What Is llms.txt, and Should Your Business Have One?

Every few months a new file lands in the search world and gets sold as the thing that will finally make AI notice you. llms.txt is the current one. You may have seen it described as "robots.txt for AI" or as a switch that makes ChatGPT prefer your site. Neither is quite right, and the gap between what it actually is and how it gets pitched is worth closing before you spend a single afternoon on it. So let's be plain about three things: what llms.txt is, how it differs from the two files people keep confusing it with, and whether it is worth your time in 2026.

What llms.txt actually is

llms.txt is a single markdown file that you place at the root of your website, at the address /llms.txt, so it sits alongside your robots.txt and sitemap.xml. Its job is to give a large language model a clean, curated summary of what your site is about and a short set of links to the pages that matter most. The proposal came from Jeremy Howard of Answer.AI in 2024, and it is community-driven rather than an official standard handed down by Google, OpenAI or anyone else.

The reasoning behind it is sensible enough. A modern web page is a mess from a machine's point of view: navigation menus, cookie banners, related-article widgets, adverts and scripts, all wrapped around the actual content. A model trying to understand your business has to wade through that clutter and guess at what is important. llms.txt sidesteps the guessing. It is a plain-markdown document, written by a human, that says here is who we are, here are the pages you should read, and here is a line about each one. Think of it as a tidy front desk for machines rather than a rebuilt website.

What actually goes in it

The format is deliberately simple, which is part of the appeal. An llms.txt file starts with a single H1 heading that is the name of your site or business. Under that sits a short blockquote, a one-line summary of what you do. Then come optional paragraphs of context, followed by sections, each introduced by an H2 heading, that hold markdown lists of links to your key pages. Each link can carry a short note explaining what it covers, so the model knows why the page is worth reading before it fetches it.

Here is a small, complete example for a fictional plumbing firm, so you can see the shape of it:

# Riverside Plumbing

> Emergency and scheduled plumbing and heating services for homes and small businesses across Bristol, available seven days a week.

Riverside Plumbing is a family-run, Gas Safe registered plumbing and heating company covering Bristol and the surrounding area. This file lists the pages most useful for understanding what we do, where we work and how to reach us.

## Core services

- [Emergency plumbing](https://example.com/services/emergency): Same-day call-outs for leaks, burst pipes and blockages, 24 hours a day.
- [Boiler installation and repair](https://example.com/services/boilers): Gas Safe boiler fitting, servicing and breakdown cover.
- [Bathroom fitting](https://example.com/services/bathrooms): Full bathroom design and installation.

## About and contact

- [About us](https://example.com/about): Who we are, our accreditations and our service area.
- [Areas we cover](https://example.com/areas): The Bristol postcodes and nearby towns we serve.
- [Contact and booking](https://example.com/contact): Phone, email and online booking.

## Useful reading

- [Pricing and call-out fees](https://example.com/pricing): How our charges work, with no hidden fees.
- [Frequently asked questions](https://example.com/faq): Common questions on timings, guarantees and payment.

That is genuinely all there is to it. No code, no configuration, no special syntax beyond ordinary markdown. The whole point is that it stays short and readable, so keep it to your best pages rather than dumping every URL you have. The convention also allows an optional expanded file at /llms-full.txt that inlines the full text of those pages for tools that want everything in one place, but the core /llms.txt is the summary-and-links file above.

How it differs from robots.txt and sitemap.xml

This is where most of the confusion lives, and it clears up fast once you see that the three files do three separate jobs. They are not competing versions of the same idea, and having all three makes perfect sense.

robots.txt is about permission. It is a long-standing file that tells crawlers which parts of your site they are allowed to access and which they should leave alone. It is a gatekeeper: yes to this path, no to that one. It says nothing about what your content means or which pages matter, only who may go where. If you want to control whether AI crawlers can reach you at all, that is a robots.txt conversation, and we cover it in our guide on whether your website is blocking AI crawlers.

sitemap.xml is a complete list of URLs. It is a machine-readable inventory, in XML, of every page you want search engines to know about, often with extra data like when each page last changed. Its job is coverage: making sure crawlers can find all your pages. It is exhaustive and built for software, not for reading. It does not summarise, prioritise or explain anything.

llms.txt is a curated summary in plain language. It is neither a permission file nor a full inventory. It is a short, human-written document that points a model at your most important content and tells it, in a line or two, what each page is about. Where sitemap.xml says here is everything, llms.txt says here is what matters and why. That difference, exhaustive-and-for-machines versus curated-and-readable, is the whole distinction, and it is why the file is written by a person rather than generated automatically.

The honest bit: does anyone actually use it?

Here is where a lot of the coverage goes quiet, and where we will not. llms.txt is an emerging, community-driven proposal, and as of 2026 the major AI crawlers do not all officially honour it. It is an idea with momentum, not a settled standard that the big platforms have committed to reading.

Google has been explicit on this point. Its representatives have said publicly that Google Search does not use llms.txt files, and John Mueller has compared the situation to the old keywords meta tag, something sites can fill in that the search engine simply does not act on (Search Engine Land, 2025). That is a direct statement from the largest player, and it matters. On the other side, some developer-facing tools and documentation platforms do read llms.txt today, and a number of technical companies publish one, so the file is not ignored everywhere. But the picture is patchy, adoption is uneven across the crawlers that shape AI answers, and nobody should tell you it is a confirmed route into ChatGPT or AI Overviews.

So the honest framing is this. llms.txt is low-cost and low-risk. It is a small markdown file that cannot hurt your site, and it may help the tools that do read it understand you more cleanly. That makes it reasonable hygiene and sensible future-proofing, in case adoption grows. What it is not is a magic bullet or a guaranteed lever for rankings or citations. If a supplier is selling you an llms.txt as the thing that will get you named by AI, they are overselling a proposal that the biggest crawlers have not signed up to. Add it because it is tidy and cheap, not because you have been promised an outcome.

How to create one, step by step

If you have weighed that up and want one, it takes about half an hour. Here is the process from start to finish.

1. List your most important pages. Not every page, just the handful that best explain your business: your core services, your about page, your contact and location details, your pricing, and a few genuinely useful resources. Aim for a curated shortlist, not a full sitemap.

2. Write the header. Open a plain text file. Put a single H1 line at the top with your business name, prefixed with a hash and a space. Under it, add a blockquote, a line beginning with a greater-than sign and a space, that summarises what you do in one sentence.

3. Add a short paragraph of context. A sentence or two under the summary, in plain markdown, giving any background a model would find useful: where you operate, what makes you distinct, any accreditations that matter.

4. Group your links into sections. Add H2 headings for logical groups, such as "Core services", "About and contact" and "Useful reading". Under each, write a markdown bullet list where every item is a link in the form [Page name](full URL), followed by a colon and a short note on what the page covers. Use full, absolute URLs, not relative paths.

5. Keep it lean and save it as llms.txt. Resist the urge to add everything. Save the file as plain text named exactly llms.txt.

6. Publish it at your root. Upload it so it is reachable at yourdomain.co.uk/llms.txt, in the same place your robots.txt lives. On most platforms that means the root of your public files; on hosted site builders you may need a redirect or a plugin to serve a file at that path. Once it loads in a browser, you are done.

For the record, we practise what we write: Aeora publishes its own at https://aeora.uk/llms.txt, and you are welcome to read it as a working example.

Where it fits in the bigger picture

llms.txt is worth keeping in proportion. It is one small, tidy file among the things that actually shape whether AI names your business, and on the current evidence it is nowhere near the top of that list. The heavier levers are the ones we would point your budget at first: clear, answer-shaped content in visible HTML that models can read, consistent and accurate facts about your business, a presence in the third-party sources these tools trust, and clean crawler access so the tools can reach you at all. Structured data belongs in that group too, and we set out what it does and does not do in our guide to technical foundations and schema.

Add llms.txt as part of good housekeeping once those bigger pieces are in hand. It costs almost nothing, it cannot do harm, and if adoption grows you will already have one in place. Just hold it in its right size: useful hygiene and quiet future-proofing, not the switch that changes everything.

Questions people ask

What is an llms.txt file?

llms.txt is a proposed convention: a single markdown file you place at the root of your site, at /llms.txt, that gives a large language model a clean, curated summary of what your site is and links to your most important pages. The idea is to hand a model a tidy map of your best content in plain markdown, rather than making it wade through cluttered HTML and navigation to work out what matters. It was proposed by Jeremy Howard of Answer.AI in 2024 and is community-driven, not an official standard from any search or AI company.

Is llms.txt the same as robots.txt or sitemap.xml?

No, they do three different jobs. robots.txt sets permissions and tells crawlers which paths they may or may not access. sitemap.xml is a machine-readable list of every URL on your site, meant to help search crawlers with coverage. llms.txt is neither a permission file nor a full URL list: it is a short, human-written summary in markdown that points a model at your most important content and explains it. You can sensibly have all three, because they do not overlap.

Do AI crawlers actually use llms.txt in 2026?

Support is limited and evolving, so be honest with yourself about it. llms.txt is an emerging community proposal, and as of 2026 the major AI crawlers do not all officially honour it. Google has said publicly that its Search systems do not use llms.txt files. Some developer-facing tools and documentation platforms read them, but you should treat the file as good hygiene and future-proofing rather than a guaranteed ranking or citation lever. It is low-cost and low-risk to add, and that is the honest case for it, not a promise of visibility.

Sources

• llmstxt.org, The /llms.txt file proposal (Jeremy Howard, Answer.AI)
• Search Engine Land, Google: llms.txt is not used by Search (2025)
• Search Engine Journal, What Is llms.txt, And Why Does It Matter? (2025)