What is llms.txt, and does your site need one?

An llms.txt file is a short plain-text file at yoursite.com/llms.txt that tells AI tools what your site is and links to the pages most worth reading. It does not change your Google rankings, and Google says you need no such file to appear in its AI answers. For most small businesses it is a half-hour job worth doing only once the pages it points to are good, and never worth paying extra for.

That is the short answer. The longer one matters, because a lot of what is written about llms.txt is sold by people who charge to make the file. I make websites and sell SEO, and this site has an llms.txt of its own, so I have a view. Where a fact comes from a source, I link it. Where it is my opinion, I say so.

What is llms.txt?

llms.txt is a proposal, not a standard that any search engine has adopted. Jeremy Howard published it in September 2024, and a second version followed in August 2026 (llmstxt.org (opens in a new tab)). The idea is simple: a website is built for people and browsers, full of menus, scripts and layout, and an AI tool reading it has to dig through all of that to find the point. An llms.txt file gives it the point up front, in Markdown, the plain-text format with # for headings and [text](link) for links.

The proposal asks for very little. The only required part is a heading with the site's name. After it, by convention, come a one-paragraph summary, any short notes, and lists of links grouped under headings, each link with an optional sentence about what is on that page.

The authors say they expected the file to be used at inference, when an AI tool is answering someone, rather than for training, "and that is how it has been used" (llmstxt.org (opens in a new tab)). In plain terms: the file is read when an AI agent is working on a task and needs to understand a site quickly, not when a company is gathering text to build a model.

What does an llms.txt file look like?

Here is one for an imaginary plumbing company. Yours would use your real pages:

# Riverside Plumbing

> Riverside Plumbing is a licensed plumbing company serving
> Springfield and the surrounding county. We fix leaks, replace
> water heaters and clear drains for homes and small businesses.

## Services
- [Water heater replacement](https://example.com/water-heaters): Tank and tankless, same-week installs, typical prices.
- [Drain cleaning](https://example.com/drains): What we clear, how long it takes, what it costs.
- [Emergency plumbing](https://example.com/emergency): Hours, call-out fee and the area we cover.

## About
- [About us](https://example.com/about): License numbers, insurance and who does the work.
- [Reviews](https://example.com/reviews): Customer reviews, unedited.
- [Contact](https://example.com/contact): Phone, address and booking.

That is the whole thing. No code, no special tags, nothing a search engine has to process. It reads like a well-organized table of contents, because that is what it is.

Where does llms.txt go, and how do you see one?

It goes at the root of your domain, the same place as robots.txt: yoursite.com/llms.txt. The second version of the proposal also allows a file inside a folder, such as yoursite.com/docs/llms.txt, which then covers only the pages under that folder (llmstxt.org changes (opens in a new tab)).

To see any site's file, type /llms.txt after its domain. Anthropic, OpenAI and Google's Gemini API documentation each publish one; I opened all three while writing this. A 404 means the site does not have one, which is still true of most small-business sites.

How is llms.txt different from robots.txt and a sitemap?

All three are plain files at the root of a site, and people mix them up. They do different jobs:

The important row is the fourth. llms.txt curates; it does not control. If you want to keep an AI company's crawler off your site, that is robots.txt's job, or your firewall's. I cover which crawlers to allow in my guide to answer engine optimization.

Who actually reads llms.txt?

This is the question most guides skip, and the honest answer is: fewer tools than the hype suggests, and not the ones most business owners care about.

Here is what the companies themselves have said:

• Google Search does not need it. Google's guidance for its AI features says: "You don't need to create new machine readable files, AI text files, or markup to appear in these features" (Google Search Central (opens in a new tab)).
• Google's own search advocate has been blunt. In June 2025, John Mueller wrote: "FWIW no AI system currently uses llms.txt" (Bluesky (opens in a new tab)).
• Google's Chrome team now checks for it. Lighthouse, the site-testing tool built into Chrome, has an llms.txt audit in its "agentic browsing" section. It only flags a file that causes a server error, and it marks a missing file "Not Applicable," because "providing the file is optional at the moment" (Chrome for Developers (opens in a new tab)).
• AI companies publish their own. Anthropic, OpenAI and Google's Gemini API each serve an llms.txt for their developer documentation, which is where the file has caught on most: helping coding assistants find the right page of a manual.

So Google is on both sides of it: its search team says you do not need the file, and its browser team audits for it. That is less of a contradiction than it looks. Search engines decide which pages to show by crawling and judging the pages themselves, and a file in which a site describes itself adds little they can trust. AI agents are different: when a person asks an agent to do something on a site, such as compare prices, book a service or read the documentation, a short map saves it from wandering.

What I could not find, as of October 2026, is any statement from Google, OpenAI, Anthropic or Perplexity that their AI search uses llms.txt to decide which businesses to name in an answer. If one appears, I will update this post.

Does llms.txt help SEO or AI visibility?

Not for Google rankings. Google ranks pages, and says it needs no AI text file to show yours in AI Overviews or AI Mode. Anyone promising you better rankings from an llms.txt file is promising something Google has said it does not do.

For AI answers more broadly, there is no published evidence that it helps either. What gets a business named in an answer is still the slow work: pages that answer real questions in plain text, a site the AI crawlers are allowed to read, and being described consistently across the web. That is what my guide to answer engine optimization covers, and an llms.txt file is item seven of seven on that list for a reason.

Where it can plausibly help is the newer case: an AI agent sent to your site by a person to get something done. That use is small today. I think it will grow, but that is my opinion, and the file costs so little that you do not need to bet on it to justify one.

Does your site need one?

Need, no. Here is how I would decide, and this part is my judgment, not a rule:

Add one if:

• Your site already has clear pages for your services, prices or price ranges, service area and contact details. The file points to good pages; it cannot stand in for them.
• You run WordPress with Yoast SEO, where it is a single setting (see the box below).
• You publish documentation, a product catalog or a lot of articles, where a map genuinely saves a reader, human or machine, from getting lost.

Leave it for later if:

• Your site is missing the basics above. An hour spent on a missing prices page will do more than any file.
• Someone is charging you for it as an "AI visibility" upgrade. It is a text file you can write in half an hour, and the benefit they are selling has not been shown.

How do you create an llms.txt file?

If you decide to add one, this is the whole job:

1. Write the heading and summary. One # line with your business name, then one > paragraph saying who you are, what you do and where. Write it the way you would describe the business to a new customer on the phone.
2. List the pages that answer customer questions. Group them under ## headings (Services, About, Contact). Give each link one plain sentence about what is on the page. In my experience, five to fifteen links is plenty for a small business; the point is to choose, not to repeat your sitemap.
3. Save it as plain text named llms.txt and upload it to the root of your site, next to robots.txt.
4. Open yoursite.com/llms.txt in a browser and check that the text appears, not a 404, a login page or a security check.
5. Keep it current. A file that links to deleted pages is worse than none. The easiest way is to have it generated, so it changes when the site does.

A note on llms-full.txt: some tools, documentation platforms especially, also produce a second file with the full text of every page in one place. It is common on software documentation sites, but it is not part of the proposal at llmstxt.org, and for a small-business site I would not bother; the second version of the proposal points agents to Markdown copies of individual pages instead.

How I built mine

This site's llms.txt is not written by hand. It is generated every time the site is built, from the same data that builds the pages: a summary of who I am and where I work, every service page with its description, my about, case study, review, FAQ and process pages, the blog posts that are published (drafts never appear), and how to reach me. When I add a post or change a service, the file changes with it, so it cannot drift out of date.

The one thing I check by hand is that the file can actually be fetched. While writing this, I requested my own llms.txt from a cloud server and got a security-check page instead of the file. Cloudflare, which sits in front of my site, was challenging that server's address, as it does with many data-center addresses. That was a test from a data-center address, not an AI crawler, so it does not prove AI tools are turned away. It does show how easily a firewall can hide the file from the very tools it was written for. If you add one, look at it from outside your own browser, and check your firewall's bot settings.

What llms.txt will not do

• It will not block anyone. It is a suggestion of where to look. Use robots.txt or your firewall to keep crawlers out.
• It will not raise your rankings. Google has said it does not need the file for its AI features, and nothing in its search documentation treats it as a ranking signal.
• It will not fix thin pages. If the pages it links to do not answer the question, the map just leads an agent to a dead end faster.
• It will not keep itself current. Generate it, or put a reminder on your calendar.

If you are letting AI agents do more than read your site, such as editing it, that is a different job with real risks, and I cover it in my guide to WordPress MCP.

The short version

• llms.txt is a plain Markdown file at yoursite.com/llms.txt: your name, a one-paragraph summary and links to your most useful pages.
• It is a proposal from September 2024, revised in August 2026, not a search engine standard.
• Google Search says you do not need it for AI Overviews or AI Mode, and it does not affect rankings.
• It is used today mostly by AI agents and coding tools reading documentation; Chrome's Lighthouse checks for it.
• Add one when your key pages are already clear, keep it short, and generate it so it stays current.
• Never pay extra for one, and make sure your firewall lets it be read.