What Is an llms.txt File, and Does Your Business Need One?
Short answer: almost certainly not. An llms.txt file is a plain-text briefing you publish for AI tools, and through 2025 it was widely recommended, including here. In mid-2026 Google confirmed its Search products ignore the file, and server logs across 137,000 domains showed AI crawlers almost never request it. Here is what it is, what the evidence now says, and the one file that genuinely does decide whether assistants can read your site.
If you've been hearing about llms.txt files and wondering whether your website needs one, you're not alone. It's new, it sounds technical, and most of what's been written about it reads like it was written for developers, not business owners.
This guide fixes that. Plain terms only. We'll tell you exactly what it is, what it doesn't do (just as important), and what you should actually do about it.
The short version: what is it, exactly?
An llms.txt file is a small, plain-text file you place on your website that gives AI tools a clean, organized summary of your business. Instead of forcing an AI to dig through your entire website, reading navigation menus, cookie banners, footer links, tracking scripts, and everything else a normal website contains, it hands the AI a curated briefing: here's who we are, here's what we do, here are the most important pages.
Think of it as a welcome packet for AI. The file sits at yourdomain.com/llms.txt, the same way your robots.txt file lives at yourdomain.com/robots.txt. It's a text file. It requires no software. It doesn't affect how your website looks or works for human visitors at all.
Why does this exist? What problem does it solve?
When an AI tool tries to understand your website, it has to read the actual HTML code, the raw source behind every page. That code is messy. A typical service page contains 2,000 words of useful content buried inside 15,000 or more lines of code: navigation, headers, footers, JavaScript, image tags, schema markup, tracking pixels, and everything else that makes a website work visually.
The AI has to guess what matters. Sometimes it guesses right. Sometimes it doesn't, and that's when you get inaccurate descriptions of your business, missing services, or a competitor getting cited instead of you.
Jeremy Howard, co-founder of Answer.AI and one of the most respected AI researchers working today, proposed llms.txt in September 2024 specifically to address this. The idea: give AI a clean, readable version of your site and it can represent your business accurately, quickly, and consistently, without having to reverse-engineer your website every time.
How it's different from robots.txt and your sitemap
These three files often get confused. They serve completely different purposes:
| File | Who it's for | What it does |
|---|---|---|
| robots.txt | All web crawlers | Tells bots which pages they're allowed to visit, a gatekeeping file |
| sitemap.xml | Google & Bing | Lists every URL on your site so search engines can find and index your content |
| llms.txt | AI language models | Gives a curated, readable summary of your business so AI can describe you accurately |
None of these replace the others. They work alongside each other. You still need your sitemap for Google. You still need your robots.txt. The llms.txt is a separate file for a separate audience, one that's growing fast.
Do the AI tools actually read it?
Almost never, and we now have numbers rather than opinions. Through 2025 this looked like sensible preparation. The evidence published in mid-2026 changed the answer.
From Google Search Central's AI optimization guide: "You don't need to create new machine readable files, AI text files, markup, or Markdown to appear in Google Search (including its generative AI capabilities), as Google Search itself doesn't use them."
The same page adds that creating one "will neither harm nor help your site's visibility or rankings in Google Search." That is the operator of the surface being optimised for, stating plainly that it ignores the file.
The server logs agree. Ahrefs examined 137,000 domains in June 2026. About 38,000 had a valid llms.txt. Ninety-seven percent of those files received no requests at all during May 2026. Roughly 1,100 domains saw any traffic to the file, and AI retrieval bots from ChatGPT and Perplexity made up about 1 percent of the requests that did arrive.
Two smaller studies point the same way. A vendor analysis in May 2026 counted 408 fetches of llms.txt across more than 500 million AI bot visits over 90 days. An independent log review across two sites recorded zero requests out of 68,759 AI bot visits in about a month. Small samples, same direction.
No AI search provider has publicly committed to reading llms.txt for search or citation. Not Google, not OpenAI, not Anthropic, not Perplexity.
Two things that sound like counter-evidence, and are not
Coding assistants and documentation tools do read it. That is real, and it is why large software companies publish one. It is a different job: helping a developer's tool understand an API. It has no bearing on a customer asking an assistant for a plumber in Kelowna.
Google's Chrome team added an experimental llms.txt audit to Lighthouse. That is for agentic browsing, not for Search, and Google Search's guidance above still stands. If a Lighthouse report tells you an llms.txt is missing, that is not a reason to add one.
So does your business need one?
No. If you do not have an llms.txt, you are not behind and you are not missing anything measurable. Spend the afternoon on your Google Business Profile instead.
If you already have one, leave it where it is. It costs nothing and does no harm. Just stop maintaining it, and stop paying anyone else to maintain it.
I used to recommend these. The reasoning was that the file was free, took under an hour, and might matter later. That was defensible in 2025. It stopped being defensible once the platform everyone is optimising for published the opposite in its own documentation and the server logs showed the file sitting untouched.
Updating a recommendation when the evidence moves is the job. Keeping one because it was already written into a proposal is how businesses end up paying for work that does nothing.
What to check instead, and this one genuinely matters
There is a file AI crawlers definitely read, and getting it wrong is expensive. It is robots.txt, and it has been on your website since the day it launched.
robots.txt tells crawlers what they are allowed to access. Plenty of sites permit Googlebot while quietly blocking every AI crawler, usually because a plugin or a previous developer added a blanket rule years ago. The site looks healthy in Google and is invisible to the assistants.
- GPTBot and OAI-SearchBot, OpenAI
- ClaudeBot, Anthropic
- PerplexityBot, Perplexity
- Google-Extended, Google's AI products
- Bingbot, which feeds Copilot
Open yoursite.com/robots.txt in a browser. If you see Disallow: / under any of those names, or a blanket rule that catches them, that is worth fixing today.
Beyond that one file, what gets a local business named by an AI assistant is unglamorous and familiar: a verified Google Business Profile with real photographs, the same business name, address and phone everywhere a machine can check, service pages written in plain language, and reviews that describe specific work. Assistants build an answer from sources that agree with each other, so consistency is the signal.
The AI search visibility page covers that work in the order it should happen.
llms.txt was a reasonable bet in 2025 and the evidence has gone against it. Google says Search does not use it, the crawlers are not fetching it, and no AI provider has committed to reading it. There is nothing here to catch up on.
Check that your robots.txt is not blocking the AI crawlers, get the Google profile right, and keep your details identical everywhere. That is what the assistants are actually reading.
