Create llms.txt and implement it: step by step with an example

You know what llms.txt is and you have decided to create one. Good, then this is your practical worksheet. Below you walk step by step through creating and implementing an llms.txt, with a working example you can copy almost as is. No developer needed, no tooling, no plugin. But there are a few choices that make the difference between a file that neatly points AI models in the right direction and a file that only adds noise.

Do you want to understand first what llms.txt actually does and whether it is worth it for your brand at all? Then read the explainer: what llms.txt is and whether you need one, or our brief explanation of how llms.txt guides AI models through your site. On this page we assume that question is settled and we focus entirely on execution.

Before you work through the individual steps, here is the complete route at a glance. From an empty file to a checked, working llms.txt takes five steps:

What you need before you start

To create llms.txt you only need three things: a text editor (Notepad or VS Code will do), an overview of your most important pages, and access to the root of your website to upload a file. That last one is usually the only point where you need someone from your web team, and even that is, in most CMSes, a matter of putting a file in the right place.

First make a shortlist of the pages that genuinely matter to an outsider. Think of your service pages, your key knowledge articles and your about page. Not your cookie policy, not every tag page, not your internal search results. The whole idea of llms.txt is that you offer a clean, curated menu, so be strict in your selection.

Step 1: create the markdown file

Open your text editor and create a new file with exactly the name llms.txt. Watch the lowercase letters and the correct extension. No LLMS.txt, no llms.md, no llms.txt.txt that Windows quietly turns it into. You write the content in markdown, the same light formatting format you know from README files.

The structure the convention prescribes is deliberately simple: an H1 with your brand name, a blockquote that sums up in one sentence who you are and who you serve, and below that a number of sections with links. Every link gets a short description so a model immediately understands what the page is about.

Step 2: write the header and the summary

Start with your brand’s identity. The H1 is your name, the blockquote underneath is your positioning in one clear sentence. This is not marketing prose but a factual description that helps a model place your entity correctly.

# Your Company
> B2B growth partner that helps brands in the Benelux score in Google and get cited in AI answers.

Keep it factual and concrete. Who you are, what you do, for whom, and in which region. A model that later has to summarise your brand in an answer happily leans on exactly that kind of clean sentence.

Step 3: add your sections and links

Below the header, split your most important pages into logical sections. Use H2 headings such as “Services”, “Knowledge” and “About us”, and put a bulleted list underneath with full links plus a short description behind each link. Always use absolute URLs, so including your full domain, so there is no doubt about where a page lives.

# Your Company
> B2B growth partner that helps brands in the Benelux score in Google and get cited in AI answers.

## Services
- [SEO](https://yourdomain.be/services/seo/): findability in Google and AI, steered on pipeline.
- [AI search / GEO](https://yourdomain.be/services/geo-agency/): visibility in ChatGPT, Perplexity and Google AI.

## Knowledge
- [What is SEO](https://yourdomain.be/blog/seo/): the complete explanation for B2B brands.
- [B2B SEO](https://yourdomain.be/blog/b2b-seo/): how to win customers with a long sales cycle.

## About us
- [About Your Company](https://yourdomain.be/about-us/): who we are and how we work.

This is a complete, working example. Adjust the names, links and descriptions to your own situation and you have a usable llms.txt. The craft is not in the formatting but in the selection: ten pages that genuinely matter beat a hundred that clutter your menu.

Step 4: place the file in the root of your domain

This is where things go wrong most often in practice. The file has to be reachable at yourdomain.be/llms.txt, so in the root of your domain, not in a subfolder. How you do that depends on your platform: on a static site you upload the file to the main folder, in many CMSes you put it in the public or root directory, and some hosting environments have a separate place for this type of file next to your robots.txt and sitemap.xml.

Not sure where that is? Look at where your robots.txt sits, because that is where your llms.txt belongs too. If you find robots.txt at yourdomain.be/robots.txt, then llms.txt has to sit at exactly the same level.

Step 5: check that the file is served correctly

Creating is not the same as working. After uploading, open your browser and go to yourdomain.be/llms.txt. You should see the plain text of your file, without a download starting and without a 404 error. If your content shows up neatly, the implementation is technically successful.

Watch one more detail: the file is ideally served as plain text (text/plain). If it comes in as a download or as HTML, your hosting configuration may be throwing a spanner in the works. In most cases it is right straight away, but it is a ten-second check that saves you headaches later.

Do not confuse implementing with managing crawler access

An important nuance, because this is where people regularly get stuck. Creating an llms.txt arranges nothing about access. If you want to steer which AI bots are allowed to fetch your content at all, you do that through your robots.txt and crawler directives, a very different exercise that we worked out separately in our guide on blocking AI bots via robots.txt. llms.txt only improves how a model understands the content you already allow; it keeps nobody in or out.

Keep the file alive

An llms.txt that you create once and then forget quickly loses its value. If you launch a new service, restructure your knowledge section or change URLs, your llms.txt has to follow. A file full of dead links and outdated descriptions gives AI models exactly the wrong picture of your brand. Set a recurring reminder, quarterly for instance, to run through the file.

And stay realistic about what it delivers. llms.txt is not a ranking factor and no guarantee of mentions in AI answers. It is a tidy, cheap support layer on top of the real work. The heavier levers for AI visibility sit elsewhere, in the five core indicators of AI visibility that determine whether a model recognises and recommends your brand. You build a strong entity, for example, with a Wikidata item for your company, which helps AI models place you correctly. An llms.txt without a strong entity and citable content is a beautiful menu for an empty restaurant.

From file to real visibility

See creating and implementing llms.txt as one small tick in a much bigger story. The real work sits in content that AI models like to cite, a consistent entity and a site that scores on buying intent instead of on vanity visitor numbers. That is what we steer on as an SEO specialist: findability that delivers leads, not a checklist of technical files. SEO and AI visibility are, to us, two sides of the same acquisition layer, not separate little projects.

Do you want to understand how it all fits together? Start with the pillar what is SEO for the full picture, and then read how B2B SEO wins you customers with a long sales cycle. That is where the gain sits, not in a little text file.

Are you wondering whether your brand shows up today in Google and in AI answers, and which moves genuinely feed your pipeline? We are a small team that moves fast and tells you honestly when something is not worth it.

Book your free intake

Frequently asked questions about creating llms.txt

Do I need a developer to create llms.txt?

No. You create the file in an ordinary text editor and filling it in takes you five minutes. The only moment where you could use a hand is uploading it to the root of your domain, and even that is easy to do yourself in most CMSes.

Where exactly should I place the llms.txt file?

In the root of your domain, so that it is reachable at yourdomain.be/llms.txt. A handy rule of thumb: put it at exactly the same level as your robots.txt. If that file sits at yourdomain.be/robots.txt, then llms.txt belongs there too.

How do I check whether my llms.txt works?

Open yourdomain.be/llms.txt in your browser. If you see the plain text of your file without a 404 and without a download starting, the implementation is technically fine. Ideally it is served as plain text.

Does llms.txt control whether AI may use my content?

No. llms.txt handles readability, not access. If you want to steer which AI bots may fetch your content, you do that through your robots.txt and crawler directives, not through llms.txt.

How often should I update my llms.txt?

Run through the file at every larger change to your services, structure or URLs, and set a reminder, quarterly for example, to check it. A file with dead links and outdated descriptions gives AI models a wrong picture of your brand.