llms.txt explained: the AI discovery file every business site should ship
What llms.txt is, in one sentence
llms.txt is a plain-text file at the root of a domain that gives AI tools a short, prioritized map of your most important pages. Think of it like robots.txt and sitemap.xml, but written for a person to read instead of a protocol parser.
Why it matters now
Tools like ChatGPT, Perplexity, Claude, and Google AI Overviews cite a handful of pages per answer. They lean on whatever they can summarize most confidently. A clean llms.txt, with a one-line summary next to each URL, gives them a near-perfect briefing on what your site is about and which pages matter. They no longer have to guess from a hundred low-signal blog posts.
A workable structure
A useful llms.txt has four parts: a title and one-line description, a short "about" paragraph with the organization, service area, and contact, grouped lists of canonical URLs with one-line summaries, and a footer section pointing at sitemap.xml and social profiles. Date-stamp the file so AI tools can see when it last changed.
Common mistakes
Two failure modes show up over and over. First, dumping every URL from the XML sitemap into llms.txt. AI tools do not want a list of 800 blog posts. They want the 30 pages a person would point at. Second, writing the summaries in marketing voice. Short, factual descriptions perform far better than "we are the leading provider of..." prose.
How to validate it
Two checks. First, fetch https://yourdomain.com/llms.txt with any HTTP client and confirm it renders as plain text with no auth wall. Second, run a prompt-based citation test before and after publishing it. Ask a handful of LLMs "What does [your business name] do?" The post-publish answers should land much closer to the description in your file.
Written by Mustafa Karim, founder and principal consultant at NavoTech Digital Solutions. Have a project or counter-example? Get in touch.