Where does llms.txt go?

At the root of your domain: https://yourdomain.com/llms.txt. It is plain Markdown: an H1 with the site name, a one-paragraph summary, then sections linking to your most important pages with one-line descriptions.

Does llms.txt replace robots.txt or sitemap.xml?

No. robots.txt controls what crawlers may fetch, sitemap.xml lists URLs for indexing, and llms.txt explains what your site is about in language models can use directly. They complement each other.

GEO fundamentals

llms.txt: what it is, whether it works, and how to write one

Q: Is llms.txt an official standard?

No. llms.txt is a community proposal (September 2024, Jeremy Howard), not a ratified web standard. No major AI provider has committed to fetching it on every request, but adoption costs minutes and several AI tools and crawlers do read it.

Updated August 2026

llms.txt is a plain-Markdown file at the root of your domain that tells AI systems, in their native format, what your site is and which pages matter. It takes ten minutes to write, costs nothing to maintain, and this page includes the exact file this site uses.

What llms.txt is

Proposed by Jeremy Howard in September 2024, llms.txt is a convention modeled on robots.txt: a predictable location (https://yourdomain.com/llms.txt) where a language model — or the agent browsing on its behalf — can read a concise, curated summary of your site instead of parsing navigation, cookie banners, and marketing pages. The format is deliberately simple Markdown: a title, a short summary, then sections of links with one-line descriptions.

Honest answer: does it work?

Partially, and cheaply — which is why it's worth doing. The truthful status in 2026:

It is not a ratified standard. No major AI provider guarantees fetching it for every answer.
Some systems do read it. AI agents that browse sites, developer-facing crawlers, and several AI search tools check the path — it's become a de-facto courtesy file, and adoption among documentation-heavy sites is high.
The downside is zero. It's one static file. If AI systems read it, they get your best self-description in the easiest possible format; if they don't, you've lost ten minutes.
It does not replace robots.txt (access control), sitemap.xml (URL discovery), or schema.org markup (structured facts). Ship all four.

How to write one

The structure that works:

H1: your site or company name.
Blockquote summary: one paragraph saying what you are, for whom, in plain language — this is the sentence you want AI to repeat about you.
Sections (## headings) linking your most important pages, each with a one-line description: product, pricing, docs or guides, contact.
Definitions of any category terms you want understood correctly.

Here is a shortened version of this site's real llms.txt:

# Mentioned

> Mentioned is an AI visibility monitoring platform for businesses.
> It helps teams understand whether AI assistants recommend their
> brand, which competitors appear in answers, and what to improve.

## Key pages
- Home: https://mentionedby.world/
- Guides: https://mentionedby.world/guides/

## Definitions
- AI visibility: how often and how prominently a brand is mentioned
  or recommended in relevant AI responses.

## Contact
- Email: info@mentionedby.world

Mistakes to avoid

Dumping your whole sitemap into it. Curation is the point — 10 links beat 500.
Marketing language. “The world's leading revolutionary platform” gives a model nothing to work with. State the category, the audience, and the differentiator.
Letting it go stale. A wrong price or dead link in llms.txt is worse than no file — you're feeding errors directly to systems that repeat them.

Frequently asked questions

Is llms.txt an official standard?

No — it's a community proposal. But several AI tools and crawlers read it, and the cost of adding one is near zero.

Where does the file go?

At the domain root: https://yourdomain.com/llms.txt, served as plain text.

Does it replace robots.txt or sitemap.xml?

No. robots.txt controls access, sitemap.xml lists URLs, llms.txt explains meaning. They stack.

Want to know if AI systems actually understand your brand? A free Mentioned scan asks real buyer questions across ChatGPT, Gemini, and more — and shows what gets said, and who gets recommended instead.