What is llms.txt (and do you need one)?
llms.txt is a plain markdown file you publish at yoursite.com/llms.txt that lists the pages you most want AI assistants and agents to read first. Think of it as a short, curated map of your site: your best answer pages, grouped and linked, written for a machine. It is useful, low-cost hygiene, but be clear-eyed: adoption by the big engines is still partial and unconfirmed, so it is a nice-to-have, not the thing that wins you the recommendation.
What llms.txt actually is
The file lives at the root of your domain, alongside robots.txt and your sitemap. It is just markdown: a heading with your brand name, a one-line summary of what you do, then a few sections of links to the pages that best answer your buyers' questions. The idea is to hand an AI a short, curated reading list instead of making it guess which of your hundreds of pages matter.
It is not a magic switch and it is not a way to block or grant access, that is what robots.txt does. llms.txt only suggests priority. An engine still has to choose to fetch it, and most of the big ones have not publicly committed to reading it yet.
What a good llms.txt looks like
Keep it short and point each link at a page that genuinely answers a question, not at your homepage. A simple structure for a fictional brand looks like this:
- # Acme Pumps, your brand name as the top heading
- > Industrial water pumps for food and beverage plants, a one-line summary of what you do, on a quote line
- ## Start here, links to your one or two best overview pages, e.g. Best pumps for food-grade lines and How to choose a sanitary pump
- ## Guides, links to the pages that answer specific buying questions a customer would ask an assistant
- ## Trust, links to your About page, customer proof, or specs that back up your claims
Each line under a section is a normal markdown link with a short description. That is the whole file. There is no hidden syntax to learn, if you can write a list of links, you can write an llms.txt.
Be honest: it is hygiene, not the lever
Two files matter far more than llms.txt, and you should get them right first. A sitemap tells engines every page you have, and a robots.txt that allows the AI crawlers, GPTBot, ClaudeBot, PerplexityBot, Google-Extended, decides whether they can read you at all. If those crawlers are blocked, nothing else you do counts, llms.txt included.
Order of operations: make sure your robots.txt allows the AI crawlers, publish a clean sitemap, then add llms.txt as the finishing touch. Doing it the other way round is polishing a door that's still locked.
The real work of getting recommended is upstream of all three: publishing the page that directly answers your buyer's question, in the format the question implies, and stating your facts in structured data (schema.org JSON-LD) a model can read. llms.txt helps an engine find those pages faster; it cannot make a thin page worth recommending.
Who should bother, and how to publish one
It is worth ten minutes for almost anyone, because the cost is near zero and the downside is none. It earns its place soonest if you have a large or messy site where your best answer pages are buried, or if you are early enough that being a clean, well-mapped option helps an engine pick you out. If you are a five-page site, the file matters less, your pages are already easy to find.
- 1
Pick your priority pages
Choose the handful of pages that best answer the questions your buyers actually ask an assistant, not your cleverest marketing pages, your most useful ones.
- 2
Write the file
Add your brand heading, a one-line summary, and two or three sections of links with short descriptions. Save it as plain markdown.
- 3
Publish it at the root
Upload it so it loads at yoursite.com/llms.txt, served as plain text or markdown. Re-check it whenever your priority pages change.
MentionLM does this for you. When it scans your site it checks whether you already have an llms.txt and, if you don't, generates a complete, paste-ready one, your brand, summary and the sections that map to the exact questions you're losing, alongside the schema and content briefs. It even bundles a one-paste setup prompt for Cursor or Claude Code so a developer can publish the whole set in one go.
Frequently asked questions
Do the big AI engines actually read llms.txt?
Some tools and agents do, but the major answer engines, ChatGPT, Claude, Gemini, Perplexity, have not all publicly confirmed they use it, and support is partial and changing. Treat it as a low-cost hint that may help, not a guarantee that an engine will follow it.
Is llms.txt the same as robots.txt?
No. robots.txt controls whether crawlers are allowed to read your site at all, it is load-bearing and you must get it right. llms.txt only suggests which pages to prioritise once a crawler is already allowed in. You need the robots.txt sorted first.
Will adding llms.txt get me recommended by ChatGPT?
Not on its own. It can help an engine find your best pages faster, but the recommendation comes from having a page that genuinely answers the buyer's question, readable structured data, and crawlers that can reach you. llms.txt is the finishing touch, not the cause.
Where exactly do I put the file?
At the root of your domain so it loads at yoursite.com/llms.txt, served as plain text or markdown, the same place your robots.txt lives. If it only opens at a deeper path, engines won't find it where they expect.
How do I know if AI already recommends me?
Ask the engines the questions your buyers ask and read the answers. MentionLM scans ChatGPT, Claude, Gemini and Perplexity in about ten seconds, shows your AI visibility score, names the rivals AI recommends instead of you, and hands back the fixes, including a ready-to-publish llms.txt.
See where AI recommends a rival instead of you.
Scan your site free →