Almost certainly not. Google states plainly that its Search systems ignore the file, neither OpenAI nor Anthropic asks site owners for one, and a month of server logs found that 97 out of every 100 of these files were never requested at all. Adding one to a plumbing or roofing site is cheap and harmless, but there is no published evidence it gets you named anywhere.
What an llms.txt file actually is
The idea came from one person, not from a search engine and not from an AI lab. Jeremy Howard published the proposal at llmstxt.org on September 3, 2024. The page now carries a second version of it, written after two years of adoption, and says it was last modified on August 10, 2026.
It asks you to put a plain text file at the root of your site, written in Markdown. Inside goes a heading with the name of the business, a short summary under it, and tidy lists of links to the pages that matter. The reasoning on that page is that “web pages are built for people”, with the useful part wrapped in navigation and ads, so a machine has to dig for it. One curated file would hand it over instead.
That is a sensible idea. It is still a proposal.
The name is what fools people. robots.txt is a real file, first defined in 1994, and both OpenAI and Anthropic tell site owners to use it. llms.txt is a lookalike that nothing is obliged to open.
Google says Search ignores it
Google has answered this question in writing, and the answer is no.
Its Search Central guide, optimizing your website for generative AI features on Google Search, last updated July 10, 2026, says: “You don’t need to create new machine readable files, AI text files, markup, or Markdown to appear in Google Search”.
That same page names llms.txt directly. It says it is “completely fine” to keep one for other services or systems that use such files, then adds: “Doing so will neither harm nor help your site’s visibility or rankings in Google Search, as Google Search ignores them.”
Google’s separate page on AI features and your website, last updated December 10, 2025, says the same about the AI answers on top of the results: “There are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary.” That page does not mention llms.txt at all.
Read those two together and the position is clear. Neither harm nor help.
Almost nothing is asking for the file
This is the part with a number you can go and check.
Ahrefs published a study on June 15, 2026 by Louise Linehan and Xibeijia Guan. It looked at all 137,210 domains in Ahrefs Web Analytics that received traffic in May 2026. Of those, 28% had published a valid llms.txt file. Of the files that existed, 97% were requested zero times that month. Not rarely. Zero.
Among the 3% of files that were fetched at all, most of the fetching was not an AI assistant reading your business. SEO audit tools accounted for 21.7% of requests, and other or unidentified bots another 14.9%. OpenAI’s GPTBot made 4.51%. The bots that go and fetch pages to answer a live question made up 1.1%.
So the most common reason anything opens an llms.txt file is an audit tool checking whether you have one.
The AI companies point you at a different file
Look at what the labs tell site owners in their own documentation, and llms.txt is not the answer they give.
OpenAI’s developer page on its crawlers says its robots.txt rules are how site owners manage what its bots do with their content. That page carries no update date. Anthropic’s support article on ClaudeBot, dated April 7, 2026, says the same, and gives the exact robots.txt lines to use. It never mentions llms.txt once.
There is a wrinkle worth knowing. OpenAI does publish an llms.txt for its own developer documentation, and so does Anthropic. That is often quoted as proof the format works. Publishing one for your own docs and reading one on somebody else’s site are two different things, and only the second one would help you.
The file those companies do read on your site is robots.txt, and getting that wrong can quietly keep an assistant out. That is worth checking on its own, and it is covered in whether your site is blocked from ChatGPT.
What this evidence does not prove
Ahrefs says so itself. Its sample is drawn from its own analytics customers, and the study warns that they “skew more technical and SEO-aware than the web at large, so treat the 28% adoption figure as an upper bound”. That caveat cuts against adoption being high. It does not change the zero request finding.
One month of logs is also one month of logs. Zero requests in May 2026 does not prove that no system will ever read the file. Smaller AI tools may read it now without saying so. The proposal could be picked up later, and if it is, this article will be out of date.
What can be said is narrower. Nothing Google, OpenAI, or Anthropic publishes for site owners promises to read it, and the request logs match that silence.
Here is the honest part. If somebody has quoted you a monthly fee to write and maintain an llms.txt file, or sold it as the way to get into AI answers, that is a charge for something none of those companies has published a commitment to open, and a monthly fee should list the work it paid for. Do not pay for it. If you want one anyway, it is a text file you can write yourself in ten minutes, and it will sit there costing nothing.
What to do this week instead
Four steps, in order, and none of them cost money.
- Type your own domain followed by /llms.txt into a browser. If text appears, you already have one. Leave it. If nothing appears, nothing is missing.
- Type your domain followed by /robots.txt. This is the file the crawlers really read. If you cannot make sense of what is in there, that is worth an hour of somebody’s time.
- Open your own access log and look for the lines naming GPTBot or ChatGPT-User. That is your version of the study above, and it will show you whether anything ever asked for /llms.txt. A name in a log is easy to fake, so check the address the visit came from before you count it.
- Ask an assistant the question your customers ask, such as who fixes burst pipes in your town, and write down which companies get named, the way this ten minute check does it. Do it twice, on different days.
That last step is the one most owners skip, and it is the only way to see where you actually stand today. There is a free scan that runs it for you, asks the questions a customer would ask, and writes down who gets named. It is a report, not a fix, and a person reads it before it goes out.
If the answers come back with your competitors named and not you, the work that follows is ordinary and slow. Pages that answer real questions, a clean site an assistant can read, and a listing that matches. That is the job of search checks, and no file at the root of your site is a shortcut around it.