Short answer. Not for getting recommended, on the evidence so far. Ahrefs found 97% of llms.txt files got no requests at all in a month, SE Ranking found no link between having one and being cited by AI, and Google says the file isn't needed for its AI features. It may help AI coding tools read documentation.
Some agencies and tools now offer to write an llms.txt file for your website, as a way to help ChatGPT and other assistants find and recommend you. It is a reasonable-sounding idea. The evidence so far says it doesn't do that.
What llms.txt is
llms.txt is a proposal published by Jeremy Howard in September 2024. The idea is to put a plain Markdown file at yoursite.com/llms.txt that summarises the site and links to its most useful pages, written for language models rather than people. The stated aim is to help AI agents use a website, for example a coding assistant reading a software library's documentation.
It isn't a standard adopted by any of the big AI companies. It is a convention that anyone can follow, and whether it does anything depends entirely on whether the AI systems choose to read it.
Ahrefs: 97% of files got no requests at all
The clearest evidence comes from server logs. In June 2026 Ahrefs analysed 137,210 sites that use its web analytics, and looked at who actually requested their llms.txt files in May 2026.
- About 28% of the sites, roughly 38,000, published an llms.txt file. Ahrefs notes this is an upper bound, because its customers lean technical.
- 97% of those files received zero requests in the month. The remaining 3%, about 1,100 sites, accounted for all of the traffic measured.
- Of the requests that did happen, 96% came from bots. Named AI bots made 19.5% of requests, split across AI agents, training crawlers, assistants and retrieval bots. GEO and AEO tools, the kind that check whether your file exists, made 5.8%.
- AI bots made no requests at all for llms.txt files that don't exist. In other words, they weren't routinely checking for one.
Ahrefs is careful to call its numbers ceilings: a request shows that something fetched the file, not that anything used it.
Google: "comparable to the keywords meta tag"
Google has been consistent. In April 2025, Google's John Mueller compared llms.txt to the keywords meta tag, the old HTML tag that search engines stopped trusting years ago because site owners could write anything in it. He added that server logs show the AI services don't even check for the file.
In June 2026 he went further. A file a site writes about itself can't help an AI decide which sites to show, because every site would describe itself favourably. He allowed it a narrow role: once an automated tool is already working on your website, a guide to the site might help it.
According to Ahrefs, Google's own guidance on generative AI features in Search also says files like llms.txt aren't needed to appear in them.
SE Ranking: no link to being cited
Requests are one question. Citations are the one that matters for a business. SE Ranking crawled around 300,000 domains and found llms.txt on about 10% of them. It then tested whether having the file was linked to how often a domain was cited by AI assistants. It found no meaningful relationship, and its prediction model got slightly more accurate when the llms.txt signal was removed (Search Engine Journal, November 2025).
What the AI companies say
We looked for documentation from OpenAI, Anthropic and Perplexity stating that their search crawlers use llms.txt to choose what to cite. We didn't find any. OpenAI's crawler documentation explains how OAI-SearchBot and its other crawlers work and how robots.txt controls them, and says nothing about llms.txt as a ranking or citation input.
That could change. For now, nobody who runs a major assistant has said the file affects whether you get recommended.
The evidence in one table
| Question | Finding | Source |
|---|---|---|
| Do AI systems request llms.txt? | 97% of files got zero requests in a month | Ahrefs, June 2026 |
| Do AI bots look for it when it's missing? | No requests for files that don't exist | Ahrefs, June 2026 |
| Does it affect AI citations? | No measurable link across about 300,000 domains | SE Ranking, November 2025 |
| Does Google use it? | No; compared to the keywords meta tag | John Mueller, April 2025 |
| Do OpenAI, Anthropic or Perplexity document using it? | We found no such statement | Their crawler documentation |
Where it might still help
There is one honest use case, and it is the one the proposal was written for. If your product has documentation that developers use with AI coding tools, a well-made llms.txt can help those tools find the right pages once they are already on your site. Mueller's narrow allowance points the same way.
For a business that wants to be recommended by ChatGPT when a customer asks for a supplier, that isn't the problem llms.txt solves.
What works instead
The same evidence points at what does matter:
- Let the search crawlers in. OpenAI says sites that block OAI-SearchBot won't appear in ChatGPT search answers. Check your
robots.txtand any security plugin for blocks on AI crawlers. - Publish pages that answer the question. The assistants cite pages they find through search. A clear page that says what you do, who it's for, where you work and what it costs gives them something to cite.
- Rank in ordinary search. Assistants that search the web find pages through search, so pages that rank well are the ones they are most likely to read.
- Measure the actual answers. Instead of guessing which file or tactic works, ask the assistants your customers' questions and see whether you are named, and which pages they read. Here is how to do that by hand.
Should you add one?
If you already have one, it does no harm. Adding one takes minutes. But don't pay for it, and don't expect it to change whether AI assistants recommend you. On today's evidence, it won't.
For what it's worth, aipozi.com does publish an llms.txt, because it costs nothing, and our robots.txt welcomes the AI crawlers by name. We log which crawlers ask for the file and will publish what we find. Most of our effort goes on pages worth citing.
Frequently asked questions
Does ChatGPT read llms.txt?
We found no documentation from OpenAI saying its search crawler uses llms.txt to decide what to cite. In Ahrefs' server-log study, named AI bots made 19.5% of the requests for llms.txt files, and 97% of files got no requests at all in the month.
Is llms.txt an official standard?
No. It is a proposal published by Jeremy Howard in September 2024. None of the big AI companies has adopted it as a standard, so whether it does anything depends on whether their systems choose to read it.
Should I pay someone to write an llms.txt file?
No. Writing one takes minutes, and the evidence so far shows no effect on whether AI assistants cite or recommend a site.
Does aipozi.com publish an llms.txt?
Yes, because it costs nothing. We log which crawlers request it and will publish what we see, alongside our other measurements.