what is llms.txt, and do you actually need one.

elisabeth hitz · july 20, 2026 · 6 min read

every few months a new file appears that you're told you must add or fall behind. llms.txt is the current one. here's what it actually is, and the honest answer on whether you need it.

llms.txt is a plain-text markdown file at your site root that hands ai systems a clean overview of what your site is, who it's for, and links to your most important pages, so a model can understand you without parsing rendered html. it was proposed as a convention in 2024 (llmstxt.org). as of 2026 no major ai platform has confirmed it as a ranking signal, so treat it as cheap insurance and a clarity exercise, not a magic citation button.

that honest framing matters, because most of what you'll read about llms.txt is either "you must have this now" or "it's useless." both are wrong. it's a small, sensible file that helps in specific situations and does nothing in others. let's sort which is which.

what llms.txt actually is

a website is built for browsers and humans: navigation, scripts, styling, popups. an ai system trying to understand your site has to wade through all of that to find the few sentences that say what you do. llms.txt skips the wading. it's a curated markdown file, usually at yoursite.com/llms.txt, that reads like a clean table of contents for a language model:

# The Closer Method
> AI setup and AI visibility for small businesses and creators.

## Core pages
- [AI Visibility Scan](/seo/): check if AI recommends your business
- [The Toolkit](/skills): claude skills for operators
- [Field Notes](/blog/): plain-english guides on using AI well

that's the whole idea. no tracking, no styling, no javascript. just the plain statement of what the site is and where the important things live, in the format a model reads most easily.

llms.txt vs robots.txt vs pricing.md

these three files get lumped together and they do completely different jobs. the difference is the whole point:

robots.txt is a permission file: it can block or allow the ai crawlers (GPTBot, ClaudeBot, PerplexityBot, Google-Extended). if those are blocked, no context file saves you, because the platform can't read you at all. so robots.txt comes first: allow the bots, then give them context. the robots.txt teardown covers the permission side.

is it a ranking factor? no. is it worth doing? usually.

here's where honesty beats hype. no major ai platform has publicly confirmed that it reads llms.txt or uses it to decide citations. so if someone sells you llms.txt as the reason you'll show up in chatgpt, they're overselling. the confirmed levers, the ones in geo vs seo, are still the boring ones: extractable answers, cited statistics, and getting mentioned on third-party sites.

but the cost of adding one is near zero, and it has a real side benefit that has nothing to do with ai: writing an llms.txt forces you to state, in three sentences, what you do and which pages matter. most business sites can't do that cleanly. if the exercise is hard, that's not an llms.txt problem, that's a positioning problem you just found for free.

who actually benefits

add one if any of these is you:

• you sell to buyers who use ai to shortlist. b2b buyers increasingly ask an assistant "who are the options for X" before they ever visit a site. a clean context file makes you easy to summarize correctly.
• your site is heavy. lots of javascript, lots of navigation, thin on plain text. llms.txt gives models the clean version.
• you have a few pages that matter far more than the rest. the file lets you point at them instead of hoping a crawler guesses.

skip it, or deprioritize it, if your content is already clean, well-structured html with clear headings and you haven't fixed the bigger levers yet. a context file on top of uncitable content is a bow on an empty box. fix the box first. and the market is proving the point: in our field scan of 51 businesses, 43% had an llms.txt while only 12% had the answer schema that actually gets extracted.

the takeaway

llms.txt is a small, honest, low-cost file that tells ai systems what your site is and where the good stuff lives. it is not a ranking hack and anyone selling it as one is guessing. add it after you've allowed the ai crawlers and made your actual content extractable, treat it as clarity insurance, and move on to the levers that are proven to move citations. the file is easy. being worth citing is the work.

browse all 30 free tools →