What is llms.txt, and does it actually work?

What is llms.txt supposed to do?

The idea, proposed in 2024, is straightforward and reasonable: a language model arriving at a website has to wade through navigation, markup and boilerplate to find the substance. A curated plain-text file at /llms.txt would hand it a clean summary and a map instead.

A companion file, llms-full.txt, carries the extended version. Think of it as robots.txt's helpful cousin: not a set of rules, but an offer of context.

Does llms.txt improve AI citations?

The available evidence says no. An analysis of roughly 300,000 domains found no relationship between having an llms.txt file and how often a domain was cited in major AI answers. In that study's citation-prediction model, removing the llms.txt variable made the model more accurate — the file was contributing noise rather than signal.

Adoption data points the same way. Around one site in ten had the file, and adoption was essentially flat across traffic tiers, so the sites winning citations were not the sites publishing it.

Google's position is explicit: "You don't need to create new machine readable files, AI text files, markup, or Markdown to appear in Google Search (including its generative AI capabilities), as Google Search itself doesn't use them." As of early 2026 no major AI lab had publicly committed to consuming the file in production.

So why does BuildRocket publish one?

Three honest reasons, none of which is "it gets us cited".

It costs us nothing. Ours is generated from the same typed content files that render the pages, so it updates itself on every deploy and can never drift out of sync with the site. It adds no weight to any page a visitor loads.

It has a real use case that isn't citation: coding assistants and documentation tooling do consume these files, and it is a tidy way to state our services, price ranges and contact routes in a form a machine can lift without parsing prose.

And it is cheap insurance. If an engine does start honouring it, we are already there. That is a defensible reason to spend an hour on something — and a terrible reason to put it on an invoice.

Should I add llms.txt to my site?

If it takes you an hour and you enjoy tidy things, sure — the downside is zero. If an agency is charging you meaningfully for it as an AI visibility service, that is a reasonable moment to ask what evidence they have, because the published evidence points the other way.

The effort it displaces is what actually costs you. An hour on llms.txt is harmless; a retainer line item on llms.txt instead of writing genuinely quotable content is not.

Related questions