On this page
llms.txt is a plain Markdown file at the root of a website (/llms.txt) that lists the pages an AI model should read first, with one line of context for each. Jeremy Howard of Answer.AI proposed it in September 2024. No major AI search engine has said it uses the file to rank or cite sources, and Google says you don't need it. For a B2B service site it takes about 20 minutes to add, so add one if you like, but it won't do much for your visibility in AI search.
We added one to maggnett.com on our second day live. Below is the file we publish, what server logs from 2026 show about who reads llms.txt, and the work that does more for AI visibility.
What is llms.txt?
The proposal lives at llmstxt.org. It came out of a practical problem. AI models work inside a limited context window, and a normal web page is mostly navigation, scripts and layout. An llms.txt file gives a model a short, clean map of the site in Markdown, which models read easily.
The format has one required part and a few optional ones, in this order:
- An H1 with the name of the site or project. This is the only required element.
- A blockquote with a one or two sentence summary.
- Optional paragraphs or lists with more detail.
- Optional H2 sections, each holding a list of links written as
[name](url): note. - By convention, an H2 called "Optional" for secondary links an agent can skip when it is short on context.
The spec also suggests publishing a clean Markdown copy of important pages, for example /pricing.md next to /pricing. Some sites add an llms-full.txt that inlines the full text of every page. That is a common convention, but the core proposal doesn't define it.
Here is the file we publish, trimmed:
# Maggnett > Maggnett is an AI-powered marketing studio. It builds and runs a marketing > "magnet" around a business (content, outbound, search intent, website, > AI growth ops, paid amplification) so the right leads come in. ## Pages - [Your Maggnett](https://www.maggnett.com/your-maggnett): the six magnets. - [Packages](https://www.maggnett.com/packages): how engagements work. - [Free tools](https://www.maggnett.com/tools): a 3-minute growth system score. - [Field notes](https://www.maggnett.com/blog): practical notes for founders and lean teams. ## Notes for agents - Sitemap: https://www.maggnett.com/sitemap.xml
Does llms.txt help you show up in ChatGPT or Google?
On the evidence so far, it does very little for AI search visibility. Three sources point the same way.
Google says you don't need it. Google's guide to AI features in Search says you don't need to create "new machine readable files, AI text files, or markup" to appear in AI Overviews or AI Mode. Gary Illyes of Google said in July 2025 that Google doesn't support llms.txt and has no plans to, and John Mueller has called the file a "temporary crutch".
Most llms.txt files are never requested. In June 2026 Ahrefs published an analysis of server logs from 137,210 domains. About 28% of them published an llms.txt file, and 97% of those files received zero requests in May 2026. SEO audit tools made the largest share of the requests that did happen. Among named AI bots, GPTBot fetched the file most, followed by Claude Code. AI bots never went looking for the file on sites that don't have one. The authors call the file "largely decoration" for AI search. You can read the full Ahrefs study.
Nobody has measured a link to citations either. SE Ranking analysed about 300,000 domains and found no meaningful relationship between having an llms.txt file and how often a domain gets cited in LLM answers. Only about 10% of those domains had the file. The authors hedge with "at least not yet", which is fair. The study shows correlation, not cause.
These studies counted requests, though. A fetch doesn't tell you whether a model used what it read, and a missing fetch today doesn't mean nobody will read the file next year. Still, nothing in the data supports treating llms.txt as a ranking lever.
Who actually reads llms.txt
Most of the reads come from coding assistants and documentation tools. When a developer asks Claude Code or a similar agent to use a library, the agent goes looking for clean docs. That is where llms.txt came from, and developer-tool companies such as Anthropic, Stripe and Cloudflare publish detailed llms.txt files for their documentation.
A B2B service business is in a different position. Your buyers are founders and marketing leads asking ChatGPT or Perplexity "who can help us fix our pipeline", not coding agents reading your API reference. Those answer engines find you through their own search indexes and crawlers, which follow robots.txt and your normal pages.
Should a B2B service website have an llms.txt file?
Yes, as long as you treat it as housekeeping. The file costs almost nothing, it can't hurt your Google rankings, and it makes your site a little easier to read for the AI tools that do fetch it. Writing it also forces a useful exercise: you have to describe your company in two sentences and pick your five most important pages.
Skip it, or leave it for later, when your site is still missing the basics covered further down. An llms.txt file pointing at a vague homepage and a thin services page doesn't fix the vague homepage. If you are not sure where your site stands, our B2B marketing audit checklist covers those basics in 28 checks.
How to write an llms.txt file in 20 minutes
- Write the summary first. One or two sentences: what you do, for whom, and the outcome. If you can't write it, that is the real problem to solve.
- Pick 5 to 10 pages. Services or offers, pricing or packages, two or three of your best articles, about, and contact. Leave out legal pages and tag archives.
- Add a note to each link. One line on what the page answers. Notes help a model choose which page to open.
- Save it as
llms.txtat the site root, served as plain text. On a static site, drop it in the public folder. On WordPress, a plugin or a file upload works. - Keep it in sync. Review it whenever you change your offer or add a cornerstone page. A stale file that lists retired services is worse than none.
You can check it by opening yourdomain.com/llms.txt in a browser. If it shows as readable Markdown, you're done.
What actually moves AI visibility for a B2B site
If the goal is to get mentioned when buyers ask ChatGPT, Perplexity or Google's AI Mode about your category, spend your time here first:
- Check that
robots.txtdoesn't block OAI-SearchBot, PerplexityBot, ClaudeBot or Googlebot. We allow AI search bots and block training-only crawlers like CCBot, since blocking those doesn't affect citations. - Answer the question in the first paragraph. AI answers tend to quote self-contained passages of 40 to 60 words. Each key page should open with a direct answer that still makes sense out of context.
- Use the same description everywhere. Your company name, category and one-line description should match across your site, LinkedIn, directories and partner pages. Models trust what they see repeated across sources.
- Get cited by other sites. AI answers lean on third-party sources such as reviews, roundups, podcasts and community threads. A mention on a page that already ranks does more than any file on your own server.
- Track referral traffic from chatgpt.com, perplexity.ai and similar sources in GA4 so you can see whether any of this works. Most teams never set this up.
This is the part of the work we do in the Search Intent magnet: map the questions your buyers ask, make sure your site answers them, and track whether AI answers start sending people your way. You can see how it fits with the rest on Your Maggnett.
FAQ
What is the difference between llms.txt and robots.txt?
robots.txt tells crawlers what they may and may not access, and the major search engines and AI companies document that they follow it. llms.txt is a voluntary map of the content you want AI models to read first. It doesn't grant or block access to anything. If you want to control AI crawlers, robots.txt is the file that does it.
Can llms.txt hurt my SEO?
No. Google has said the file neither helps nor harms visibility in Search, and it doesn't change how your pages are indexed. The only real risk is an outdated file that sends AI tools to pages you have retired, so keep it current.
Do I need llms-full.txt too?
For most B2B service sites, no. llms-full.txt inlines the full text of your pages, which makes sense for documentation that coding agents load into context. A marketing site with a dozen pages gets little from it, and it adds another file to maintain.
Does ChatGPT read llms.txt?
OpenAI hasn't said that ChatGPT search uses llms.txt to choose sources. Server logs show GPTBot does fetch the file on some sites, but a fetch doesn't show that the content was used in an answer. ChatGPT search finds and cites pages through its own index and the OAI-SearchBot crawler.
Where should the llms.txt file go?
At the root of your domain, so it loads at https://yourdomain.com/llms.txt. Serve it as plain text. A copy inside a subfolder won't be found.