How is llms.txt different from sitemap.xml?

sitemap.xml is built for classic indexing bots, while llms.txt targets conversational agents specifically. The first lists URLs to crawl; the second favors content that's readable, condensed and ranked, so a model can consume it directly instead of parsing tangled HTML.

Is llms.txt an official standard?

No, it's an emerging convention. No major AI provider has officially adopted it yet, though uptake is growing on the publisher side. It's a low-cost bet: the file is simple to produce and does no harm if nothing reads it.

What does an llms.txt file look like?

The format stays deliberately simple: a level-1 heading for the site name, a one-sentence blockquote summary, then sections of links to your reference pages, each with a short description. A documentation site would list its core guides, its API reference and its getting-started pages.

llms.txt: What It Is and Why It Matters for AI Search

llms.txt is a Markdown file at a site's root that points AI models to its most important, structured content — a sitemap built for LLMs, not classic crawlers.

May 30, 2026 · 2 min read

Example of an llms.txt file at the root of a website

GEO

llms.txt is a plain-text file written in Markdown, placed at the root of a website, that hands language models a ranked map of the site's most important and best-structured content.

Proposed in September 2024 by Jeremy Howard (co-founder of Answer.AI and fast.ai), the llms.txt format answers a real limit of AI engines: they work with a fixed context window and struggle to pull the essentials out of an HTML page buried under scripts, menus, and ads.

Why it matters for AI visibility

The llms.txt file works like a sitemap built for machines: it points models at your reference content, raising the odds they lean on your pages — and cite you — rather than a competitor's.

Unlike sitemap.xml, which is built for classic indexing bots, llms.txt targets conversational agents specifically and favors content that's readable, condensed, and ranked. It's an emerging convention: no major AI provider has officially adopted it yet, but uptake is growing on the publisher side.

A concrete example

A technical documentation site can publish an llms.txt file that lists its core guides, its API reference, and its getting-started pages — each in a lightweight Markdown version. An AI assistant asked about that product goes straight to those resources instead of guessing from the HTML of the interface.

The format stays simple: a level-1 heading for the site name, a blockquote summary, then sections of links. The model no longer has to parse tangled HTML — packed with scripts and secondary tags like ads, menus, and pop-ups — just to find what matters.

A minimal example, served at https://yoursite.com/llms.txt:

# Site title

> One-sentence description of the site.

## Documentation

- [Getting started](https://yoursite.com/start.md): up and running in 5 minutes
- [API reference](https://yoursite.com/api.md): every endpoint

## Optional

- [Changelog](https://yoursite.com/changelog.md): version history

For the full picture, see the complete GEO guide.

FAQ

What is the llms.txt file?: llms.txt is a plain-text file written in Markdown, placed at the root of a website, that hands language models a ranked map of the site's most important and best-structured content. It works like a sitemap built for machines, pointing models at your reference pages.
Who proposed the llms.txt format?: Jeremy Howard, co-founder of Answer.AI and fast.ai, proposed it in September 2024. The format answers a real limit of AI engines: they work with a fixed context window and struggle to pull the essentials out of an HTML page buried under scripts, menus and ads.
How is llms.txt different from sitemap.xml?: sitemap.xml is built for classic indexing bots, while llms.txt targets conversational agents specifically. The first lists URLs to crawl; the second favors content that's readable, condensed and ranked, so a model can consume it directly instead of parsing tangled HTML.
Is llms.txt an official standard?: No, it's an emerging convention. No major AI provider has officially adopted it yet, though uptake is growing on the publisher side. It's a low-cost bet: the file is simple to produce and does no harm if nothing reads it.
What does an llms.txt file look like?: The format stays deliberately simple: a level-1 heading for the site name, a one-sentence blockquote summary, then sections of links to your reference pages, each with a short description. A documentation site would list its core guides, its API reference and its getting-started pages.

Nicolas Meridjen

Fondateur de LightSpot.ai — outil d'audit de visibilité IA (46 critères SEO + GEO)

Je construis LightSpot.ai et j'analyse comment les moteurs de recherche IA (ChatGPT, Perplexity, Google AI Overviews) choisissent les sources qu'ils citent. J'écris sur le GEO et le SEO à partir de données d'audit réelles.

X LinkedIn Site

On the same topic

GEO3 min read

AI Citation: What It Is and How to Get Cited by AI Engines

An AI citation is when a generative engine — ChatGPT, Perplexity, AI Overviews — picks your content as a source in its answer and links back to your page.

May 30, 2026

GEO3 min read

GPTBot, ClaudeBot & PerplexityBot: AI Crawlers Explained

GPTBot, ClaudeBot, and PerplexityBot are the crawlers that let ChatGPT, Claude, and Perplexity cite your content. Block them and you get zero AI citations.

May 30, 2026

GEO2 min read

Google AI Overviews: What They Are and How to Get Cited

AI Overviews is Google's AI-generated answer at the top of search results. It cites two to four sources picked for how extractable and trustworthy they are.

May 30, 2026