llms.txt is a short markdown file at the root of your website that lists your best pages for AI tools to read. It won't get a business site cited in AI answers: Google says its Search ignores it, and the largest studies find no link to citations. Publishing one is harmless and quick, and it pays off mainly for sites with documentation that coding assistants read.
Most posts on this subject argue about rankings. This one asks a narrower question, which is who actually reads the file, because the answer decides whether it's worth your time. It sits inside the wider work of organic search and AI visibility, where a lot of cheaper effort comes first.
What is llms.txt?
llms.txt is a proposal from 2024 for a markdown file at /llms.txt that gives AI tools a short, curated map of a site: a name, a summary and links to the pages worth reading. It saves an agent from digging through menus and scripts. It's a convention, and Ahrefs notes that no major AI platform has committed to reading it.
The format is small. According to the proposal at llmstxt.org, the file has these parts, in this order:
- A single H1 with the name of the site. This is the only required part.
- A short summary in a blockquote.
- Optional paragraphs or lists with more background.
- Sections under H2 headings, each a list of links with an optional note after a colon.
The newer version of the proposal also suggests that pages worth reading offer a clean markdown copy, at the same address with .md added. The idea is that an agent reads the map first, then fetches only the pages it needs. The file is meant to stay small enough to fit in an AI tool's working memory.
It isn't a robots.txt for AI. It controls nothing and blocks nothing. If you want to decide which bots may crawl your site, that's a different file, covered in our guide to blocking AI training bots while allowing AI search.
Does llms.txt help your site show up in AI answers?
The evidence says no. Google states that Google Search doesn't use llms.txt. A study of 13,911 domains found no measurable visibility lift once site authority was controlled for. And Ahrefs found that 97% of the valid files it checked received no requests at all in May 2026.
Start with Google, because it's the one platform that has said so in writing. Its guide to optimizing for generative AI features lists llms.txt files among the things you can ignore. You don't need new machine-readable files, AI text files or markdown to appear in Google Search, including its generative AI features, "as Google Search itself doesn't use them." It adds that creating the files is fine and "will neither harm nor help" your visibility or rankings in Google Search.
Then the studies. Search Atlas compared llms.txt files on 13,911 domains with 83,387 visibility records across six AI platforms: Copilot, Gemini, Google AI Mode, Grok, OpenAI and Perplexity. The median difference between sites with and without the file was 0.42 points on a 0 to 100 scale. Better-written files did seem to go with better visibility, but that advantage largely disappeared when sites of similar authority were compared. Its reading: stronger sites tend to publish tidier files, and the file itself adds no measurable lift.
Ahrefs looked at the other side, whether anything fetches the file. Across 137,210 domains with traffic in May 2026, about 28% published a valid llms.txt, and 97% of those files had zero requests that month. The 3% that were read, around 1,100 domains, accounted for all the traffic measured. Ahrefs also found no requests from AI bots for llms.txt files that don't exist, so the bots aren't going looking. Two caveats from Ahrefs itself: its customers skew technical, so 28% is an upper bound, and it measured only the index file.
None of this proves that no AI product anywhere uses the file. It does mean that nobody has shown a citation gain from publishing one, and a plan that depends on it is a gamble.
Who actually reads llms.txt?
Mostly bots built for developers and, increasingly, auditing tools. Of the requests Ahrefs recorded, 96% came from bots, and about 19.5% came from named AI tools, with GPTBot first and Claude-Code second. Another 12% came from tools that study or check llms.txt files themselves.
That points to the one group where the file makes sense. Coding agents fetch a library's documentation to get an API call right, and a tidy index of that documentation saves them effort. The proposal itself says llms.txt files are used most heavily for software documentation, and notes that OpenAI, Anthropic and Gemini publish them for their own developer docs. Ahrefs reports that Google's John Mueller has described the file as most useful for AI coding agents.
A brochure site for a dentist or a design studio has no API for an agent to call. Its content is a handful of pages, and any crawler can already read them.
Should a business website publish one anyway?
Publish one if your site has documentation, a help centre or product guides that developers or AI assistants consume. For a typical service or ecommerce site, skip it for now, or treat it as a small job with no expected return. Spend the effort on pages that answer questions clearly.
A simple way to decide:
- Software with docs or an API: yes. This is the use the file was designed for.
- A publisher or a content-heavy site that depends on Google: no. Google has told you it ignores the file.
- A service business or small shop: optional. It costs little and does no harm, but nothing in the data suggests it moves citations.
If you do publish one, keep it honest and small. Link only to pages that exist and matter. Write the summary as plain facts about what the business does and where. Update it when the site changes, because an outdated map is worse than none. Check that /llms.txt returns a normal page, not an error. And expect silence: in Ahrefs' data, most files were never requested.
If the goal is to be quoted by AI assistants, the work that counts is elsewhere. Make sure the right crawlers can reach you, answer real questions in plain sentences, and say things other pages don't. We covered that in how to get your business cited in AI answers, and the traffic side in what Google AI Overviews change for your traffic.
If you'd like onlyspell to look at where your site's AI visibility effort should go first, talk to us about your project.
Frequently asked questions
Is llms.txt the same as robots.txt?
No. robots.txt tells compliant crawlers which parts of a site they may fetch. llms.txt is a markdown map of your best pages meant to help AI tools that choose to read it. It blocks nothing and grants nothing, and no platform is required to use it.
Does Google use llms.txt?
Google says Google Search doesn't use it, including its generative AI features, and that publishing one will neither help nor harm your rankings. Google adds that it's fine to create the file for other services or systems that use it.
Will ChatGPT or Claude cite my site more if I add llms.txt?
There's no evidence of that. Search Atlas found no measurable visibility lift across six AI platforms after controlling for site authority, and Ahrefs found most published files receive no requests. Some AI bots, such as GPTBot and Claude-Code, do fetch files that exist, but fetching isn't citing.
Sources
If you still need additional help in developing and designing your website, our magic is here to help. Contact us today, and our wizards will help you immediately with your digital marketing needs!
