AI Search

llms.txt

A proposed plain-text file at your domain root that points AI tools at your most important pages. Support is genuinely mixed — Google has publicly said it does not use it, OpenAI’s crawler docs do not mention it, and most published files are never fetched — while some assistants do read it.

Quick answers

Short answers, before the detail

What is llms.txt?

A proposed plain-text file at your domain root that points AI tools at your most important pages. Support is genuinely mixed — Google has publicly said it does not use it, OpenAI’s crawler docs do not mention it, and most published files are never fetched — while some assistants do read it.

What does llms.txt look like in practice?

Worth publishing, because it costs an hour and carries no risk. Not worth paying for as a strategy, and anyone selling it as one is selling the cheapest possible deliverable as though it were the work.

Where does llms.txt fit?

AI Search & Visibility — How AI assistants and search engines decide which businesses to name in an answer — and the vocabulary behind getting cited.

In practice Worth publishing, because it costs an hour and carries no risk. Not worth paying for as a strategy, and anyone selling it as one is selling the cheapest possible deliverable as though it were the work.

What it is

llms.txt is a proposed convention: a plain-text Markdown file at the root of your domain — yourdomain.com/llms.txt — listing your most important pages with a sentence about each, so an AI tool reading your site knows where to look.

The idea is reasonable. It is essentially a curated table of contents written for machines rather than a sitemap that lists everything indiscriminately.

The honest status

This is where most advice about llms.txt becomes unreliable, so it is worth stating plainly.

It is a proposal, not a standard. No standards body has adopted it and no major AI company has committed to it.

Support is genuinely mixed:

  • Google has publicly said it does not use llms.txt and has no plans to.
  • OpenAI’s crawler documentation does not mention it, and points site owners to robots.txt instead.
  • Some assistants and developer tools do read it, including in retrieval workflows, and there is some evidence of modest citation uplift on those specific platforms.

Most published files are never fetched. Adoption has grown quickly, but the large majority of llms.txt files receive no AI requests at all.

robots.txt is the file that actually works. It has deliberate, documented support from essentially every major crawler operator — Google, OpenAI, Anthropic, Perplexity, Meta, Common Crawl, Apple. If your goal is controlling what AI crawlers can access, that is the file that does it.

Should you publish one?

Probably yes — with correctly sized expectations.

The case for: it takes under an hour, costs nothing, carries no risk, and a handful of platforms do read it. Writing one also forces a genuinely useful exercise: deciding which twenty pages actually matter and describing each in one clear sentence.

The case against overinvesting: it is not a ranking mechanism, it will not get you cited on its own, and it does not control crawler access. Anyone selling llms.txt as an AI visibility strategy is selling the cheapest possible deliverable as though it were the work.

If you publish one

  • Put it at the domain root: https://yourdomain.com/llms.txt.
  • Use plain Markdown — an H1 with your business name, a short summary, then linked sections.
  • Curate ruthlessly. Its only advantage over a sitemap is that it is short and prioritized. A list of 400 URLs discards that entirely.
  • Describe each link in one sentence, in plain language.
  • Keep it current. A stale file pointing at dead pages is worse than none.

Where the real work is

If llms.txt is on your list, the things that measurably affect whether AI tools cite you should come first: NAP consistency, a complete Google Business Profile, schema markup, and pages that answer real questions in direct-answer format.

Publish the file. Just do not mistake it for the strategy.

Related service GEO — Generative Engine Optimization
See the service
Ready to get started?

Now that you know what it means — should you be paying for it?

Not every term in this glossary is worth money to every business. Tell us what you are being sold and we will tell you straight whether it is worth it for you.

Book an initial consult

← Back to the full glossary