llms.txt and AI search: what the file is, and what Google actually asks for

Where llms.txt comes from

The llms.txt proposal was authored by Jeremy Howard and first published on 3 September 2024; the site llmstxt.org shows a second version of the proposal, updated in August 2026 “based on what I learned from two years of adoption”. Its aim is to standardise a /llms.txt file that helps agents use a website.

The reasoning is practical: pages built for people carry navigation, scripts and layout that an AI model has to wade through. A short Markdown file can give the essentials in one place.

What goes in the file

According to the proposal, the file sits at the site root (or at a path within it, covering the pages under that path) and follows a simple Markdown structure, in this order:

1. An H1 with the name of the project or site. This is the only required section.
2. A blockquote with a short summary containing the key information.
3. Optional detail in paragraphs or lists.
4. Zero or more H2 sections containing file lists: each item is a Markdown link name, optionally followed by a colon and notes.

What to put in it, and what to keep out

For a small business, a useful llms.txt reads like a fact sheet a journalist could rely on. Take an imaginary bakery: an H1 with its name, a one-line summary saying what it sells and where, then a short H2 list linking to the pages that hold the facts, each with a few words of explanation.

• Put in: what you do, where you operate, who you serve, how to contact you, and links to the pages that prove it (services, prices page if you publish one, opening hours, terms, about).
• Keep every statement true and checkable on the linked page. A fact sheet that contradicts the website is worse than none.
• Keep out: slogans, claims you cannot back up, private information and anything you would not want quoted out of context.
• Update it when your offer, address or opening hours change, at the same time as the pages themselves.

The companion idea: Markdown versions of pages

The proposal also suggests offering a clean Markdown version of useful pages at the same URL with .md added (page.html.md) or with the extension replaced by .md. This is optional and only worth it if you can keep those versions in sync with the real pages.

What Google says about AI Overviews and AI Mode

Google Search Central has a page for site owners on AI features. Its key points, in Google's words:

• There are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimisations necessary.
• You don't need to create new machine-readable files, AI text files or markup to appear in these features, and there is no special schema.org structured data you need to add.
• To be eligible to appear, a page must be indexed and eligible to be shown in Google Search with a snippet, meeting the Search technical requirements.
• To limit what appears, use the existing controls: nosnippet, data-nosnippet, max-snippet or noindex.
• Traffic from AI features is included in the overall search traffic in Search Console's Performance report, under the Web search type.

So should you add an llms.txt?

Our view: treat llms.txt as a cheap, accurate fact sheet for any agent that chooses to read it, and spend most of the effort where Google says it matters, on indexable pages that answer the question clearly and are worth citing.

What we do on our own sites at Takat

On all our sites, since 4 October 2026, we have: an llms.txt fact sheet, AI crawlers explicitly allowed in robots.txt, an opening sentence that answers the page's question, JSON-LD structured data matching the visible content, and IndexNow notifications. None of this is presented as a guarantee of being cited by an AI. It is the readiness work we also offer in our AI search service.