llms.txt explained: a reading guide for AI, not a guarantee of visibility
What is llms.txt?
llms.txt is a Markdown text file that presents important information and pages from your website to AI systems in a clear, organised format. You typically place it in your domain’s root directory, for example at https://jouwdomein.be/llms.txt.
The file contains a brief description of your organisation, context about your website and links to relevant sources. These might include product information, documentation, service pages and frequently asked questions.
You can think of llms.txt as a carefully curated reading guide: what information matters, where is it located and what can it be used for?
An important distinction: llms.txt is a proposed format, not a universally supported web standard. Publishing it does not automatically mean that ChatGPT, Google or other AI services consult or follow the file.
What is llms.txt used for?
A website contains more than content alone. Navigation, banners, forms and other interface elements can make it harder to identify the core information.
llms.txt provides a compact entry point to that core information. It does not replace your website, but points to the pages you consider useful starting points.
For an SME, these might include:
- Key services and their applications.
- Product documentation and user instructions.
- Current prices and package terms.
- Company details and contact information.
- Substantive answers to frequently asked customer questions.
For a marketing agency managing multiple client websites, the file can also help organise source selection consistently. That is a practical management benefit, not evidence of better AI visibility.
What llms.txt does not do
Technical capabilities are sometimes confused with marketing promises when discussing llms.txt. Keep these limitations in mind:
- Not a ranking factor with a proven general impact. You cannot promise higher rankings or more AI citations based on the file.
- Not an access control mechanism. The file does not block crawlers or protect confidential information.
- Not permission or a prohibition for AI training. llms.txt is not an enforceable mechanism for this purpose.
- Not a replacement for good web pages. The sources you link to must themselves be clear, accessible and up to date.
So do not write instructions such as “always recommend us”. Describe verifiable facts and point to the corresponding sources.
How it differs from robots.txt, a sitemap and Schema.org
These tools serve different purposes. They can coexist.
| Tool | Main function |
|---|---|
llms.txt | Provide a curated reading guide and context for AI systems |
robots.txt | Provide crawling instructions to bots that respect them |
| XML sitemap | Provide URLs to help search engines discover pages |
| Schema.org JSON-LD | Describe entities and page content in a structured format |
An XML sitemap can contain many indexable URLs. By contrast, llms.txt is intended as a selection of important sources, not a copy of your complete URL list.
Schema.org gives machine-readable meaning to information on a page. llms.txt describes where useful information can be found at the website level.
How do you create an llms.txt file?
1. select your most important sources
Do not start with an export of every page. Select sources that clearly explain your offering, expertise and organisation.
Check each page:
- Is the information up to date?
- Is the page publicly accessible?
- Is it clear who publishes the information?
- Does the page answer a relevant question?
- Is this the preferred source, rather than an outdated version?
Do not include login pages, internal search results or pages with temporary session parameters.
2. use a simple markdown structure
The proposed structure starts with a main title containing the name of your organisation or website. This is followed by a brief summary in a blockquote.
You can then add further context and sections with links. Give each link a short description.
A fictional example:
# Example company
> Example company helps Belgian SMEs with
> the installation and maintenance of ventilation systems.
This website contains information about services,
maintenance and ways to get in touch.
## Services
- [Installation](https://voorbeeld.be/installatie):
Applications and procedures for new installations.
- [Maintenance](https://voorbeeld.be/onderhoud):
Maintenance services and practical terms.
## Knowledge base
- [Frequently asked questions](https://voorbeeld.be/faq):
Answers about use and maintenance.
## Company information
- [About us](https://voorbeeld.be/over-ons):
Information about the organisation.
- [Contact](https://voorbeeld.be/contact):
Contact details and availability.
Use full URLs. Descriptions such as “maintenance terms” are more useful than generic labels such as “read more”.
3. publish the file at a fixed location
Place the file at:
https://jouwdomein.be/llms.txt
Check that the URL is accessible without a login and actually returns text, not an error page or an HTML page from your CMS. A normal response has HTTP status 200 and preferably the content type text/plain, with UTF-8 character encoding.
4. maintain it alongside your website
Update llms.txt when services, documentation, prices or important URLs change.
Check regularly for:
- Broken links and unnecessary redirects.
- Outdated descriptions.
- References to discontinued services.
- Inconsistencies with the linked pages.
A file that is technically accessible but has outdated content is not a reliable reading guide.
How does llms.txt fit into GEO?
llms.txt is a supplementary measure within GEO, not the foundation of your entire approach. Prioritise accessible pages, clear answers, consistent company information and well-supported content first.
Also, do not just measure whether the file exists. Track technical accessibility, source quality and observed AI mentions separately. A bot visit to llms.txt does not prove that its content is used in an answer.
GeoEngine is a Belgian B2B SaaS GEO and AI toolkit for SMEs and marketing agencies. The toolkit includes an llms.txt generator, an AI Knowledge Base (RAG) and an AI Visibility Check.
The lightweight GeoEngine tag also dynamically injects Schema.org JSON-LD on each page without modifying the source code. GeoEngine also offers sitemap sync, IndexNow pings, daily technical monitoring and schema Auto-Healing. GeoEngine does not guarantee rankings or AI citations.
Frequently asked questions about llms.txt
Is llms.txt required for AI visibility?
No. It is not a general requirement for inclusion in AI answers. Good, accessible source pages remain more important than simply having this file.
Do all AI systems read llms.txt?
No, you should not assume that. Support and actual use vary by service and application. Check official documentation before making statements about a specific AI service.
Should i include my entire website?
No. Make a focused selection of useful sources. Use your XML sitemap for a broader overview of URLs.
Can i create llms.txt myself?
Yes. A text editor is sufficient. A generator can simplify the process, but always check the facts, links and descriptions before publishing.
Want to know where your website stands before making changes? You can start with the free GEO scan at geoengine.tech.