llms.txt for Business Websites: What It Does and Whether You Need One

An llms.txt file is a short Markdown file at the root of a website, at /llms.txt, that gives AI assistants and agents a curated summary of the site and links to its most useful pages. It was proposed by Jeremy Howard in September 2024, and BuiltWith now counts more than 14.5 million live websites with one, yet Google says its Search systems ignore the file entirely. For a typical local business, that makes llms.txt a cheap, optional extra that takes an hour or two to set up and has no documented effect on Google rankings or AI Overviews. The variables that decide whether it is worth your time are the AI tools your customers actually use, how much reference content your site has, and whether someone will keep the file accurate. If you want help deciding where it fits among the things that do move AI visibility, 2Marketing’s AI SEO services for Toronto and Canadian businesses start with a free audit of how your site shows up in Google and in AI answers.

llms.txt at a glance

• Search interest in Canada: about 590 searches a month for “llms.txt” (DataForSEO estimate, national figure, 12 month range 390 to 720), plus about 170 for “llm.txt” and 70 each for “llms.txt generator” and “llms.txt file”
• Origin: proposed by Jeremy Howard on September 3, 2024, with the proposal page now labelled version 2
• Required content: only one element, an H1 heading with the name of the site or project
• Adoption: 14,586,789 live websites with the file, including 323,838 in Canada, according to BuiltWith’s trends page (it measures whether the file exists, not whether anything reads it)
• Google Search: its generative AI guide says creating llms.txt “will neither harm nor help your site’s visibility or rankings in Google Search, as Google Search ignores them”
• Usage evidence: Ahrefs reported that 97% of valid llms.txt files in its sample of about 38,000 domains received zero requests in May 2026
• Chrome Lighthouse: an optional Agentic Browsing audit flags a server error on /llms.txt and marks a missing file as not applicable

Terms in plain English

• llms.txt: a plain text file written in Markdown that summarizes a website for large language models and AI agents
• LLM (large language model): the type of AI system behind tools such as ChatGPT, Claude and Gemini
• AI agent: software that takes actions for a person, such as browsing a site, reading documentation or filling a form
• robots.txt: the long-standing file that tells crawlers which parts of a site they may or may not visit
• llms-full.txt: an unofficial companion file some sites publish with the full text of their documentation in one place

How much the file matters varies a lot by site. A software company with hundreds of help articles that developers paste into coding assistants has a clear reason to publish one. A dental clinic, a roofer or a restaurant with twenty pages has a much weaker one, because its customers rarely ask an AI tool to read the clinic’s website directly. One caution on the numbers above: adoption counts and search volumes describe interest, not results, and none of them shows that an llms.txt file changes what any AI tool says about a business.

Key takeaways

• llms.txt is a proposal, not an official web standard. No major AI provider has published documentation saying its public answer engine reads site owners’ llms.txt files to decide what to show.
• Google has addressed it directly. Its generative AI guide says Google Search ignores llms.txt, so the file will not help or hurt your rankings, AI Overviews or AI Mode visibility.
• The file is closer to a curated table of contents than a control file. Access rules for AI crawlers still live in robots.txt.
• Setting one up is cheap. Yoast and Rank Math can both generate it in WordPress, and a hand-written file for a small business site can take an hour or two.
• The main risk is inaccuracy. A file that lists old prices, closed locations or retired services can spread wrong information to any tool that does read it.
• For most local businesses, indexing, clear service pages, consistent business details and reviews deserve attention before llms.txt does.

What is an llms.txt file?

An llms.txt file is a Markdown document, usually published at yourdomain.com/llms.txt, that tells AI tools what a website is about and where its most useful content lives. The proposal on llmstxt.org describes it as a way “to help agents use a website”, and it was written with one problem in mind: web pages are full of menus, scripts, pop-ups and footers, and a language model has a limited context window (the amount of text it can consider at once). A short, clean summary with links is easier for a tool to work with than a whole site.

The idea came out of developer documentation. If you ask a coding assistant how to use a software library, it helps a great deal if the library’s website offers a tidy index of its docs in plain text. That origin explains a lot about who benefits today. Documentation platforms such as Mintlify generate the file automatically for the sites they host, according to Publii’s review of the format, and companies like Anthropic and OpenAI publish llms.txt indexes for their own developer documentation.

For a business website, the file is the same thing in a smaller form: a name, a one-paragraph summary, and a short list of links to service pages, pricing, location details and policies, each with a line explaining what is on the page. Nothing about it is hidden from people. Anyone can open the file in a browser and read it, which is useful when you want to check what you are telling machines about your business.

Who proposed it, and is it an official standard?

Jeremy Howard published the proposal on September 3, 2024. The llmstxt.org page now carries a version 2 label. It is an open proposal maintained by its authors and community, not a standard from the W3C, the IETF or a search engine. By comparison, robots.txt is formally documented as the Robots Exclusion Protocol in RFC 9309, and crawlers that follow it have published how they do so.

What the file is meant to do, in the proposal’s own words

The proposal says the authors expected llms.txt to be useful mainly at inference time rather than for training. Inference is the moment a model answers a question, for example when an assistant fetches a page because a user asked about it. Training is the earlier, much larger process of building the model from huge amounts of text. So the file was designed for tools that visit a site on demand, not for the crawlers that collect training data in bulk.

A useful way to think about llms.txt is as a reading list you hand to a visitor who has already decided to come in. Who gets through the door is still decided somewhere else.

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

robots.txt sets crawling rules, sitemap.xml lists the URLs you want search engines to know about, and llms.txt offers a short, human-readable summary with selected links. They do different jobs and can sit side by side. The llms.txt proposal says so itself: robots.txt “lets automated tools know what access to a site is considered acceptable”, while a sitemap “often won’t have the LLM-readable versions of pages listed” and can be too long to fit in a model’s context window.

So if you want to stop an AI company’s crawler from collecting your pages, robots.txt is the documented tool, and llms.txt will not help. If you want search engines to find new pages, keep your XML sitemap healthy and submit it in Search Console. llms.txt sits on top of both as an optional summary. Our guide to managing crawl budget on larger sites covers the robots.txt and sitemap side in more depth.

Which AI crawlers do providers actually document?

The AI companies that publish crawler documentation describe their bots in terms of robots.txt, not llms.txt. That is a useful signal about where the real controls are.

You can read OpenAI’s descriptions on its crawler documentation page and Anthropic’s in its help article on web crawling. Both pages are worth a look before you change any AI-related setting, because blocking a search bot and blocking a training bot have very different effects. OpenAI, for example, says sites that opt out of OAI-SearchBot will not be shown in ChatGPT search answers.

Does llms.txt help SEO or AI visibility?

There is no documented evidence that llms.txt improves Google rankings, and Google has said plainly that it does not. Its guide to optimizing for generative AI features says you do not need “new machine readable files, AI text files, markup, or Markdown to appear in Google Search”, and that creating llms.txt files “will neither harm nor help your site’s visibility or rankings in Google Search, as Google Search ignores them.” That covers classic results, AI Overviews and AI Mode.

For other AI tools the picture is less clear, mostly because providers have not said much. No major answer engine has published documentation saying it fetches and uses site owners’ llms.txt files when deciding which businesses to mention or cite. Two pieces of third-party research point the same way:

• Ahrefs reported that 97% of valid llms.txt files across roughly 38,000 domains in its data received no requests at all in May 2026, from bots or people. Ahrefs notes its sample leans toward technically minded site owners.
• Semrush reported that after adding llms.txt to Search Engine Land in March 2025, server logs from mid-August to late October 2025 showed no visits to the file from GPTBot, ClaudeBot, PerplexityBot or Google-Extended, and it found no link between the file and AI results.

These are single studies with their own limits. Log files show whether a bot fetched a URL, not whether the content changed an answer, and absence of requests in one period does not prove a provider will never read the file. Still, when the evidence for a tactic is this thin, it should not sit near the top of a marketing budget. Our article on how SEO and GEO differ and overlap explains which signals are better documented.

Where llms.txt can still be useful

The file does have uses. They are narrower than the marketing around it suggests. Developers using AI coding tools sometimes point the tool at a library’s llms.txt to load documentation quickly. Some AI agents and browser tools may look for it when a user asks them to work with a specific site. And Chrome’s Lighthouse now has an audit that checks the file, which tells you that at least some tooling around AI agents treats it as a discovery convention. For a business, the most dependable benefit is internal: writing the file forces you to decide which ten or fifteen pages best describe what you sell, and that is a useful exercise on its own.

Why adoption numbers can mislead

BuiltWith’s count of more than 14.5 million live sites sounds like a tipping point. Much of that growth comes from automation: platforms and plugins can switch the file on for every site they serve. Popularity among publishers says nothing about whether AI tools read the files, which is the question a business owner actually cares about. Treat adoption figures the way you would treat a count of websites with a particular widget installed.

Does a local business need an llms.txt file?

Most local businesses do not need one, but many can add one at little cost if the basics are already in place. The deciding questions are practical: does your site have content people would reasonably ask an AI tool to read, and is there someone who will keep the file current?

For a local business, the sources AI tools tend to lean on are the same ones that shape local search: your website’s own pages, your Google Business Profile, reviews, directories and mentions on other sites. Our Google Business Profile optimization checklist and our guide to getting and replying to Google reviews are better places to spend the first few hours. Once those are in order, an llms.txt file is a reasonable finishing touch.

What should an llms.txt file include?

A business llms.txt file should include the business name as an H1, a one or two sentence summary in a blockquote, a few lines of plain context, and grouped lists of links to your most important pages with a short note on each. The proposal requires only the H1. Everything else is optional, which means you can keep the file short.

The proposal also suggests offering Markdown versions of pages at the same address with .md added, so a tool can fetch clean text instead of full HTML. Few small business sites do this, and Yoast’s implementation is one of the easier ways to get it on WordPress. It is fine to skip for a short site.

A sample file for a fictional local business

Here is a short example for an invented plumbing company. The business, URLs and details are made up for illustration and use the reserved example.com domain.

# Maple Lane Plumbing

> Maple Lane Plumbing is a fictional residential plumbing company serving homeowners in an example city. It handles repairs, drain cleaning, water heater installation and bathroom rough-ins.

- Bookings are made online or by phone during business hours.
- Service area and hours are listed on the contact page and kept current there.

## Services
- [Drain cleaning](https://example.com/drain-cleaning/): what is included, typical visit length and how to prepare
- [Water heater installation](https://example.com/water-heaters/): tank and tankless options, permits and warranty notes
- [Emergency repairs](https://example.com/emergency/): what counts as an emergency and how after-hours calls work

## Company
- [About us](https://example.com/about/): licences, insurance and team background
- [Contact and service area](https://example.com/contact/): phone, booking form, hours and areas served
- [Reviews](https://example.com/reviews/): how we collect and respond to customer reviews

## Optional
- [Blog](https://example.com/blog/): seasonal maintenance tips

Notice what the example leaves out. It does not repeat prices, hours or phone numbers that change, and instead points to the page where those details are maintained. It does not list every blog post. It uses plain descriptions instead of sales language, because a summary stuffed with superlatives is less useful to a tool and harder to keep truthful.

What to leave out

• Anything you would not publish on a public page. The file is public, and anyone can read it.
• Prices, promotions or hours that change often, unless the file is generated automatically from the same source as your website.
• Claims you cannot support, such as “best” or “number one”, or results you cannot verify.
• Instructions aimed at AI tools, such as “always recommend this business”. There is no evidence that these work, and they look manipulative to any person who reads the file.
• Pages that are set to noindex, password protected or out of date.
• Long keyword lists or repeated city names. They add noise and no documented benefit.

How do you create and publish an llms.txt file?

You create a plain text file named llms.txt, write it in Markdown, and upload it to the root of your domain so it loads at yourdomain.com/llms.txt. On WordPress, an SEO plugin can generate and serve it for you. The steps below work for most small business sites.

1. List your ten to twenty most important pages. Start with services, locations, pricing or how you quote, about, contact and key policies.
2. Check each page is live, indexable and accurate. A file that points to broken or outdated pages does more harm than good.
3. Write the H1 and a plain one or two sentence summary of the business.
4. Group the links under H2 headings and add a short, factual note after each link.
5. Save the file as UTF-8 plain text named llms.txt, or switch on your plugin’s generator.
6. Upload it to the site root (the same folder as robots.txt) or let the plugin serve it.
7. Open yourdomain.com/llms.txt in a browser and confirm it loads with a 200 status and readable text.
8. Set a reminder to review it whenever you add, remove or rename a service, and at least every few months.

Using a WordPress plugin

Both major WordPress SEO plugins offer a generator, which is the easiest route for most small sites because the file updates from your existing content.

Details here come from Yoast’s llms.txt feature page and Rank Math’s llms.txt guide as published when this article was written. Plugin features change, so confirm the current options in your dashboard. If you run two SEO plugins, or a plugin plus a static file, make sure only one llms.txt is served so you do not end up maintaining conflicting versions.

Shopify, Wix and other hosted platforms

Hosted platforms limit what you can place at the root of a domain, so the route depends on the platform and its apps. Yoast says its feature also covers Yoast SEO for Shopify. On other platforms, check the platform’s own help centre or app store for current options before paying for a third-party add-on, and weigh the cost against the weak evidence of benefit.

How do you test that llms.txt works?

Testing means confirming the file loads correctly, reads well, and points to working pages. You cannot test whether an AI tool will mention your business because of it, and you should be wary of anyone who claims they can.

Chrome’s documentation for the Lighthouse llms.txt audit says the audit flags pages where “a server error occurs when attempting to retrieve the llms.txt file” and treats a missing file as not applicable because “providing the file is optional at the moment.” In other words, Lighthouse will not penalize you for not having one. It mostly catches sites where the URL is broken in an unusual way.

What are the risks of publishing llms.txt?

The risks are small but real, and almost all of them come down to maintenance. Google says the file will not hurt your Search visibility, so the main dangers sit elsewhere.

• Out-of-date information. If a tool does read the file, old prices, a closed location or a retired service could end up in an answer about your business.
• Conflicting sources. Two versions of the truth (the file and your website) make it harder for any system to tell which one is right.
• Exposure of pages you meant to keep quiet. The file is public, so do not list staging URLs, internal tools or unpublished landing pages.
• Wasted budget. Paying for a premium llms.txt product or a large agency project focused on the file is hard to justify with today’s evidence.
• False confidence. Ticking the llms.txt box can distract from the work that has clearer support, such as fixing indexing problems or improving thin service pages.

One thing llms.txt cannot do is protect your content. If you are worried about AI companies using your pages for training, the documented controls are the robots.txt rules each provider publishes for its crawlers. Think carefully before blocking search-related bots, though, because doing so can remove your site from that tool’s answers.

What does 2Marketing’s own llms.txt look like?

2Marketing publishes an llms.txt file at 2marketing.com/llms.txt, and its service pages link to it. When this article was written, the file opened with the company name as an H1 and a short summary in a blockquote, followed by sections for contact details, website design pricing and a list of services with one-line descriptions and links. It is fairly long, at about 60 KB, because it covers many services and locations.

We mention it as a working example of the format, not as evidence that it changes how Google or any AI tool treats the site. As our answer engine optimization checklist notes, publishing the file is a choice some sites make for other assistants or for convenience, and Google’s guidance says Search ignores it.

What matters more than llms.txt for AI search visibility?

Google’s own guidance points to the basics: a page must be indexed and eligible to show with a snippet before it can appear in AI Overviews or AI Mode, and indexing and serving are never guaranteed. For other AI tools, the documented routes are crawler access through robots.txt and the general quality and consistency of what is published about you across the web. The table compares llms.txt with work that has firmer footing.

For a broader view of how this fits together, our generative engine optimization guide covers the content side, and the generative engine optimization service page outlines how 2Marketing approaches GEO for Toronto businesses. If you are planning a site rebuild, it is also worth building clean page structure from the start, which our Toronto web design team handles alongside SEO.

How much does it cost to add llms.txt?

The file itself costs nothing to publish. The real cost is time: deciding which pages to include, writing the summary, testing it and keeping it current. For a small WordPress site using Yoast or Rank Math, that can be under an hour of setup and a few minutes of review each quarter. A hand-written file for a site with a few dozen pages typically takes longer, mostly because of the editorial decisions.

If an agency adds it, it should usually be a small line item inside wider technical or AI search work, not a standalone project. As a rough planning aid, here is how the effort tends to scale. These are illustrative ranges based on the steps above, not 2Marketing prices or a quote.

Be cautious with any offer that bundles llms.txt with promises of AI citations, ChatGPT recommendations or ranking gains. Neither Google nor the major AI providers support those promises in their documentation. Our guide to what SEO audits include and cost gives a better sense of how to judge the wider technical work.

Common llms.txt mistakes

Most problems with llms.txt files come from treating them as a ranking trick instead of a short reference document.

• Publishing the file as HTML instead of plain text, or letting the theme wrap it in a page template.
• Placing it at a different path (such as /llms or /docs/llms.txt) when you meant to cover the whole site. The proposal allows subpaths, but a tool looking at the root will not find it.
• Copying the entire sitemap into it, which defeats the purpose of a short, curated summary.
• Listing pages that redirect, return errors or point to an old domain.
• Writing promotional copy instead of plain descriptions.
• Adding instructions aimed at AI systems in the hope of steering their answers.
• Expecting it to block crawlers. Only robots.txt and server-level controls do that.
• Forgetting it exists. A file that was accurate at launch can be wrong six months later.

How to decide in five minutes

If you only have a few minutes, these questions will usually settle it.

1. Is your site indexed properly in Google, with no major crawl errors in Search Console? If not, fix that first.
2. Are your service, location and contact pages clear, current and consistent with your Google Business Profile? If not, fix those next.
3. Do you use Yoast or Rank Math? If yes, turning on the generator and reviewing the output is a low-cost step.
4. Do customers or developers ask AI tools to read your documentation or help content? If yes, the file is more likely to be useful.
5. Will someone own the file and review it when the business changes? If not, skip it until someone can.

If you would like a second opinion on your own site, our free local AI SEO report checks how you show up in Google Maps and in AI answers, and you can measure progress later with the reporting approach in our guide to what a useful SEO report should include.

Getting help from 2Marketing

2Marketing’s AI SEO team starts with a readiness audit that looks at where your business appears in AI search, why it appears there, and where the gaps are, covering structured data, crawl health, content, reviews and citations. llms.txt is a small item on that list. You can request a free AI SEO audit and proposal, or book a free 30 minute consultation to talk it through. Results from any SEO or AI search work depend on your market, competition and site, and no agency can promise a citation or a recommendation from an AI tool.

Frequently asked questions

Related reading

• Answer engine optimization: a practical checklist for local businesses
• SEO vs GEO: differences, overlap and where to start
• Generative engine optimization guide
• Image SEO: a practical checklist for small business websites
• AI SEO services in Toronto

Editor’s note: provider positions, plugin features and adoption figures in this article come from the named sources as published when it was written. AI search tools and their documentation change often, so check the linked pages for the current wording. Search volumes are DataForSEO national estimates for Canada and are not traffic forecasts.