llms.txt

llms.txt is a plain-text, Markdown file you publish at your site's root (/llms.txt) to hand AI models a curated map of the pages you most want quoted — docs, product overviews, key guides — with a short note on each. Instead of leaving models to crawl your whole site and guess what matters, you point them straight at the good stuff. It is a proposed standard, not yet honored by every model, but it is cheap to ship and sits right next to the rest of your AI-search setup.

How it works

The file lives at https://yoursite.com/llms.txt. It is written in Markdown, so it reads fine to both a person and a machine. The convention is simple: an H1 with your site name, an optional short summary in a blockquote, then H2 sections holding lists of links. Each link is a Markdown link with a short note on what the page covers.

That structure matters. A normal HTML page buries your real content under navigation, scripts, and styling. An AI crawler has to strip all that away. With llms.txt you give it the clean version up front — the URLs that matter, in plain words, with context attached. Some teams also publish an llms-full.txt that inlines the actual page text, so a model can read the content without a second fetch.

Why it matters for B2B sites

B2B buyers increasingly ask ChatGPT and Perplexity before they ever visit your site. If the model can read a tidy map of your strongest pages, it is more likely to surface and cite the right one. That is the core goal of generative engine optimization — being the answer, not just a result. The same file also feeds the systems behind AI Overviews in Google.

We treat llms.txt as one supporting layer, never the whole strategy. It does not replace clear, quotable content or proper schema. To see where it fits, read our guide on how to rank in ChatGPT and Perplexity. The honest take: ship it, keep it current, but expect it to help at the margins while your content does the heavy lifting.

FAQ

Is llms.txt an official standard?

No. llms.txt is a proposed convention, not a ratified standard, and no model is obligated to read it. Adoption is uneven and still changing. Treat it as a low-cost addition to your answer engine optimization setup rather than something that is guaranteed to be consumed.

Do ChatGPT and Perplexity actually read llms.txt?

There is no guarantee any given model fetches or uses the file today, and support can change over time. The safe approach is to ship it because it is cheap and easy to keep current, while assuming your clear, quotable on-page content and schema are doing the heavy lifting.

What is the difference between llms.txt and llms-full.txt?

llms.txt is the curated index: links to your key pages, each with a short note. llms-full.txt goes further and inlines the actual page text, so a model can read the content without a second fetch. Many teams publish the index first and add the full version only if it earns its keep.