What is llms.txt?+
llms.txt is a plain text file in Markdown at the root of your domain, at https://yourdomain.com/llms.txt. It tells a large language model who you are, what you offer and where the substance on your site lives. Jeremy Howard of Answer.AI proposed it in September 2024. Adoption across the top ten thousand sites rose fivefold in twelve months, from 1.04 percent in July 2025 to 5.61 percent in June 2026.
Do AI crawlers read llms.txt?+
Our own server logs record them doing so. OpenAI fetches these files through GPTBot, OAI-SearchBot and ChatGPT-User, and Anthropic through ClaudeBot and Claude-User. The other crawlers in these logs are Google-CloudVertexBot from Google, Amazonbot from Amazon and DeepSeekBot from DeepSeek. We log the user agent, path, address and timestamp of each bot request to the sites we run, and more of these crawlers turn up every month.
What does a good file change?+
A good file changes how accurately an engine describes you when it answers a question you should be the answer to. It states your identity, your catalogue, who each part is for and the situations you are the right recommendation in, all in one place a model can read in a single fetch. The engine can then use your own description of the business in place of a summary from whichever page a crawler landed on.
Why buy one instead of generating it myself?+
A generator takes your homepage and your sitemap as input and outputs a list of links. We read your pages and extract your entities, offers and audiences. We then write the recommendation logic stating when a model should name you, and we verify every URL live before adding it. The value lies in that research, because anyone can copy the format in an afternoon. The same run includes a full SEO and GEO audit of your site.
What do I receive?+
You can download the complete llms.txt as .md or .txt, and we email you a copy to keep. You also receive the headline result of the full SEO and GEO audit from the same run: the number of issues and how severe they are. We verify every URL in the file live before writing it in.
How much does it cost?+
One complete run costs a single payment of twenty five dollars or twenty euros in the currency you choose. Buying a run does not require a subscription or an account. The price covers both the researched llms.txt and the full SEO and GEO audit from the same run. If the run fails, you are refunded automatically.
How long does a run take?+
A run lasts as long as your site needs. We read your pages with no sampling, so a large site takes longer than a small one, and we would rather not quote you a figure we cannot hold to. You can close the tab: the run continues on our infrastructure, and the finished file is waiting when you return and in your inbox.
How many pages do you read?+
We read your site and place your commercial and editorial content first. Discovery builds a full URL inventory from robots.txt, sitemap.xml and nested sitemap indexes, or from a link crawl when there is no sitemap. We classify every URL by kind before the deep read begins, and a large site takes a little longer than a small one.
Is this a template with my details filled in?+
No template is involved. On every run the agent fetches your robots.txt and sitemap and builds a full URL inventory. It then reads your pages in parallel, extracts your entities and offers and composes the file from its findings. Each file follows the same fifteen section structure and takes its content from the pages of the site, so two different sites get two different files.
Do you verify the links you put in the file?+
We fetch every URL and confirm it resolves before we write it into the file. An engine that follows your file reaches a real page. We run this check because a model stops relying on a file after two dead links. The check takes place as the agent writes the file section by section and covers every URL in the finished llms.txt.
What language is the file written in?+
We write the file in your site's own language. If your site is in French, the file is in French with correct accents. The file contains the words your business uses. The agent writes every section in the language of your pages, whatever that language is.
What is the difference between llms.txt and llms-full.txt?+
llms.txt states who you are, what you offer and where things are. llms-full.txt inlines your content so a reader gets everything in one fetch. This tool writes your llms.txt. A good llms.txt links a reader to the pages that hold your substance. In our logs llms-full.txt gets fetched more than llms.txt does.
How is this different from robots.txt and sitemap.xml?+
robots.txt sets permission, and it is the one file the whole industry honours. sitemap.xml holds the inventory of URLs that exist. llms.txt gives orientation: what this place is and what matters in it. llms-full.txt, the fourth file at your domain root, inlines your content. All four have to stay consistent with each other, and our audit checks that. We often find sites that block crawlers in robots.txt and then publish an llms.txt.
What does the SEO and GEO audit cover?+
The audit runs a fixed, versioned checklist, so two runs are comparable. It covers indexability and crawler permissions, canonicals, titles and descriptions, heading structure, structured data coverage and validity, internal linking depth, orphan pages, image alt coverage, language and region signals, sitemap health, AI crawler access, entity grounding, citation readiness, answer shaped content and freshness signals.
Why do you not show me the full audit findings?+
You get the number of issues and how severe they are. We record each finding with a code, a severity, its area, the evidence we saw and what to do about it, and an adviser who has read your site goes through them with you. That call covers what each issue is and the order to fix them in. It lasts twenty minutes and is free to book from the result screen.
Which AI engines have you seen fetch these files?+
Our server logs name five operators. OpenAI fetches through GPTBot, OAI-SearchBot and ChatGPT-User, and Anthropic through ClaudeBot and Claude-User. The other crawlers in these logs are Google-CloudVertexBot from Google, Amazonbot from Amazon and DeepSeekBot from DeepSeek. New ones keep appearing. On 25 July we logged a Claude-User fetch carrying a coding agent user agent, which means a developer had pointed a tool at the documentation.
Do I need technical skills to use it?+
You need no technical skills. You get two finished text files, and putting them live means uploading them to the root of your domain, which takes fifteen minutes. On most platforms you drag them into your public folder. If your team would rather we did it, an adviser can sort that out on the call, which you book from the result screen.
Will it work for a site that is not in English?+
It does, because we write the file in your language with no translation step. A French site gets a French file with correct accents. The file uses the words of your business. The engines answering French questions read French. The agent composes every section in your site's language and verifies every URL live before committing it.
What makes yours different from the free generators?+
A free generator takes your homepage and your sitemap as input and outputs a list of links that an engine already had. We read your pages and write your identity and registration, your catalogue with live verified URLs, who each area is for, the situations you are the right recommendation in and how to cite you correctly. The fetches come from that research.
Where do I put the file once I have it?+
Put it at the root of your domain so that its address is https://yourdomain.com/llms.txt. On most platforms that means dropping it in your public or static directory, which takes fifteen minutes. If your framework serves routes instead of files, generate it from a route so the content stays bound to your real data. Your server or CDN access logs then show which crawlers request it.
How often should I regenerate it?+
Regenerate it whenever your offering changes materially, and otherwise a few times a year, so the engines read the current version of your business. A file written once and never verified accumulates dead links as URLs move and products get retired. Each regeneration is one run and one payment. The versioned audit checklist lets you compare two runs months apart.
Do you store my site content?+
We keep the files and the audit from the run for your return visit and for the email we send you. We do not resell your content or use your site to train anything. All of this works without an account. The email contains a permanent copy of the file.
What happens if the run fails?+
You are refunded automatically, because the promise is a finished deliverable and a failed run delivers none. We tell you what happened. The price covers one complete run with one payment, no subscription and no account, so you have nothing to cancel after a failed run.
Can I use the file commercially and modify it?+
The file is yours outright. You can extend it with your own edits and regenerate it, and we encourage you to keep it under version control like code. Agencies hand it over to clients as their own deliverable. Regenerate it whenever your offering changes materially, so your edits stay on a current version of your business.
Can I run it for a client site?+
Agencies do. Each run is one site and one payment, with the files delivered to whichever email address you give us, so you can run it for a client and hand over both files as your own deliverable. If you are doing this at volume, talk to us and we will sort out a better arrangement than paying for twenty separate runs.
What if I already have an llms.txt?+
We fetch it first, so we know what you are currently telling the engines. The run then replaces it with a version built from your pages. The audit reports errors in the old file, such as URLs that have since moved. We verify every URL in the new file live before writing it in.
Who built this?+
La Boétie, a technology studio that builds AI and infrastructure systems and runs its own content properties, built this tool. The evidence quoted on this page comes from our own production logs, where named AI crawlers fetch these files week after week. Your file follows the same structure as the file we run on our own sites.