Do you need llms.txt?
For Google, no: Google says its Search ignores llms.txt files. No other major AI company has said its tools read them. This guide explains the file, where it helps, and why we don't sell one.
What llms.txt is
The llms.txt format is a short Markdown file that a site publishes at its main address, such as example.com/llms.txt, to give AI tools a summary of the site and a list of links to the pages worth reading. Jeremy Howard first proposed it in September 2024 on llmstxt.org, and the current text is version 2, revised in August 2026.
The format asks for very little. Only one part is required: a top heading naming the site. A one-paragraph summary, extra notes and headed lists of links are all optional. A list headed Optional marks links an AI tool can leave out when it is short of room.
As a format, llms.txt remains a community proposal, not a web standard. The llmstxt.org page calls itself an informal overview and invites public comment through a GitHub repository and a Discord channel.
Short definition in the glossary.
llmstxt.org, "The /llms.txt file, v2", published 2024-09-03, modified 2026-08-10, read 2026-10-27. Read the proposal.
What Google says about llms.txt
Google Search does not use llms.txt. Google's guide to its generative AI search features puts the file on a list of things site owners can ignore, and says that making one will "neither harm nor help" a site's visibility or rankings, "as Google Search ignores them."
The same answer holds for both of Google's AI formats. The same guide says: "You don't need to create new machine readable files, AI text files, markup, or Markdown to appear in Google Search." For a local business, what counts in Google hasn't changed: pages that Google is able to crawl and index, and facts that are correct everywhere.
One Google product does look for the file. Chrome's Lighthouse tool checks for llms.txt in its agentic browsing audits, which concern AI agents using a site, and sit apart from its SEO audits. Lighthouse marks a missing file as not applicable, since providing one is "optional at the moment", and only flags a file that fails with a server error.
Why AI answers don't need special markup.
Google Search Central, "Google's Guide to Optimizing for Generative AI Features on Google Search", last updated 2026-07-10. Chrome for Developers, Lighthouse "llms.txt" and "Lighthouse agentic browsing scoring", both last updated 2026-05-05. All read 2026-10-27. Read Google's guide.
What ChatGPT and Claude document
OpenAI and Anthropic have not documented their crawlers reading llms.txt. OpenAI's crawler page explains how site owners control OAI-SearchBot and GPTBot through robots.txt, and Anthropic's help article does the same for ClaudeBot, Claude-User and Claude-SearchBot. Neither page, read on 27 October 2026, says those crawlers request or read llms.txt files. OpenAI's page mentions the file once, only as a link to its own documentation index.
Both companies do offer llms.txt files on their own developer documentation sites, as the llmstxt.org page points out. Publishing a file for your own docs is a different thing from reading everyone else's, so it tells a local business nothing about how ChatGPT or Claude pick which businesses to name.
Where an AI company hasn't documented something, AI SEO Vancouver doesn't fill the gap with guesses. If OpenAI or Anthropic publish guidance on the file, this page will say so and the review date will change.
How ChatGPT search finds local businesses.
OpenAI, "Overview of OpenAI Crawlers" (page carries no date). Anthropic, "Does Anthropic crawl data from the web, and how can site owners block the crawler?", last updated 2026-04-07. llmstxt.org, "The /llms.txt file, v2", modified 2026-08-10. All read 2026-10-27. OpenAI crawler page. Anthropic crawler article.
Where llms.txt can be useful
llms.txt earns its keep on sites that AI coding tools read. The proposal says the files are "used most heavily for software documentation", where an agent has to locate the right reference page for a product or code library quickly.
The proposal also suggests a business could use the format to set out its structure and policies. For a local service business with a handful of pages, AI SEO Vancouver hasn't found any primary source showing that the file changes what AI tools say. A clear, accurate website already gives those tools the same facts.
Some platforms now make the file for you. The llmstxt.org page lists Wix, which makes one for each site on its platform, and the Yoast SEO and AIOSEO plugins for WordPress. If your site already has one, there's no need to remove it, but open it once and check that it lists your real pages.
llmstxt.org, "The /llms.txt file, v2", modified 2026-08-10, read 2026-10-27.
llms.txt versus robots.txt
The two files have separate roles. A robots.txt file sets which parts of a site crawlers may visit, and it is an internet standard, RFC 9309. llms.txt only suggests what is worth reading: it grants nothing and blocks nothing.
The llmstxt.org proposal agrees, stating that "robots.txt and llms.txt have different purposes." For AI tools, robots.txt is the file with documented effect: OpenAI and Anthropic both name it as the way a site lets their crawlers in or keeps them out.
Neither file replaces the other, and neither one removes a page from Google. Google says robots.txt "is not a mechanism for keeping a web page out of Google."
IETF, RFC 9309 "Robots Exclusion Protocol", September 2022. Google Search Central, "Introduction to robots.txt", last updated 2025-12-10. llmstxt.org, modified 2026-08-10. OpenAI and Anthropic crawler pages as above. All read 2026-10-27. Read RFC 9309. Read Google's robots.txt introduction.
How to add an llms.txt file safely
Writing the file by hand is quick. Before publishing one, run four checks so it doesn't cause problems.
- Copy your facts exactly. Business name, address, phone, opening hours and services should read the same as on your website and Google Business Profile. A file that disagrees with your site hands AI tools two versions to choose between.
- Link only to live pages you'd be happy to see quoted, such as services, service area, about and contact.
- Save it in the top folder of your site, then open example.com/llms.txt in a browser. It should load as plain text, not as your error page.
- Treat it as the last job, not the first. Letting crawlers in through robots.txt, accurate pages and a correct Google Business Profile all come first, and none of them depend on it.
Here is a sample for a business called Example Plumbing Co., in the order llmstxt.org describes: heading, summary, notes, then headed link lists.
# Example Plumbing Co.
> Licensed plumber serving Burnaby and New Westminster, BC. Repairs, water heaters and drain cleaning for homes and small businesses.
Phone and hours are on the contact page. Prices are quoted after a site visit.
## Services
- [Water heater repair](https://example.com/water-heaters/): tank and tankless units, repair or replacement
- [Drain cleaning](https://example.com/drains/): kitchen, bathroom and main line blockages
## About
- [Contact and hours](https://example.com/contact/): phone, email, hours, service area
- [About us](https://example.com/about/): licence number and insuranceWhat we do about llms.txt
AI SEO Vancouver doesn't sell llms.txt as a fix. If a client wants one, we write it using the same facts as their site. Our time goes first to what AI tools can actually use: crawler access, correct facts and clear answer pages.
How answer engine optimization works.
What we won't promise about llms.txt
An llms.txt file won't change Google's AI answers, because Google Search ignores it, and AI SEO Vancouver won't claim it changes any other AI tool's answers either.
- That adding llms.txt gets a business named by ChatGPT, Claude, Gemini or Google's AI features.
- That any AI company reads the file. Google says its Search doesn't, and OpenAI and Anthropic haven't said either way.
- That passing Lighthouse's llms.txt check means anything for rankings. Lighthouse files that check under agentic browsing, apart from its SEO audits.
Google Search Central, "Google's Guide to Optimizing for Generative AI Features on Google Search", last updated 2026-07-10. Chrome for Developers, "Lighthouse agentic browsing scoring", last updated 2026-05-05. Both read 2026-10-27.
How we check, as of October 2026
This guide rests on five primary sources, each read again on 27 October 2026: the llmstxt.org proposal, Google's AI search guide, Chrome's Lighthouse pages, and the crawler pages from OpenAI and Anthropic. When one of them changes, the page is revised and the date below moves.
For clients, AI SEO Vancouver's tracker saves what ChatGPT, Google AI Overviews, Google AI Mode, Gemini and Claude say each month. If a client publishes llms.txt, the months before and after sit side by side. A change in the same month still wouldn't prove the file caused it, and the report says so.
Still unsure? Send us your question and a person will answer it plainly, by email.
Send a questionQuestions about llms.txt
For Google, no. Google says its Search ignores llms.txt, AI Overviews and AI Mode included. No other major AI company has documented reading the file either. It's optional, and only worth adding if it stays accurate.
A Markdown file published at a site's main address that summarizes the site for AI tools and links to its key pages. It comes from a 2024 community proposal on llmstxt.org and is not a web standard.
Not in Search. Google's AI search guide says Google Search ignores llms.txt files, so one neither helps nor hurts a site there. Chrome's Lighthouse checks for the file, but only in its audits for AI agents, which are separate from SEO.
Nothing OpenAI has published says so. Its crawler documentation covers access through robots.txt and doesn't describe its crawlers reading llms.txt. The route OpenAI does document is letting OAI-SearchBot reach your pages.
Not in Google, which says the file neither harms nor helps. The real risk is a file that disagrees with your website, which gives any tool that reads it a wrong version of your details.
robots.txt is an internet standard that sets which pages crawlers may visit, and OpenAI and Anthropic both use it for site owners' choices about their crawlers. llms.txt only suggests what to read and blocks nothing.
Free snapshot
See what AI says first
The free snapshot runs 10 of your customers' questions through five AI tools and emails you what they said within three business days, before you change any file. Prefer to talk it through? Book a call.
Last reviewed 27 October 2026
What you get
- Your questions, word for word.
- What each AI tool said, with the date.
- Who they named and in what order.
- A few plain notes on what might help.