What Is llms.txt — and Why Your Business Website Needs One

The short answer: llms.txt is a plain-text file that lives at yourdomain.com/llms.txt and gives AI assistants a clean, curated map of your website — what your business is, and where the pages that matter are. As more customers ask an AI assistant for recommendations instead of searching Google, it's one of the cheapest steps toward being represented accurately. It's a proposed standard with mixed adoption, not a magic ranking switch — details below, honestly.

What is llms.txt, exactly?

It's a markdown-formatted text file, proposed by Jeremy Howard in 2024, with a simple recipe:

That's the whole thing. Where robots.txt tells crawlers what they may read, llms.txt tells AI systems what's worth reading and what it means. You can see ours live at getgoodtwin.com/llms.txt — this site practices what this post preaches.

Why this matters now

A growing share of "how do I…" and "who should I use for…" questions never touch a search results page — they're asked directly to ChatGPT, Claude, Perplexity, and the assistants built into phones and browsers. When those assistants answer, they often fetch and read websites on the spot, on the customer's behalf.

Here's the problem: many business websites are nearly unreadable to those assistants — text that only appears after JavaScript runs, key facts buried in images, answers hidden three clicks deep. The businesses that get described accurately (and recommended) are the ones whose sites are effortless for a machine to read. llms.txt is the front-door version of that: your business, summarized by you, in the format AI reads natively.

llms.txt vs. robots.txt vs. sitemap.xml

FileWritten forWhat it says
robots.txtSearch crawlers"Here's what you may and may not crawl."
sitemap.xmlSearch crawlers"Here's every URL that exists."
llms.txtAI assistants"Here's who we are and which pages matter, explained."

They're complements, not replacements. Keep all three.

What to put in yours

A small-business llms.txt can be a dozen lines:

Only list pages that are already public. It's a map, not a key — it grants no access and should contain no secrets.

The honest limits

Three things nobody selling "AI SEO" will tell you, and we will:

We recommend it anyway because the cost is minutes, the downside is zero, and it helps every assistant that does look — a bet with no losing side.

The rest of the AI-readable checklist

llms.txt is the doormat. The house is your content, and AI assistants reward the same things good writing always rewarded, plus a few technical ones:

  1. Text in plain HTML — if the words only exist after JavaScript runs, many AI readers see an empty page.
  2. Direct answers up front — state the answer in the first paragraph, then elaborate. Assistants quote pages that answer cleanly.
  3. Headings phrased as real questions — the ones customers actually ask.
  4. Schema.org markup — structured data (Article, FAQ, Organization) that machines parse without guessing.
  5. Honest claims — assistants cross-check across sources; overclaiming now gets caught by software, not just regulators.

Questions people ask

Is llms.txt an official standard?

No — a community proposal from 2024 with mixed, changing adoption. Cheap to add, honest expectations required.

Does Google use it?

No confirmation. Keep robots.txt and sitemap.xml for search; llms.txt is aimed at AI assistants. It won't hurt SEO.

Can it leak private info?

Only if you write private info into it. List public pages only — it's a map, not a key.

What matters more than llms.txt?

Readable, truthful pages: plain-HTML text, answers up front, question headings, schema markup. The doormat helps; the house decides.