In the SEO field, the lms.txt file is increasingly being discussed, proposed to optimize websites in their relationship with generative AIs like ChatGPT and Copilot. But is it really useful or just another passing trend? In this article, we analyze what it is, how it works, what its supporters say, and what the most common doubts are.
What is the lms.txt file
The lms.txt file (often also called llms.txt, where LLM stands for Large Language Models) was proposed in 2024 by researcher Jeremy Howard (Answer.ai project). The goal? To provide artificial intelligences with a simplified and well-structured version of a website’s main content, making it easier to understand when generating responses.
What is it for
The lms.txt file is not intended for traditional search engines like Google, but for models like ChatGPT, Claude, Gemini, Copilot, and others. It is a Markdown document that concisely and clearly summarizes the key sections of the site. Instead of analyzing every HTML page full of superfluous elements, the AI finds in lms.txt a sort of “conceptual map” that guides it to the most relevant content.
What it contains
- A concise description of the site
- Links to the most important pages, with brief explanations
- Hierarchical structure (titles, lists) to facilitate parsing by the AI
It is a completely opt-in file: it doesn’t exclude anything (like robots.txt does), but suggests to the AI what to read to better understand the site.
Advantages according to proponents
- Better understanding by AIs: generated responses are potentially more accurate and up-to-date
- Communicative consistency: prevents the AI from inventing or misunderstanding brand messages
- Efficiency: reduces computational load for the AI, facilitating content access
- Preparation for the future: if it becomes a standard, those who have already implemented it will have an advantage
The strongest criticisms
- Nobody reads it: currently, no official AI engine (ChatGPT, Bing, Google) uses it
- Does not improve ranking: Google does not consider it, nor does it improve classic SEO
- Terrible user experience: if an AI directly quotes from the file, the user lands on a raw Markdown page
- Easy to manipulate: being auto-generated, it can be used to “inflate” keywords or insert spam
Who is using it
Some tech companies and open-source projects are experimenting with the llms.txt file. Among them are ElevenLabs, WordLift, LangChain, Hugging Face, and others. Some online directories collect sites that have already implemented it, but at the moment they remain isolated cases and for exploratory purposes.
Is it really needed?
It depends. If you manage a site with a technical knowledge base or API documentation, it might be worth testing. But if you have a blog, an e-commerce store, or a magazine, lms.txt today will not bring you more visibility, traffic, or better rankings.
The AI industry has not yet adopted this standard. Without an official statement from OpenAI, Google, Microsoft, or others, the usefulness of this file remains theoretical.
Conclusion
lms.txt is a fascinating but still immature idea. It could represent the future of AI optimization, or it could end up forgotten like the old meta keywords tag. For now, it’s best to observe its evolution and focus on quality content, structured data, and solid SEO.
Click here for the PDF (internal use)
Frequently asked questions about lms.txt
What is the lms.txt file?
It is a text or Markdown file designed to make it easier for generative artificial intelligences to read a site’s content. It is not intended for users or traditional search engines.
Where should it be placed?
In the site’s root directory, just like robots.txt or sitemap.xml, with URLs like https://www.tuosito.it/lms.txt or llms.txt.
Do Google or Bing read it?
No. Currently, neither Google, Bing, nor ChatGPT use or read the lms.txt file. No major AI player has officially adopted it.
Does it affect SEO?
No. It does not improve search engine rankings. It is a tool oriented towards artificial intelligence, not classic SEO.
Is it worth implementing?
Only if you want to experiment for the future, especially if your site provides technical content, guides, or documentation. For most websites today, it has no concrete impact.
Be the first to comment