llms.txt: Does Your Website Need One for AI Search? What the Evidence Says

Every few months a small file becomes the thing marketers are told they must add. Right now that file is llms.txt. Before you spend an afternoon on it, here is what is actually known, and what is not.

1. Where llms.txt Came From

The idea was published by Jeremy Howard of Answer.AI on 3 September 2024, with a specification at llmstxt.org. It proposes a markdown file at the root of your site that lists your most useful pages in a form that is easy for a language model to read. The stated problem was practical: web pages are cluttered with navigation and scripts, and models have limited context.

Some commentators point out that the original proposal was aimed at coding tools and AI agents reading documentation at the moment of use, not at winning citations in AI search. That framing is easy to lose when it gets repeated as an SEO tactic.

In practical terms, the file is plain text in a simple format: a title, a short summary, and a list of links with a short description for each. Anyone can write one in a text editor. Its simplicity is the appeal, and also the reason it is easy to over-sell: a list of links does not make your content better, it only points to it.

2. What Has Been Said About Using It

The picture is mixed, and most of what is public comes from secondary reporting rather than official documentation, so treat the details with care.

• Google: reporting on remarks by Google's Gary Illyes at Search Central Live in July 2025 says Google does not support llms.txt and has no plans to. John Mueller is reported to have compared it to the old keywords meta tag. Check Google's own current documentation, as guidance changes.
• Other AI providers: one source claims some assistants retrieve the file, while others report that no major provider has confirmed using it and that server logs show AI crawlers rarely request it. We could not verify any of these claims against a primary provider statement.
• Correlation studies: at least one analysis reportedly found no link between having the file and being cited by AI tools. We have not independently verified it.

The takeaway is caution, not dismissal. Official positions can change, and different tools behave differently. The safest approach is to treat claims about llms.txt as unconfirmed unless they come from the provider's own documentation, and to avoid paying for services that promise visibility from it.

3. So Should You Add One?

It costs little and does no harm, so there is no strong argument against it. The honest case for it is modest: it is a tidy summary of your key pages that some tools might read. The honest case against expecting much is that nothing confirms it influences visibility in search or AI answers.

If you decide to add one, take it as a small housekeeping task rather than a campaign. Put the file at the root of your domain, list your most important pages (about, pricing, key guides), describe each in one factual sentence, and update it when you publish something major. Then move on to the work that matters more.

4. What Matters More for AI Search Visibility

These five areas are where your effort is most likely to pay off for both traditional search and AI tools. They are also useful to human visitors, which is a good sign that they are worth doing.

1. 1Answer questions directly. Lead each page or section with the plain answer, then explain. AI tools and search features both prefer text that states the answer clearly.
2. 2Use visible FAQs with matching structured data. The question and answer on the page should be the same text your schema describes.
3. 3Keep facts consistent. Your name, services, pricing and policies should read the same everywhere. Contradictions make you harder to quote.
4. 4Make pages crawlable. Content that needs scripts to appear can be missed. Plain HTML text is the safest.
5. 5Earn real mentions. Third-party coverage, reviews and directories carry more weight than any file you control. For a broader view see how customers are starting to find you through AI assistants.

These points are not new; they are the same fundamentals that help people. A visitor who finds a direct answer, a consistent set of facts and a page that loads reliably is better served, and so is any system that reads your site on their behalf. If you improve those, you improve your chances in both traditional and AI-driven search, whatever happens with any particular file.

5. A Simple llms.txt Example

If you decide to add the file, the format is easy. It is a plain text file named llms.txt at the root of your site, such as yourdomain.com/llms.txt. It starts with a title, adds a one-line description, and then lists your most useful pages with a short note on each. Think of it as the contents page of a short guide to your business, written for a reader in a hurry.

• Title and summary: your business name and one sentence on what you do, in plain words.
• Key pages: about, pricing, contact and your main guides, each with a one-sentence description.
• Honest notes: anything an AI tool should not claim about you, such as features that are not yet available.
• Maintenance: update it when you launch a major page, and remove links that no longer work.

The last point is worth stressing. A file that lists pages that no longer exist, or describes features you do not offer, can do more harm than having no file at all. If you cannot commit to keeping it accurate, it is better to skip it and spend the time improving the pages themselves.

Finally, measure honestly. If you add the file, note the date, and keep an eye on how your site is described by AI tools over the following months. Without a before-and-after record, it is easy to credit a file for changes that came from something else entirely, such as new content or improved reviews.

Summary

llms.txt is a harmless, low-cost idea with unproven benefits for search visibility. Add it if you like, but spend the real effort on clear answers, consistent facts and crawlable pages, which help both traditional search and AI tools. Revisit the question when a major provider documents how it uses the file.