Some of the posts you read about AI visibility probably tell you to create an llms.txt file. Some even present it as the first thing to do. In this post we briefly explain what llms.txt is, what Google says about it, and how much effort you should put into the file. Here is the short answer up front: it is not needed for Google, and for the other engines we found no document that shows it has an effect.
What is llms.txt?
llms.txt is a text file written in Markdown, usually placed in a site's root directory. According to the proposal on llmstxt.org, its purpose is to present a site's content to AI models and agents in a clean, short form. The reasoning behind the proposal is this: web pages are designed for people. Menus, ads and complex page structure can make it harder for an AI tool to read the content efficiently.
The file usually contains the site's name, a short summary, and links to important pages with short descriptions. The proposal was first published in September 2024. According to the page on llmstxt.org, the proposal was updated with its second version in August 2026.
There is an important detail here: llms.txt is a proposal. It is not a rule that search engines have followed for years, like robots.txt. Whether an AI engine reads this file or not is that engine's own decision.
Are llms.txt and robots.txt the same thing?
No. The two files have similar names but do different jobs:
- robots.txt tells bots which parts of the site they can enter. OpenAI, Perplexity and Google explain in their own documents how their search bots follow this file.
- llms.txt is not a permission file. It does not stop a bot, and it does not allow a bot either. It only offers a guide that summarizes the site's content.
So a block in robots.txt cannot be removed by adding llms.txt. First you need to make sure your robots.txt and security settings are correct.
What does Google say about llms.txt?
Google is clear on this. Its AI features document says you do not need to create new machine-readable files, AI text files or special markup to appear in AI Overviews or AI Mode.
Google's guide for AI features is even clearer:
- You do not need to create new machine-readable files, AI text files, markup or Markdown to appear in Google Search.
- Google Search does not use these files.
- It is fine to create llms.txt or similar files for other services. But this neither hurts nor helps your visibility or ranking in Google Search, because Google Search does not take these files into account.
According to the guide, this also applies to Google Search's AI features. So, by Google's own statement, llms.txt has no effect on the Google side, including AI Mode and AI Overviews. We explained how Google runs these features in what is Google AI Mode and what has changed for brands?.
Do ChatGPT and Perplexity read llms.txt?
In OpenAI's and Perplexity's crawler documents, how they reach sites is explained through robots.txt and bot names. OpenAI's document says OAI-SearchBot is used to show sites in ChatGPT's search features. Perplexity's document recommends allowing PerplexityBot to appear in search results.
We found no statement in these documents about the effect of llms.txt on appearing in search answers. So we have no basis for saying "if you add llms.txt, ChatGPT or Perplexity will mention you more." If you see a post that says this, ask for its source.
So how much effort should you put into llms.txt?
The honest answer: very little. The file does no harm. If you already have a tidy list of pages, it can be prepared in a few minutes. But it is not the first thing that will raise your AI visibility.
These are the tasks whose effect is clearly stated in the documents and that give more back for your effort:
- Make sure AI search bots can enter your site. If OAI-SearchBot or PerplexityBot is blocked, llms.txt does not get around that block. We covered this in does blocking GPTBot affect visibility in ChatGPT?.
- Check that your pages are indexed in Google and eligible to be shown with a snippet. This is the basic technical requirement Google sets for its AI features.
- Make sure your service pages say clearly in the first paragraph what you do, who you serve and where you are.
- Find the list pages the engines cite as sources and check whether your name appears there.
If you still want to create one, what should it look like?
If you want to create an llms.txt for other tools, or just to keep things tidy, the proposal on llmstxt.org describes a simple structure:
- At the top, the site's name, as a single main heading in Markdown.
- Below it, a short summary of what the site is, as a blockquote.
- Then, under subheadings, links to important pages, each with a one-sentence description.
A few points to keep in mind when you create it:
- The usual place for the file is the site root, that is, the /llms.txt address. The proposal also allows subfolders.
- Include only pages and information that actually exist on the site. Writing a claim in this file that is not on the site damages trust here, just as it does anywhere else.
- Update the file when you add or remove pages. A file full of outdated links is of no use.
- After creating the file, do not feel as if you have done the other work. The tasks backed by documents in the list above still come first.
How do you know what is working?
The way to separate cosmetic tasks from measured work is to measure. After you make a change, ask the same question list to the engines again and see whether the result changed. Moving to the next task without seeing the effect of a change means not knowing where your effort went.
The technical check on AIShortlist's Site screen looks at bot access, robots.txt, the sitemap, titles, meta descriptions, schema and question headings. For each finding it writes what to do. On the measurement side, it asks the questions you approve to ChatGPT, Gemini and Perplexity. With the free trial you can run a measurement for 1 brand and 10 questions and also see the site check. No card is required.