llms.txt is a proposed standard: a plain-text markdown file at the root of your site (yoursite.com/llms.txt) that gives large language models a curated map of your most important content. It was proposed by Jeremy Howard of Answer.AI in September 2024.

Do AI models actually use llms.txt?

As of mid-2026, no major AI provider has publicly confirmed that it uses llms.txt for training or for choosing citations, and independent observers have reported seeing little crawler traffic to these files. Google’s John Mueller has publicly compared the idea to the old keywords meta tag. Treat it as a low-cost, unproven signal, not a ranking lever.

Is llms.txt the same as robots.txt?

No. robots.txt tells crawlers what they may and may not access. llms.txt does the opposite: it is an opt-in map that highlights and curates your best content for language models. They solve different problems and you can use both.

Where does the llms.txt file go?

At the root of your domain, so it is reachable at yoursite.com/llms.txt. Some sites also publish an llms-full.txt with the full text of key pages inlined.

Should I bother adding an llms.txt file?

It is cheap and will not hurt, so adding one is reasonable once your higher-impact work is done. Just calibrate expectations: crawlability, answer-first content and off-site authority move AI visibility far more than a curated index file does today.

llms.txt: What It Is and Does It Work?

llms.txt is one of the most talked-about and most misunderstood ideas in AI search. Some people treat it as a magic switch that gets you cited by ChatGPT. Others dismiss it entirely. The truth sits in between, and it is worth getting right, because the answer changes whether you should spend any time on it. This guide covers what the file is, what it looks like, and an honest read on whether it works.

What is llms.txt?

llms.txt is a proposed standard for a plain-text, markdown file placed at the root of your website, reachable at yoursite.com/llms.txt. Its purpose is to give large language models a clean, curated map of your most important content, so a model that wants to understand your site is not left to parse cluttered HTML navigation and marketing pages. It was proposed by Jeremy Howard of Answer.AI in September 2024.

The motivation is real. Language models work within a limited context window, and a lot of a normal web page is boilerplate: menus, banners, scripts, cookie notices. The idea behind llms.txt is to hand the model a distilled index (and optionally, in a companion llms-full.txt, the full text of key pages) so it can find the signal without the noise.

What it looks like

The format is deliberately simple: a title, an optional summary in a blockquote, then markdown headings with lists of links and short notes. A minimal example:

# Cituna

> Cituna audits how AI engines cite your brand and generates the fixes.

## Docs
- [What is AEO](https://cituna.com/learn/answer-engine-optimization): the pillar guide
- [What is GEO](https://cituna.com/learn/generative-engine-optimization): generative engine optimization
- [Pricing](https://cituna.com/pricing): plans and features

## Optional
- [Glossary](https://cituna.com/learn/glossary): AEO, GEO and related terms

That is essentially it. A human can read it, a model can parse it, and it takes minutes to write for a small site. The simplicity is the point, and also, as we will see, part of why its impact is limited.

Does llms.txt actually work?

Here is where you should be skeptical of anyone selling certainty. The honest, evidence-based picture as of mid-2026:

No major provider has confirmed using it for citations. OpenAI, Anthropic, Google and Perplexity have not publicly stated that llms.txt influences whether or how they cite a site in answers.
Observed crawler traffic is low. People who have checked their server logs after adding the file have frequently reported few or no requests for it from the big AI crawlers.
Prominent skeptics exist. Google’s John Mueller has publicly compared the idea to the old keywords meta tag, a signal search engines ended up ignoring.
It has real traction in one place: docs and tools. Many documentation sites and developer platforms publish llms.txt, and it is genuinely useful when a developer or an agent pastes your docs into a model.

The balanced read: llms.txt is a sensible, low-cost idea whose payoff today is modest and unproven for citations. It may matter more as AI agents that read sites directly become common. Add it with clear eyes, not as a shortcut around the work that actually moves AI visibility.

If a tool or agency promises that an llms.txt file will get you cited by ChatGPT, treat that as a red flag. The levers that move citations are crawlability, answer-first content and off-site authority, covered in the AEO guide and GEO guide.

How to create one

If you want to add one, it is quick:

Pick your best pages. Your core docs, key guides, pricing, and the pages you most want a model to understand. Quality over completeness.
Write the file. Start with an H1 title and a one-line blockquote summary, then group links under markdown headings with a short note on each, following the example above.
Keep the important stuff first. Put the pages you care most about near the top; use an ## Optional section for the rest.
Publish it at the root. Deploy it to yoursite.com/llms.txt and keep it in sync when your key content changes.

Cituna generates an llms.txt for you as one of its fixes, alongside schema and FAQ markup, so you can ship it in a click and move on to the higher-impact work.

Should you add one?

Yes, with the right expectations. It costs a few minutes, it does not hurt, and it is a tidy signal that you maintain your site with machines in mind. Just do it after the work that actually earns citations, and do not let it become a substitute for that work. If you only have an hour this week, spend it making your content answer-first and reachable, not on a curated index file the engines may not be reading yet.

llms.txt: what it is and whether it works

What is llms.txt?

What it looks like

Does llms.txt actually work?

How to create one

Should you add one?

Frequently asked questions

See how AI engines see your brand