What Is llms.txt?
A short, plain-text file proposed as a way to tell AI assistants what your site actually is — before they have to guess. Here's what it does, what it doesn't, and what to put in one.
llms.txt is a proposed plain-text file, placed at the root of a
domain, that gives AI assistants and crawlers a clean, curated map of a site's most important
content — a short description of what the site is, its main sections, and where to find them.
It's similar in spirit to robots.txt or sitemap.xml, but aimed at
helping generative AI understand and describe a site well, rather than controlling crawl access.
It is not an official standard. No major AI company has confirmed adopting it,
and nobody can honestly promise it changes how a specific assistant behaves — it's a low-cost
signal, not a guarantee.
The problem it tries to solve
When a language model lands on a site, it doesn't see a business. It sees markup: navigation, templates, cookie banners, breadcrumbs, and somewhere inside all of that, the actual content. It has to infer what the site sells and what matters most, the same way a distracted visitor would — except a model can't click around to check.
llms.txt was proposed in 2024 as a direct answer to that: instead of leaving a model
to reconstruct the picture from templates and clutter, a site can state it plainly, once, in a
format built for a model to read rather than for a browser to render.
What a well-formed llms.txt file contains
A useful llms.txt stays short and reads like a briefing, not a data dump:
- A one- or two-line site description — what the site is and who it's for, without slogans.
- The main sections or categories, each with a short description and its link — the handful that define the business, not an exhaustive list.
- Key pages — an about or contact page, and anything a model would need to describe the site accurately.
- Verifiable, concrete information — the kind of detail that helps an assistant describe or recommend a site with confidence, rather than marketing language with nothing behind it.
What it deliberately leaves out matters just as much: the entire catalog or every URL on the
site belongs in sitemap.xml, not here. An llms.txt is a summary, not
an index.
llms.txt vs. robots.txt vs. sitemap.xml
The three files sit at the same address pattern — the domain root — and get confused constantly. They do three different jobs.
| robots.txt | sitemap.xml | llms.txt | |
|---|---|---|---|
| What it does | Grants or denies crawl access | Lists every URL to be crawled and indexed | Summarizes what the site is and what matters |
| Built for | Any crawler | Search engine indexers | Language models |
| Format | Directives (User-agent, Disallow) | XML, machine-generated | Natural-language Markdown |
| Scope | Access rules, site-wide | Exhaustive — every indexable URL | Curated — the handful that matter |
| If it's missing | Everyone can enter by default | Search engines rely on crawling and links instead | The model guesses on its own |
| Priority | Rules over the other two | Independent of the other two | Only matters if robots.txt lets the bots in |
That last row is the one that trips people up: llms.txt can be written perfectly and
still do nothing, if robots.txt blocks the AI bots from the site in the first place.
Access comes first, description second.
Why this is genuinely low-hanging fruit right now
Almost no ecommerce site has an llms.txt file yet. It's a new proposal, it isn't
required by anything, and most stores have never heard of it — which is exactly why it's worth
doing before it becomes standard practice. A file that takes twenty minutes to write correctly,
on a domain root that almost none of your competitors have touched, is about as cheap as
visibility work gets.
That said, cheap and easy isn't the same as proven. Read the honesty box below before treating this as a strategy rather than a small, sensible addition.
llms.txt is a proposed standard, not an adopted one. Nobody — not us, not anyone
selling it as a silver bullet — can promise a specific assistant reads it or changes its answers
because of it. What we can say: it's cheap to do, unlikely to hurt,
and too early to promise it helps. Do it because it costs almost nothing and
removes an excuse, not because it's a guaranteed win. It also won't fix content that isn't
citable to begin with, and it does nothing if your robots.txt is blocking AI bots.
Further reading
Frequently asked questions
Do I need llms.txt if I already have a sitemap.xml?
Yes, they do different jobs. sitemap.xml is an exhaustive list of every URL you want search engines to crawl and index. llms.txt is a short, curated summary written for language models: what you sell, your main categories, and what actually matters. One is a directory, the other is an executive summary. Having a sitemap doesn't make llms.txt redundant.
Will llms.txt affect my Google rankings?
There's no evidence it does, and it isn't designed to. llms.txt targets AI assistants forming an answer, not the classic search index. It isn't a ranking signal for Google, the same way robots.txt isn't a content-quality signal. Treat it as a separate lever, not an SEO shortcut.
Which AI companies actually read llms.txt?
None has officially confirmed it. llms.txt is a standard proposed in 2024, and as of today neither OpenAI, Anthropic nor Google has adopted it as part of how their assistants fetch or cite pages. It costs little to add and there's no evidence it hurts, but no one can honestly promise a specific assistant reads it yet.
Is llms.txt the same as robots.txt?
No. robots.txt is access control: it tells crawlers who can enter and which paths are off-limits. llms.txt is descriptive: it tells language models what your site is and what matters, without granting or denying anything. robots.txt still rules the relationship — if it blocks the AI bots, an llms.txt file changes nothing.
Can I add llms.txt to a Shopify store?
Not the way you would on regular hosting. Shopify gives no access to the domain root, so you can't upload a standalone file there. The workarounds are an app that serves it at the right path, a proxy or subdomain you control, or skipping it and putting the effort into robots.txt.liquid and structured data instead.
Is llms.txt worth doing if it's not an official standard yet?
For most stores, yes, as a low-cost addition — not as a strategy on its own. It takes little effort to write, it's unlikely to cause any harm, and almost no ecommerce site has one yet. What it won't do is fix content that isn't citable in the first place, or replace being reachable and indexed to begin with.
llms.txt is one piece. The diagnosis covers the rest
A free diagnosis of your blog shows what's actually stopping AI assistants and Google from citing you — beyond a single text file.
Get my diagnosisno card · the diagnosis is free · we email you once