llms.txt: what it does for AI search, and what it does not

llms.txt gets a lot of attention, and a lot of the advice about it skips the important part: what the engines themselves say. This guide covers what the file is, where it came from, what Google, OpenAI and Perplexity have said about it, a simple template, and where it sits in your priorities.

The short answer: it is cheap to add and harmless, but Google says Google Search does not use it, and we found no statement that ChatGPT or Perplexity search read it. It is not what gets you cited.

What llms.txt is

llms.txt is a proposal for a plain text file, written in Markdown, that gives language models a short guide to a website: what the site is, and links to its most useful pages. It was proposed by Jeremy Howard on llmstxt.org, published on 3 September 2024. The proposal describes it as a file that offers brief background information, guidance, and links to detailed markdown files.

The idea is that a model reading your site should not have to wade through navigation, scripts and ads to find the substance. The file gives it a curated map instead. The proposal also suggests offering clean Markdown versions of your pages at the same URL with .md added.

What the format looks like

The proposal sets out a simple structure, in this order:

1. A heading with the site's name. This is the only required part.
2. A short summary in a quote block.
3. Optional detail in plain paragraphs or lists.
4. Sections of links, each under its own subheading, listing URLs with a short note on each.
5. An optional section for secondary pages a model can skip if it is short on space.

The file sits at the root of your site, at /llms.txt, or at any path within it to cover the pages under that path.

A simple template

Here is a plain example for a made up company, written out line by line. In the real file, each line below starts at the beginning of the line.

• Line 1: # Acme Analytics
• Line 2: > Acme Analytics is reporting software for US dental practices. Plans start at $99 a month.
• Line 3: ## Product
• Line 4: Features: https://www.example.com/features.md: what the software does
• Line 5: Pricing: https://www.example.com/pricing.md: plans and what each includes
• Line 6: ## Guides
• Line 7: Setup guide: https://www.example.com/setup.md: how to connect your practice software
• Line 8: ## Optional
• Line 9: Company history: https://www.example.com/about.md

In the proposal's format, each link is written as a Markdown link inside a list item. Keep it short. The point is to give a model your best pages, not all of them.

What the engines say about it

This is the part that matters.

Google

Of the three, Google is the one that addresses it directly in its documentation. Its AI optimization guide says you do not need to create machine readable files, AI text files, markup or Markdown to appear in Google Search, including its generative AI features, because Google Search itself does not use them. It adds that it is completely fine to create and maintain llms.txt files for other services or systems that use them.

So for AI Overviews and AI Mode, llms.txt does nothing. Google's separate Chrome Lighthouse tool does include an optional llms.txt check in its agentic browsing category, describing the file as an emerging convention. That is a developer audit, not a Search signal, and Lighthouse treats a missing file as not applicable.

OpenAI and Perplexity

We read both companies' crawler documentation on 31 October 2026 and found no statement that their search crawlers or assistants read llms.txt on other sites. What both do document is robots.txt: OpenAI says sites that block OAI-SearchBot will not be shown in ChatGPT search answers, and Perplexity recommends allowing PerplexityBot to appear in its results. Those rules affect your visibility. llms.txt, as far as their documentation says, does not.

Should you add one?

Yes, if it takes you ten minutes and you keep it accurate. It is harmless, Google says so explicitly, and Google's own Lighthouse describes it as a convention designed for language models and AI agents. If you run WordPress, the free Yoast SEO plugin says it generates the file with one click.

But do it in the right order. These come first, because the engines document that they matter:

1. Crawler access. Allow OAI-SearchBot and PerplexityBot in robots.txt, keep Googlebot unblocked, and check your CDN is not blocking them.
2. Indexing and snippets. Google says a page must be indexed and eligible to show with a snippet to appear in AI Overviews or AI Mode.
3. Clear pages. The answer first, facts in plain text, one topic per page.
4. The pages the engines cite. If roundups and comparisons cited for your category leave you out, that is the gap to close.

How to tell what is actually working

llms.txt will not show up in your AI visibility, because the major engines do not say they read it. What you can measure is whether the answers name you. Our free checker will check whether AI engines name you: five buyer prompts across Google AI Overviews, ChatGPT, Gemini and Perplexity, with every answer and the pages each one cited, in about two minutes, with no card and no email. If you are missing, the cited pages tell you where to work, and it will not be your llms.txt.

Questions

Does Google use llms.txt?

No. Google's AI optimization guide says Google Search does not use llms.txt or other AI text files, including for its generative AI features, and that creating one is fine if other services use it.

Does ChatGPT or Perplexity read llms.txt?

We found no statement in OpenAI's or Perplexity's crawler documentation that their search products read llms.txt on other sites. Both document robots.txt as the way to control their crawlers.

Where does an llms.txt file go?

The proposal says it can sit at the site root, at /llms.txt, or at any path within the site to cover the pages under that path.

Is there an llms.txt generator?

Yes. Some tools create one for you. For example, the free Yoast SEO plugin for WordPress says it generates an llms.txt file with one click. You can also write one by hand in a few minutes.