AI systems have to figure out in seconds what a website is about. Classic HTML with navigation, banners and marketing text makes that hard for them. The llms.txt solves this: a plain text file in the root directory (for example surfaced.at/llms.txt) that explains to an AI model, in clear language, what the company does, for whom, and which pages are most important.
What's in an llms.txt
- A short description of the company in one or two sentences — the "what and for whom"
- The most important pages as commented links: services, pricing, FAQ, contact
- Audiences, specialization and unique selling point in clear language
- Factual key data: location, owner, contact — consistent with the rest of the online presence
Why AI reads it
The llms.txt is a deliberately curated entry point for AI — comparable to what robots.txt is for search engine crawlers. Instead of guessing its way through unstructured HTML, the model gets a clean, contradiction-free summary. And freedom from contradiction is decisive, because otherwise AI fills gaps with guesses.
Important: the llms.txt doesn't replace schema markup or good content, it complements them. It works strongest when its statements match website, schema and external sources like LinkedIn. Surfaced creates and maintains the llms.txt as part of the AEO packages — derived from a single data source, so all information stays consistent.
Surfaced uses llms.txt itself too: surfaced.at/llms.txt is live. In the free mini-check we check, among other things, whether your page has an llms.txt and whether AI crawlers have access at all.
Test the dashboard free — from €0