Home Blog

AI search

llms.txt: does a small business need one?

On this page
  1. Quick version
  2. What llms.txt is
  3. Who actually reads it
  4. What llms.txt does not do
  5. If you still want one
  6. What to spend the time on instead
  7. Decision guide
  8. Frequently asked questions

No, a small business does not need an llms.txt file. Semrush reports that no major AI platform, including ChatGPT, Google's AI Overviews or Claude, has confirmed that it reads or relies on llms.txt. Semrush sells site audit tools and llms.txt validation services, so weigh that in context. If you want the file anyway, it is a small job, but treat it as optional.

Quick version

  • What it is: a short Markdown index at the root of a site that lists key pages so machines can orient themselves, per RedFork. RedFork is a Wix studio that sells llms.txt implementation.
  • Who requests it: mostly SEO audit tools. In one Ahrefs measurement, 97% of published files got zero requests in May 2026, as RedFork reports.
  • What it does not do: it does not opt you out of AI training, and Google says it has no effect on Search rankings or AI Overviews.
  • Where it helps: software documentation, where coding agents follow it to find API references and tutorials, according to the llmstxt.org proposal site.
  • Our view: skip paid setup. If you write one, keep it to an afternoon task at most and spend the real effort on pages people and search engines already use.

What llms.txt is

An llms.txt file is a short Markdown index at a site's root that lists key pages, so machines can orient themselves, as RedFork describes it. Jeremy Howard proposed it in 2024, according to the same article. RedFork sells llms.txt implementation, so it has a reason to explain the file well, and also a reason to sell it. Its own advice is to skip it and do other things.

The proposal site says the H1 heading with the name of the project or site is the only required section, and that the file can sit at the site root or at any path within it, per llmstxt.org. That site belongs to the proposal's authors, so it is an advocate, not a neutral party. On format, Semrush says the file should be written in Markdown, using # for the H1, ## for H2s and > for blockquotes. Save it as lowercase llms.txt, which is the filename the official specification uses.

Here is a made-up example, not a real business, showing the layout built from those Markdown elements:

# Example Plumbing
> Plumbing repairs, installs and call-outs for homes in one town.

## Services
- [Drain cleaning](https://example.com/services/drains): What a visit involves
- [Water heaters](https://example.com/services/water-heaters): Repair and replacement

## Company
- [Pricing](https://example.com/pricing): Call-out fee and rates
- [About](https://example.com/about): Who we are
- [Contact](https://example.com/contact): Phone and opening hours

Who actually reads it

The best public data on this comes from request logs, not from vendor claims. These are the figures we found.

  • Ahrefs looked at 137,210 domains. About 28% of them published a valid file, and 97% of published files got zero requests in May 2026. Ahrefs treats the 28% as a ceiling because its customers skew technical. RedFork reports both figures.
  • Of the requests that did arrive, AI retrieval bots such as OAI-SearchBot and PerplexityBot made up about 1.1%. SEO audit tools took roughly 21.7%. Same RedFork article.
  • Semrush ran its own test. From mid-August to late October 2025, its llms.txt page received zero visits from Google-Extended, GPTBot, PerplexityBot or ClaudeBot, it says.
  • Among large firms, only 7.4% of Fortune 500 companies had added llms.txt, against 92.8% with a standard robots.txt, according to cCoreVentures, citing ProGEO.ai from March 2026. cCoreVentures sells AI strategy consulting for small businesses.

Adoption is growing, though counts differ by source. Semrush says 951 domains had published a file in July 2025 and 16,670 had by September 2026. That is roughly 17.5x growth: 16,670 divided by 951 is about 17.5. A different count from cCoreVentures has 4,088 sites in June 2025 and 36,120 by May 2026. The two counts use different methods and we have not tried to reconcile them. Both agree on direction. Growth in publishing is not the same as growth in reading.

Where it does get used

The llmstxt.org site says llms.txt files are used most heavily for software documentation, where coding agents follow them to find API references and tutorials. It also says OpenAI, Anthropic and Gemini publish llms.txt files for their own developer docs, and that Chrome's Lighthouse audits sites for one as part of its agentic browsing checks. If you run a docs site or an API, that is a real use case. A plumber, a bakery or a consultancy is a different case.

What llms.txt does not do

Three sources address this directly. Digital Applied sells agentic SEO services that include AI crawler governance. cCoreVentures sells AI consulting. Semrush sells audit tools. None of them sells a service that depends on you ignoring the file.

What you might hope forDoes llms.txt do it?Source
Opt out of AI trainingNo. It is a navigation guide at inference time and does not control training or crawl permissions.Digital Applied
Improve Google rankings or AI OverviewsGoogle states it has no effect on either.cCoreVentures
Improve results in AI answersSemrush's testing found no correlation.Semrush
Block a crawlerNo. A firewall rule is enforced before robots.txt is even read, so a WAF block beats any robots.txt Allow.Digital Applied

Google's own guidance, as RedFork summarizes it, says you don't need new machine-readable files, AI text files or markup to appear in those features. cCoreVentures also quotes John Mueller as saying no AI system currently uses llms.txt.

One counterpoint. OnePoint Solutions describes the file as a briefing document for ChatGPT, Claude, Perplexity and Google's AI Overviews. It sells llms.txt and schema setup starting at $499. That framing does not match the request data above, and we weigh the logs more heavily than the pitch. This is our judgment, not a sourced fact.

If you still want one

The cost is low. OnePoint Solutions says it takes about thirty minutes to write a good one, though that company sells the service. The steps below combine that guidance with Semrush's.

  1. List your key pages. Semrush suggests, at the very least, product or service pages, up-to-date blog posts, a pricing page, an about page and a contact page.
  2. Write plain facts. OnePoint says to strip marketing words like "industry-leading" and stick to facts. It also says to publish prices if you have them.
  3. Keep it short. The same guide suggests 300 to 800 words and putting the most important information first.
  4. Save it as llms.txt in lowercase, in Markdown, per Semrush, at the site root, where RedFork says the file lives.
  5. Skip hand-writing if you use a docs platform. The llmstxt.org site says Mintlify, GitBook and Yoast SEO can generate the file automatically.

Do not pay a large sum for this. Our view: given the request data, a file that takes an afternoon to write is a reasonable experiment, and a paid package is hard to justify. That is an opinion, and you may weigh it differently if you sell to developers.

What to spend the time on instead

RedFork's article ends by recommending that you spend the hour on a visible last-updated date and the first sentence under each heading instead. It cites Seer Interactive, which found that three in four AI-cited pages had been updated within a year. It also says AI-cited content runs roughly 25.7% fresher than traditional search results. RedFork sells marketing services, so treat this as a pointer to check, not proof. It fits what search engines have long rewarded: pages that are current and answer the question fast.

That is also what we build at Paperboy. Paperboy writes, fact-checks and publishes SEO articles to a small business site every week. Each plan is priced per site, per our pricing page: Starter is $49 a month for 3 articles a week, Growth is $99 a month for 7 a week, and Pro is $179 a month for 14 a week. We sell this, so discount our opinion accordingly. If you want to see how an AI assistant reads your site today, our free AI check is a place to look, and our blog covers related questions.

Decision guide

Your situationOur view
Local service business or shopSkip it, or write one yourself in an afternoon.
Online store with a pricing pageOptional. Make sure the pricing page itself is clear first.
Software product with docsWorth having. Docs are where the file is used most.
Someone offers paid setupAsk what request data they can show. Compare against the published logs above.

This table is our judgment, not a sourced fact.

Frequently asked questions

Does llms.txt help a site show up in ChatGPT or Google AI Overviews?

There is no confirmation that it does. Semrush says no major AI platform has confirmed reading it, and Semrush did not find a correlation with better AI results in its own testing.

Can llms.txt stop AI companies from training on a site?

No. Digital Applied says llms.txt is not a training opt-out and does not control training or crawl permissions. Crawl rules belong in robots.txt or firewall rules.

Where does the llms.txt file go?

RedFork describes it as a Markdown index at the site root. The llmstxt.org proposal also allows it at any path within the site, covering the pages under that path.

Is llms.txt the same as robots.txt?

No. OnePoint Solutions notes that llms.txt sits at the root like robots.txt and sitemap.xml, but it serves a different purpose. It is a guide to your pages, not a set of crawl permissions.

Who sends requests to llms.txt files?

RedFork reports that SEO audit tools took roughly 21.7% of requests and AI retrieval bots about 1.1%, citing Ahrefs data from May 2026.

How this article was checked

Fact check

llms.txt: does a small business need one?

paperboyseo.com/blog/llms-txt-small-business
Held
2 issues

to fix before this article can publish. The engine held it back.

  • Every claim has a source: passed
  • Numbers match their sources: passedEvery number in a cited sentence was looked up in the saved source text.
  • Sources say what the article says: held2 items held for a fix
  • Sounds like the customer: passedVoice match 92 of 100.
  • No banned claims: passed
  • House style: passed
7 sources read
  • llmstxt.org
  • redforkmarketing.comSells something
  • ccoreventures.comSells something
  • digitalapplied.comSells something
  • onepointsolution.caSells something
  • semrush.comSells something
  • paperboyseo.com
4 small notes from the check
  • Reads as a factual claim but has no source link.
  • Reads as a factual claim but has no source link.
  • Reads as a factual claim but has no source link.
  • An opinion stated as a fact.
Sources read
7
Score
68/100
Checked
Oct 29, 2026

Sources

The pages the engine read for this article.

  1. The /llms.txt file, v2 – llms-txtllmstxt.org
  2. Do You Need an llms.txt File for Your Small Business Website?redforkmarketing.com
  3. llms.txt Won't Get You Cited by ChatGPT — Here's What Actually Will | cCoreVentures Blogccoreventures.com
  4. AI Crawler Access Control: The 2026 Decision Matrixdigitalapplied.com
  5. llms.txt Guide for Small Business and Online Storesonepointsolution.ca
  6. What is llms.txt & should you use it?semrush.com
  7. Paperboy: Fact-checked SEO articles on autopilot | Paperboypaperboyseo.com

About Paperboy: Paperboy writes SEO articles for your own website every week and checks every number against its source before publishing. It is operated by Hive Asset Group, LLC.

Want a guide like this for your own site?

Paste your address. In under ten minutes you read a finished, fact-checked article written for your site. Free, no card.

You pick a plan only after you have read it.

Prefer to run it yourself? Self-host it (repo coming soon). Using an agent? Add the MCP server.