Does llms.txt Actually Get You Cited by AI? (2026)

Adding an llms.txt file will not, on its own, get you cited by ChatGPT, Perplexity, or Google. Google has said plainly that it does not use the file, no major AI provider has committed to it as a ranking signal, and adoption across the web sits at roughly one in ten sites. It is a cheap, harmless thing to add, but it is a nice-to-have, not the lever it is sold as.

That is the short answer. Because the file gets talked about as if it were a cheat code for AI search, here is the longer one: what llms.txt is meant to do, what Google and the crawler data actually say, and where it is genuinely worth your time.

What llms.txt is supposed to do

llms.txt is a proposed standard: a plain Markdown file at the root of your site that points language models at your most important pages, in a clean, easy-to-parse format. The idea is reasonable. Crawling a modern site means wading through navigation, cookie banners, and template clutter, so a curated map of "here is what matters and where it lives" could save a model that work.

The leap people make is from "reasonable idea" to "AI search ranking factor". That leap is where it falls down.

What Google actually says

Google has been unusually direct about this, which is rare for SEO questions. The position, stated repeatedly through 2025, is that Google does not use llms.txt and does not plan to:

• John Mueller, in comments widely reported in June 2025, compared llms.txt to the long-dead keywords meta tag and noted that "no AI system currently uses llms.txt".
• Gary Illyes, at a Search Central event, said Google will not crawl llms.txt files and that normal SEO is what gets you into AI Overviews.
• Google's own guidance on its AI features is that you do not need to create new machine-readable files or markup to appear in generative results. The same pages that rank in Search are the ones eligible to be summarised.

The keywords-meta-tag comparison is the part worth sitting with. That tag also let you hand search engines a tidy list of what your page was about. It died because a signal you fully control, and that costs nothing to game, is worth almost nothing to the engine reading it. llms.txt has the same problem.

What the crawler data shows

Beyond Google, the evidence for any engine leaning on llms.txt is thin. No major provider, OpenAI, Anthropic, Google, Meta, or Mistral, has publicly committed to using it as a signal in their answer surfaces. Independent analyses of large crawl samples have found adoption stuck around 10 percent of sites and direct hits on the file vanishingly rare next to ordinary page crawling.

There is one honest caveat, and it cuts slightly the other way: some AI crawlers have been observed fetching llms.txt when it exists. Fetching a file is not the same as ranking your content by it, but it does mean the flat "no AI system touches it" line is too strong. The fair summary is that a few crawlers will read it, none has said it changes what they cite, and Google has said outright that it does not.

So should you add one?

Yes, with the right expectations. The file is a few lines of Markdown and carries no real downside, so there is little reason not to ship one. Just be clear about what job it is doing:

If your goal is...	Does llms.txt help?
Ranking or being cited in Google AI Overviews	No. Google has said it does not use the file.
Being cited by ChatGPT or Perplexity	Not meaningfully. They cite pages they can crawl and trust.
Feeding documentation to coding agents	Yes. This is the real use case.

That last row is where llms.txt genuinely earns its place. Agentic tools like Cursor, Windsurf, and Claude Code look for an llms.txt when pointed at a documentation site, so it works as a clean machine-readable map for the agent layer. If you run docs, that is a real reason to have one. Improving your AI search citations is not.

This site ships an llms.txt for exactly that modest reason: it is tidy, it costs nothing, and it helps tools that ask for it. It is the last thing on the list, not the first.

What actually gets you cited

If the goal is to show up in AI answers, spend the effort on the things the engines have actually told us they reward:

1. Let the AI crawlers in. Check robots.txt is not blocking GPTBot, ClaudeBot, PerplexityBot, or Google-Extended. A blocked crawler is the most common reason a site is invisible to AI search, and no llms.txt will rescue you from it.
2. Answer first. Lead each section with a direct, self-contained answer the model can lift in one piece.
3. Earn claims with specifics. Sourced numbers and named references get quoted; vague assertions get skipped.
4. Mark it up with schema. Article, FAQPage, and HowTo give engines clean, quotable units.

The full version of that playbook is in how to get cited by ChatGPT and Perplexity. Do those four things and you will be ahead of most sites competing for the same answers, with or without a Markdown file at your root.

The takeaway

llms.txt is not a ranking factor, not a citation shortcut, and not the keywords meta tag reborn, though Google's own people invited that last comparison. It is a small, sensible file with one real job (the agent and documentation layer) and a reputation it has not earned (AI search visibility). Add one if you like, then go and do the work that actually moves the needle.