What is llms.txt?
Format, example and whether it works
llms.txt is a Markdown text file you place at the root of your domain (yourcompany.com/llms.txt) that gives AI models a short, curated overview of your most important content. The format was proposed by Jeremy Howard in September 2024. Today OpenAI, Anthropic and Google publish the file for their own developer documentation, and Chrome's Lighthouse checks for it. But none of the major AI search services documents using it to decide whom they mention — and in our measurement of 1,912 Nordic businesses, sites with the file were not cited more often. It takes ten minutes to make: do it, but do not expect it to move anything on its own.
What llms.txt is — the definition
llms.txt is a markdown file placed in the root of your domain — the same way as robots.txt. It gives large language models (LLMs like those behind ChatGPT, Claude and Perplexity) a curated, easy-to-read overview of your business's most important content: what you do, which pages matter most, and how they fit together.
The format was proposed in 2024 by Jeremy Howard (Answer.AI) as an open standard — a suggestion for how websites can voluntarily make themselves easier for AI models to understand. The idea is simple: modern websites are full of navigation, scripts and noise that make it hard for a model to find what matters. An llms.txt cuts to the bone.
Do AI engines use llms.txt? The honest answer
The short answer: not in any documented way when it comes to getting mentioned. Three things are certain, and they point in different directions:
- The big AI companies use the format themselves. OpenAI, Anthropic and Google publish llms.txt for their own developer documentation, and thousands of sites have the file today (llmstxt.org, 2026).
- Chrome checks for it. Google's Lighthouse now has an llms.txt audit under "agentic browsing". It only fails if the server errors when the file is fetched; if the file is missing, the audit is marked not applicable, because the file "is optional at the moment".
- No AI search service has committed to using it to decide who gets mentioned. Google's John Mueller has called the question of its effect speculative, and our own measurement found no difference (see below).
So the file is a practical aid for AI agents that need to understand a site quickly — not a shortcut into ChatGPT's answers. The full review with sources is at does llms.txt work?
What an llms.txt looks like
An llms.txt is plain markdown and doesn't need to be long. The basic structure is a heading with the business name, a short summary, and then curated links to your most important content:
# Your Company
> A short description of what you do, for whom, and where you are based.
## Most important pages
- [Services](https://yourcompany.com/services): What we offer
- [About](https://yourcompany.com/about): History and team
- [Contact](https://yourcompany.com/contact): How to reach us
## Frequently asked questions
- [FAQ](https://yourcompany.com/faq): Answers to the common questionsYou can generate a complete, correctly formatted file in one minute with our free llms.txt generator.
The format rules
The specification at llmstxt.org (version 2, August 2026) is short:
- An H1 with the name (
# Your Company) — the only required part. - A blockquote with a short summary (
> …) with what a model most needs to know. - Optional text — paragraphs or lists with more detail.
- H2 sections with lists of links in the form
- [Name](url): note. - A section called "Optional" for secondary links a model can skip.
The file normally sits at the root, but may also live in a subfolder (e.g. /docs/llms.txt) and then covers pages under that path. Some sites also publish an llms-full.txt with all content in one file, and the specification suggests that every page can be fetched as clean Markdown by adding .md to its address. Both are optional and make most sense for documentation, not for a company website.
llms.txt vs. robots.txt vs. sitemap.xml
The three files live in the same place (the domain root), but do different things:
| File | Tells | To whom |
|---|---|---|
| robots.txt | What crawlers may NOT fetch | All crawlers |
| sitemap.xml | Every URL that exists | Search engines (indexing) |
| llms.txt | What is WORTH reading, and in what order | AI models (understanding) |
They don't replace each other — they complement. And one important detail: make sure your robots.txt actually grants the AI crawlers access, or they will never reach your llms.txt.
How to make your own llms.txt
- Generate the file. Use the llms.txt generator — fill in the fields and get a finished file in one minute. Free, no login.
- Place it in the root. The file must live at yourcompany.com/llms.txt (not in a subfolder).
- Check that AI can read your site. Run the AI Readability Check to confirm that the crawlers have access and can see your content.
- Keep it up to date. If your most important content changes, update the file — an outdated llms.txt sends the AI off track.
If you want it done — and maintained — automatically, the Geoa platform publishes and maintains your llms.txt directly via the WordPress or Shopify integration, together with schema markup and the other signals. From DKK 299/mo with a 14-day free trial, no credit card required.
How to check the file works
- Open the address in your browser: yourcompany.com/llms.txt. You should see plain text, not an error page or your home page.
- Check the server responds correctly. The file should load with status 200 and as text. If you run Chrome's Lighthouse, it shows up under "agentic browsing".
- Check the links in the file. Every link should point to a page that exists — a link to a deleted page sends an AI agent nowhere.
- Check robots.txt. If you block GPTBot, ClaudeBot or PerplexityBot, they never reach the file. AI Readability Check shows this in a minute.
What our measurements show
We measured 1,912 businesses in Denmark, Norway and Sweden from July to August 2026. 202 of them had an llms.txt — roughly one in ten. Sites with the file were used as a source in AI answers in 43.6% of cases; sites without it in 43.3%. The difference is too small to tell apart from chance (see the measurement with figures per country).
What did make a difference was the rest of the technical foundation: AI crawlers having access, text that can be read without JavaScript, and the business's facts in structured data. Sites that got those right were used as a source far more often (see the measurement). By all means make your llms.txt — but spend the next hour on those points.
Frequently asked questions about llms.txt
What is llms.txt in short?
A markdown file in the root of your domain (yourcompany.com/llms.txt) that gives AI models a curated overview of your most important content. The format is an open proposal from 2024; no major AI search service has documented using the file to decide whom it mentions.
Where should the file go?
In the root of your domain: yourcompany.com/llms.txt — not in a subfolder. You can generate a finished file for free with the llms.txt generator.
Is llms.txt the same as robots.txt?
No. robots.txt tells crawlers what they may not fetch; llms.txt tells AI models what is worth reading, and in what order. They complement each other — and the AI crawlers need access in robots.txt to reach your llms.txt.
Is llms.txt enough to become visible in AI answers?
No. llms.txt is one signal among several. Whether AI mentions you also depends on structured data, content citability, your brand entity and crawler access. But it is one of the fastest to complete — and most competitors don't have it yet. See the GEO guide for the full picture.
Does it cost anything to make an llms.txt?
No. The generator is free and requires no login. If you want it published and kept up to date automatically alongside your other AI signals, the Geoa platform does that from DKK 299/mo.
Who invented llms.txt?
Jeremy Howard proposed the format in September 2024. The specification lives at llmstxt.org and reached version 2 in August 2026.
Does Google use llms.txt?
Google Search has not said it uses the file, and Google's John Mueller has called the question of its effect speculative. On the other hand, Chrome's Lighthouse checks whether the file can be fetched, and Google itself publishes llms.txt for its developer documentation.
What is llms-full.txt?
An optional file with the whole site's content combined as Markdown in one file. It makes most sense for documentation and knowledge bases — an ordinary company website is fine with llms.txt.
Should llms.txt be listed in robots.txt?
No. The file lives at its fixed address, just like robots.txt itself. What matters is that robots.txt does not block the AI crawlers that need to fetch it.
How often should llms.txt be updated?
Whenever your most important pages change — a new service, a new pricing page, a deleted page. An outdated link in the file is worse than no link.
Make your llms.txt in one minute
Free generator — no login, ready to place in the root of your domain.
Open the llms.txt generator →