Glossary
What is llms.txt?
llms.txt is a plain Markdown file you place at yoursite.com/llms.txt that hands AI models a curated map of your most important pages. Jeremy Howard of Answer.AI proposed it in September 2024. It is an emerging convention, not a confirmed ranking factor, and no major AI provider has confirmed that their crawlers actually read it.
What is llms.txt, in one breath
llms.txt is a plain Markdown file that lives at yoursite.com/llms.txt. Think of it as a friendly little index card you hand to AI models: here are my most important pages, in plain language, organized the way I'd explain them to a smart intern. Jeremy Howard of Answer.AI proposed it in September 2024, and the idea spread fast.
Here's the honest part though. It's an emerging convention, not a rule. It is not a confirmed ranking factor, and as of now no major AI provider (OpenAI, Anthropic, Google, Meta) has confirmed their crawlers actually use it. So treat it as a low-cost bet, not a magic switch.
Why people made it
Regular web pages are a mess for machines. Nav bars, cookie banners, sidebars, ads, JavaScript that loads three seconds late. An AI model trying to understand your site has to wade through all that junk to find the actual content. The pitch behind llms.txt is simple: give models a clean, curated, human-and-machine-readable summary so they don't have to guess.
It's the spiritual cousin of robots.txt and XML sitemaps, but with a different job. Robots.txt says "you can or can't crawl this." A sitemap says "here's every URL I have." llms.txt says "here's what matters, and here's what each thing is about." It's curation, not a full dump. (Big difference, and the reason it can actually be useful.)
What it looks like
The format is dead simple. An H1 with your site or project name, a short blockquote summary, then sections of links with one-line descriptions. Something like:
# Outline Technologies
> Outline is an SEO, AEO, and GEO agency that gets brands cited by AI engines.
## Services
- [AEO & GEO](https://outline.ad/services/geo): get cited by ChatGPT and Perplexity
- [SEO](https://outline.ad/services/seo): rank on Google and in AI Overviews
## Optional
- [Blog](https://outline.ad/blog): posts on AI search
That's it. No schema wrestling, no JSON. If you can write a README, you can write one of these. Some teams also ship an llms-full.txt with the actual page content inlined, so a model gets the full text without a second fetch.
Does it actually work yet?
This is where you should keep your expectations on a leash. The adoption numbers are wild in both directions.
Per Ahrefs' May 2026 analysis of 137,000 domains, roughly 28% had published an llms.txt file. Sounds huge, right? But 97% of those files got zero requests that month. Zero. And Ahrefs notes that no major LLM provider has confirmed using the file as part of their crawler protocol.
| Metric (Ahrefs, May 2026) | Number |
|---|---|
| Domains studied | 137,000 |
| Had an llms.txt | ~28% |
| Of those, got zero requests | 97% |
| Major providers confirming use | 0 |
So a lot of people built the index card and almost nobody's reading it. That doesn't mean it's useless. It means it's early. Conventions like this can flip from "nobody cares" to "table stakes" quickly once a provider blesses them. But betting your whole strategy on it today would be silly.
Should you make one anyway?
Probably yes, if it's cheap for you. Here's my honest take. The downside is basically a few minutes of work and a file that might sit unread. The upside is a clean, maintained summary of your best content that AI tools can use if and when they decide to, plus it forces you to actually think about which of your pages matter. That clarity exercise alone is worth something.
A few ground rules so you don't shoot yourself in the foot:
- Keep it curated. Don't dump every URL. The whole point is signal over noise.
- Keep it current. A stale map is worse than no map.
- Don't expect rankings. It's not a ranking factor, full stop.
- Treat it as one tactic inside a broader generative engine optimization play, not the whole game.
If you want the deeper how-to, we walk through building one in our guide to llms.txt. And because this file shares DNA with crawl directives, it's worth reading how it sits next to robots.txt in our piece on AI crawlers and robots.txt. For the bigger picture of getting cited by AI engines, our AI SEO overview ties it together, and the LLM SEO glossary entry covers how models pick what to surface.
The bottom line
llms.txt is a smart, simple idea with real momentum and zero confirmation from the people who'd need to honor it. Build one because it's cheap, keep it tidy, and measure whether anything actually requests it. Just don't tell your boss it's a ranking factor. (It isn't, and the data says barely anyone's reading these yet.)
FAQ
Is llms.txt the same as robots.txt?
No. Robots.txt tells crawlers what they may or may not access. llms.txt does the opposite job: it offers a curated, plain-Markdown map of your most important pages so AI models can understand your site quickly. They complement each other rather than compete.
Does llms.txt improve my AI rankings or citations?
There's no confirmed link. llms.txt is an emerging convention, not a verified ranking factor, and no major AI provider has confirmed their crawlers use it. Per Ahrefs' May 2026 study, 97% of the files that existed got zero requests. Treat it as a low-cost experiment, not a guarantee.
Where do I put the llms.txt file?
At the root of your domain, so it resolves at yoursite.com/llms.txt. It's a plain Markdown file: an H1 with your site name, a short summary blockquote, then sections of links with one-line descriptions. Some sites also publish an llms-full.txt with the full page text inlined.
Who created llms.txt?
Jeremy Howard of Answer.AI proposed it in September 2024. The idea spread quickly across the AI SEO community, though adoption by the actual AI providers (OpenAI, Anthropic, Google, Meta) remains unconfirmed.
Related
See if AI engines cite your brand
Run a free check, or read the playbooks behind the term.