llms.txt is a small plain-text file that sits at the root of your website (yourdomain.co.uk/llms.txt) and gives AI systems a short, curated summary of who you are, what you do, and which pages matter most. It's written in Markdown, takes about an hour to produce, and exists because websites are built for human eyes while AI tools work best with clean, simple text.

It works like a reading guide you leave at the front door for machines. It's a proposed standard rather than an official one, and we'll be honest with you up front: adoption is still early, and no major AI company has formally committed to using it. We add it to every site we build anyway, and by the end of this post you'll see why. If you're new to the wider topic, start with our plain-English guide to GEO. This post covers one small, practical piece of that puzzle.

What is llms.txt, exactly?

The idea was proposed in September 2024 by Jeremy Howard, an AI researcher and co-founder of Answer.AI. The problem it tackles is simple. Web pages are cluttered with navigation menus, cookie banners, scripts and footers, and AI systems have limited room for what they can read at once. A typical homepage might be hundreds of kilobytes of code wrapped around three sentences that actually matter.

llms.txt strips all that away. It's a single Markdown file with a deliberately simple structure:

• A heading with your business name. The only required element.
• A short summary in a blockquote. One or two sentences saying what the site is and who it's for.
• Sections of links to your key pages, each with a one-line note on what the page covers.

That's the whole format. It is, genuinely, just a text file, though the web has standardised on stranger things.

What goes in one? A worked example

Here's what an llms.txt file might look like for a small trades business. "Hartley & Son Plumbing" is fictional, so treat it as a template to adapt to your own business, not something to copy word for word.

# Hartley & Son Plumbing

> Family-run plumbing and heating company covering Taunton and
> surrounding villages. Gas Safe registered. Emergency call-outs,
> boiler servicing and bathroom installation. Trading since 1998.

## Services

- [Emergency plumbing](https://www.example.co.uk/emergency/): 24/7 call-outs, typical response under two hours
- [Boiler servicing and repair](https://www.example.co.uk/boilers/): Gas Safe engineers, all major brands
- [Bathroom installation](https://www.example.co.uk/bathrooms/): design and full installation, typical job 5-10 days

## About

- [About us](https://www.example.co.uk/about/): who we are, qualifications, areas covered
- [Reviews](https://www.example.co.uk/reviews/): verified customer reviews
- [Contact](https://www.example.co.uk/contact/): phone, email and call-out request form

## Pricing

- [Price guide](https://www.example.co.uk/prices/): call-out fees and typical job costs

Notice what makes it work. Plain language. Verifiable facts: registrations, response times, trading history. And links chosen around the questions real customers ask an AI: who is this, what do they do, where do they cover, what does it cost, can I trust them. Keep it short. llms.txt is an index, not a brochure.

How does llms.txt fit alongside robots.txt and schema?

These three get muddled together, but they do different jobs and they complement each other:

• robots.txt is about permission. It tells crawlers (search engines and AI alike) which parts of your site they may access. It's where you decide whether AI crawlers are welcome at all.
• Schema markup is about facts. Machine-readable labels inside your pages that spell out your business type, address, opening hours, services and FAQs, unambiguously.
• llms.txt is about guidance. A curated summary and reading order, so an AI system can understand your site without wading through the clutter.

It isn't either/or. If you have to prioritise, do it in that order: get robots.txt right first, add schema second (search engines confirmed long ago that they use it), and treat llms.txt as the inexpensive extra on top.

The honest caveats

We'd rather you heard this from us than discovered it later:

• It's a proposal, not a standard. At the time of writing, no major AI company has publicly confirmed that its assistants use llms.txt when answering questions. Google has said plainly that it doesn't use the file.
• The evidence is patchy. Some AI crawlers have been observed requesting the file, and developer documentation platforms adopted it quickly. But "an AI recommended this plumber because of llms.txt" has not been demonstrated, and anyone claiming otherwise is selling something.
• It's not a magic lever. An llms.txt file won't rescue a thin website or contradictory listings. The fundamentals come first: checking what AI currently says about you and fixing the sources it reads.

So why bother? Cold cost-benefit. It takes an hour to write, minutes a year to maintain, and carries no downside. At worst it's ignored. If AI assistants do standardise on it, the sites that already have one are ahead. That's a cheap position to hold.

Why we add it to every site we build

We build our clients' websites on our own in-house CMS, which means when a sensible new convention appears, we can roll it out across every site we look after rather than waiting on a plugin. llms.txt earned its place on that list: we write the file as part of every build, and for clients on our care plans we keep it current when services, prices or coverage areas change. A summary file that's out of date is worse than none at all.

It's one small piece of a bigger habit: building sites that are legible to humans first and machines a close second. The typography is for your customers; the structure underneath is for everything else that reads the web on their behalf.

Common questions

Will llms.txt improve my Google rankings?

No. Google has been clear that llms.txt is not a ranking factor, and it has no effect on traditional search results. It is aimed at AI systems rather than search engines. Treat it as a low-cost bet on how AI reads websites, not as an SEO tactic.

Is llms.txt the same as robots.txt?

No. robots.txt tells crawlers what they may and may not access, so it is about permission. llms.txt tells AI systems what your site is about and which pages matter most, so it is about guidance. They sit side by side at the root of your site and do different jobs.

What is llms-full.txt?

A companion file from the same proposal. Where llms.txt is a short index with links, llms-full.txt contains the full text of your key content in a single file, so an AI system can read everything without following links. It is most useful for documentation sites; most small business websites only need llms.txt.

Do I need a developer to add llms.txt?

It helps, but it is not essential. The file is plain text written in Markdown, so the writing is the easy part. You, or whoever runs your website, then upload it to the root of your site so it appears at yourdomain.co.uk/llms.txt. On sites built on our CMS, we handle this as standard.

Which AI tools actually read llms.txt?

Honestly: adoption is still early. Some AI crawlers have been observed fetching the file, and developer documentation tools have adopted it fastest, but none of the major AI companies has publicly committed to using it in their assistants. That may change, and the file costs almost nothing to maintain, which is why we add it anyway.

Want your site to make sense to machines?

We're Bare Creative: Jack and Carly, a web design and digital studio in Somerset. Every site we build ships with llms.txt, structured data and a sensible robots.txt as standard, because being understood by AI is now part of being found at all. If you'd like to know whether your current site is machine-legible, or you'd like one that is, get in touch. We'll give you a straight answer either way.