Awwwards
404media is now part of Awwwards. Find our studio profile and work there.

llms.txt is a Markdown file that gives AI systems a map of your website. Google ignores it, and so far there is no proof it gets you cited. Here is when it still makes sense and how to write one.
Christoph SpeiserProject lead & salesIn short: llms.txt is a plain Markdown file at yoursite.com/llms.txt. It tells language models in a few lines who you are and which pages matter most, with a link and a short note for each. It does not influence Google: Google states that Search ignores the file and that it has no effect on rankings or AI Overviews. Whether ChatGPT or Perplexity use it for their answers has not been confirmed. We create it on every website because it costs almost nothing, but we don't sell it as a way into AI answers.
Hardly any GEO topic has as many myths as this one. Some agencies offer llms.txt as a separate package. This article shows what the file is meant to do, what the evidence says and how to write a good one in a few minutes.
Jeremy Howard of Answer.AI proposed the file on 3 September 2024. The spec lives at llmstxt.org and is now titled “The /llms.txt file, v2”. The idea: a language model that reads your site while answering a question has little time and limited context. Instead of working through menus, cookie banners and scripts, it reads a short summary with links to the important content.
The file sits in the root, like robots.txt, or in a subfolder such as /docs/llms.txt for that part of the site. Unlike robots.txt it doesn't allow or block anything. Unlike sitemap.xml it doesn't list every URL, only a curated selection.
The spec defines a fixed order. Only the first part is required:
| Part | Markdown | Required |
|---|---|---|
| Name of the site or company | H1 heading: # Name | Yes |
| Short summary | Quote: > one or two sentences | No |
| More details | Paragraphs or lists, no headings | No |
| Link lists by topic | H2 heading, then one line per link with title, URL and note | No |
| Less important links | H2 heading named ## Optional | No |
Version 2 adds a convention: pages that agents need should also exist as clean Markdown at the same address with .md appended. A page can point to these versions with a link tag of type text/markdown.
Here is a shortened example for a fictional joinery workshop in Graz:
# Tischlerei Muster
> Joinery in Graz since [year]. Kitchens, wardrobes and stairs made of local wood for private and business clients in Styria.
## Services
- [Custom kitchens](https://www.example.at/kuechen): planning, production and fitting, delivery time 8–12 weeks
- [Stairs](https://www.example.at/treppen): oak and larch, measured on site
## Company
- [About us](https://www.example.at/ueber-uns): team, workshop, training
- [Contact](https://www.example.at/kontakt): address, opening hours, phone
## Optional
- [Blog](https://www.example.at/blog): care tips and projectsUse the same facts as on your website and Google Business Profile: same name, same address, same services. A file that contradicts the website does more harm than good. Our own file is live at 404media.at/llms.txt and is generated from the same data as the website, so it can never fall out of date.
llms-full.txt is not part of the official spec, but documentation platforms such as Mintlify generate it automatically. It contains the full text of all pages in one file, so an AI tool can load everything in one go. That makes sense for technical documentation. Anthropic's llms-full.txt for its developer docs is around 45 MB. For a company website with twenty pages it adds nothing.
No. In its guide to generative AI features, Google writes that you don't need machine-readable files, AI text files, markup or Markdown to appear in Google Search. On 15 June 2026 Google added a clarification: these files aren't needed for Google Search and won't negatively or positively impact your visibility or rankings. Keeping one for other services is fine.
John Mueller of Google called llms.txt “purely speculative for now” in June 2026. In 2025 he had already compared it to the keywords meta tag, which Google has ignored for years.
No provider has confirmed that its search crawler uses llms.txt to choose sources. OpenAI documents OAI-SearchBot as the crawler for ChatGPT search, and it reads normal web pages. The available data points the same way:
What does get you into AI answers is described in our guide Get cited by ChatGPT: open doors for the crawlers, answer-first content, consistent facts and mentions on other websites.
| Situation | Benefit |
|---|---|
| Software documentation, APIs | High: coding tools such as Cursor or Claude Code load the file directly when developers point them to it |
| Online shop with many products | Low to medium: an overview of categories, shipping and returns helps agents that read the site |
| Company website, 10–50 pages | Low: five minutes of work, no measurable effect on AI answers |
| Website with content only in JavaScript | None: fix the website first, crawlers don't see the content anyway |
Our advice: create it, keep it accurate, and spend your time on the things that work. Which ones those are for your business is shown by our free AI visibility check.
WordPress plugins such as Yoast SEO can generate the file. On sites we build, it comes from the same content as the pages and updates itself.
We check crawler access, content and company facts and show you which steps actually get you into AI answers. Free strategy call, fixed quote.
A Markdown file in the root of a website that gives language models a short summary and a curated list of the most important pages. The format was proposed by Jeremy Howard in September 2024 and is described at llmstxt.org.
No. Google says Search does not use the file and that it has no positive or negative effect on rankings or AI Overviews.
There is no evidence for that. No AI provider has confirmed using the file for answers, and a study of around 300,000 domains found no link to AI citations.
No. robots.txt tells crawlers what they may access. llms.txt only describes your content. To let AI crawlers in, you need the right rules in robots.txt.
Yes, if it takes little effort. It does no harm and helps AI agents and developer tools that load it on purpose. Just don't expect more visibility from it.
For a company website about 15 minutes. Keeping it in sync with the website matters more than the first version.

More lessons from our projects in Vienna.

SEO & GEO • • 8 min read
Christoph Speiser
SEO & GEO • • 9 min read
Christoph Speiser
SEO & GEO • • 10 min read
Christoph Speiser
Tell us about your project. Christoph gets back to you personally.
Christoph Speiser, Project lead & sales