is Markdown rather than HTML, a model can read it without rendering anything. Meanwhile, the contrast with robots.txt matters: robots.txt tells crawlers what they may read, while llms.txt suggests what they should read. One is permission, the other is a menu.
Is llms.txt actually used by AI tools?
Here is the sentence most guides bury: no major AI engine has publicly committed to using llms.txt. OpenAI, Anthropic, Perplexity and Google have announced nothing of the kind. Google went further: its search team has said plainly that Google does not use the file, and its own documentation on AI features and your website states that AI Overviews need no special files and no special markup — a page needs to be indexed and snippet-eligible, nothing more.
At the same time, dismissing it completely misses what is happening around it. Documentation platforms such as Mintlify and GitBook now generate the file by default, Yoast added an llms.txt feature to WordPress, and Chrome’s own Lighthouse documentation describes it as an emerging convention for giving language models a machine-readable summary. A convention this cheap does not need official adoption to be worth twenty minutes.
So the honest status in 2026 is: unproven, inexpensive, and spreading. Treat any stronger claim — in either direction — as marketing. The file is also one small extra, not a strategy. Our breakdown of where GEO ends and classic SEO begins shows where it sits in the wider work.
What ours looks like, and why we keep it
We would rather show one real file than quote ten opinions, so here is the live file: morneto.com/llms.txt. We write it by hand, not generated, and it contains the business summary, our published prices, and fifteen questions with direct answers — the questions a buyer actually asks an AI assistant, such as what a website should cost in New York and who owns the domain afterwards.
Two rules keep it useful rather than decorative. First, it only says things that are also true on the site itself, because a summary that drifts from the pages it summarizes is worse than no summary. Second, it gets updated whenever a price or an important page changes, on the same day. A stale llms.txt quietly feeds AI tools last month’s facts.
For honesty’s sake: our server logs show steady traffic from AI crawlers — in one measured 24-hour window Googlebot fetched 150 pages here, Bingbot 32, Meta’s crawler 34, Amazonbot 30, PerplexityBot 20 and GPTBot 16. What we cannot prove is that any of them read the llms.txt file specifically, and we will not pretend otherwise. The file is a bet with a tiny stake, not a measured channel. The measured channels — being indexed, being quotable, being consistent across the web — are covered in our guide to getting found by AI, and the reason Bing matters so much for AI answers has its own guide.
How to create the file in twenty minutes
- Open a plain text editor and start with one H1 line:
# Your Business Name. - Add a blockquote summary of two or three lines: what you do, for whom, where, and what it costs if you publish prices.
- List your best pages in Markdown link format with a one-line description each — services, prices, and the guides that answer real questions. Ten strong links beat fifty thin ones.
- Answer the questions people actually ask an assistant. This is the part almost everyone skips, and in our view it is the most valuable: short, factual Q&A beats a bare link list.
- Upload it to your site root so it loads at yourdomain.com/llms.txt, then check it in a browser.
- Put a reminder in your calendar to update it when prices or pages change. An accurate file is the whole point.
Notably, nothing in that list requires a developer, a plugin or a budget. If someone quotes you a monthly fee for “llms.txt optimization,” you now know what the work actually is.
Worth doing, or a sales pitch?
Our verdict, as a studio that both builds sites and maintains one of these files: do it once, keep it accurate, and never pay for it. The upside is uncertain but real — a clean, factual summary that any current or future AI tool can read at a known address. The downside is twenty minutes. Bets with that shape are worth taking. Even so, the file is one small part of a full AI visibility setup, and it is the smallest part of it.
The warning sign is when the file gets sold as a ranking product. Traffic lost to AI answers is a genuine problem, and we wrote about the measured mechanics of it in why website traffic drops in the AI era — but no honest provider can promise that llms.txt wins any of that traffic back, because no engine has promised to read it. Anyone claiming otherwise is selling certainty they do not have.
Finally, if you are not sure whether your site’s real problem is AI answers, indexing or something more ordinary, that is exactly what an audit is for. Ours is free, it names your pages and your problems, and you keep it either way.
Common questions about llms.txt
Is llms.txt actually used by AI engines?
No engine has publicly committed to it. Google has said its search does not use the file, and OpenAI, Anthropic and Perplexity have announced nothing. Documentation tools generate it anyway, and AI crawlers can fetch it like any URL — so it is an unproven, low-cost bet rather than a measured channel.
Is llms.txt a Google ranking factor?
No. Google’s documentation says AI Overviews and AI Mode need no special files or markup — a page needs to be indexed and eligible for snippets. Anyone selling llms.txt as a ranking product is selling something Google says does not exist.
Is llms.txt mandatory?
No. Nothing breaks without it. It is an optional convention, like a well-written About page for machines. Skip it entirely and your SEO is unaffected.
What is the difference between llms.txt and robots.txt?
robots.txt is about permission: it tells crawlers which parts of the site they may or may not access. llms.txt is about guidance: it suggests which pages are worth reading and summarizes the facts. They live side by side at the site root and do different jobs.
How do I check whether a website has one?
Type the domain followed by /llms.txt in your browser, for example morneto.com/llms.txt. If the file exists it loads as plain text; if not, you get a 404. Most sites, including most agencies advising on it, do not have one.
