llms.txt for Small Business Websites

Man working at a desk with several computer screens, reviewing a business website

Photo on Pexels

Sooner or later someone will tell you that your website needs an llms.txt file. It might be your web agency, an SEO newsletter, or a vendor offering to “make you visible to ChatGPT” for a monthly fee. The pitch sounds plausible: a small text file that tells AI engines what your business is and which pages matter.

Here is what llms.txt actually is, what it isn't, what the big engines have said about it, and how to decide whether it's worth an hour of your web person's time. Short version: it's cheap to add, it's not a ranking lever, and it does not replace the work that gets a business named in an AI answer.

What llms.txt is

llms.txt is a proposal published in September 2024 by Jeremy Howard of Answer.AI. The idea: put a plain Markdown file at /llms.txt that gives a language model a short, clean summary of your site and links to the pages worth reading, so it doesn't have to dig through navigation menus, scripts and cookie banners to find the facts.

The format is simple. Per the proposal:

  • An H1 with your business name. It's the only required part.
  • A one-paragraph summary in a blockquote: what you do, for whom, where.
  • Sections of links under H2 headings, each pointing to a page with a one-line note: services, service area, pricing approach, policies, contact.
  • An “Optional” section for secondary pages a reader can skip when space is short.

The proposal also suggests offering clean Markdown copies of key pages. Its author has since published a second version, revised on the basis of two years of adoption.

What llms.txt isn't

  • It isn't robots.txt. It doesn't allow or block anything. If your firewall or crawler rules turn AI crawlers away, an llms.txt file won't let them in.
  • It isn't a sitemap replacement. Search engines still discover your pages the usual way.
  • It isn't an adopted standard. It's a proposal. No AI company is obliged to read it, and support has been uneven.
  • It isn't a way to buy a mention. Nothing you put in the file makes an engine recommend you. It can only describe what's already true.

Who actually reads it today

This is the part most sales pitches skip. In June 2025, Google's John Mueller wrote that “no AI system currently uses llms.txt,” as Search Engine Roundtable reported. In May 2026, Search Engine Journal reported that Google's own guidance for its AI search features lists llms.txt among the things you don't need. And in June 2026, asked again, Mueller called its future “purely speculative for now” (Search Engine Journal).

The signals aren't all in one direction. The same May 2026 report notes that Chrome's Lighthouse tool added an llms.txt check to a new “Agentic Browsing” audit, so another part of Google treats the file as an emerging convention for AI agents that browse sites on someone's behalf.

The honest summary: there's no public evidence that llms.txt helps a small business get named in ChatGPT, Gemini, Perplexity or Google's AI answers today. It may matter more as AI agents start browsing on behalf of buyers. Nobody can promise when.

So should your business have one?

For most small and mid-sized businesses the answer is: add it if it's cheap, don't pay a retainer for it, and never put it ahead of the basics. A reasonable rule of thumb:

  • Worth an hour if your site has a clear set of service, pricing-approach and policy pages, and your web person can publish a text file at the root without a ticket queue.
  • Not worth paying for as a standalone service, or as a reason to sign an “AI visibility” contract.
  • Not a priority if your core facts are still inconsistent across your site, Google Business Profile and directories. Fix that first.

Process first: the file is only as good as your fact sheet

An llms.txt file is a summary of facts your company agrees on. If sales quotes one service area, the website lists another and the directory listing still shows a phone number from 2021, the file just becomes one more place where the story disagrees. That's the same lesson as our guide to answer engine optimization for small business: write the one-page fact sheet first, then publish it everywhere, including here.

If you do write one, keep it to what a buyer would need to hire you:

  1. Business name, a two-sentence summary, the cities or regions you serve.
  2. One link per service page, each with a plain one-line description.
  3. Links to your licensing or certification page, your terms, and your contact page.
  4. Nothing you wouldn't say on the homepage. No claims you can't back up.

Where proactive AI fits: keeping it true, not writing it once

The real risk with llms.txt isn't that nobody reads it. It's that it goes stale. Someone writes it once during a website project, the business adds a service and drops a city, and a year later the file describes a company that no longer exists. Checking it depends on someone remembering to. That's a carburetor: it does less and asks more of you.

A proactive agent works like fuel injection. It reads the signal that already exists:

  • It compares your llms.txt, your service pages and your public listings against the fact sheet, every week.
  • When a page changes or disappears, it drafts the matching update to the file.
  • It tracks whether AI answers name you for your customers' real questions, as in our guide to getting your business recommended by ChatGPT, so you can see whether any of this moves anything.

What reaches you is only the exception: “Your llms.txt still lists drywall repair, which you removed from the site in September. Draft fix attached.”

The honest part: invisibility is earned

The first weeks are done mostly by hand. Someone on your team confirms the fact sheet and decides what counts as drift. That's the calibration. The agent earns autonomy with evidence: it suggests, then prepares changes for review. Anything published in your company's name, on your site or anywhere else, stays a human decision. An ECU manages the mixture. It doesn't choose the destination.

Where to start this week

Open your own site and check three things: that your crawler rules and firewall aren't turning AI crawlers away, that your service area and services read the same on every page, and that your Google Business Profile matches. If all three are clean, a basic llms.txt is an easy, low-stakes addition. If they aren't, that's where the hour should go.

If you'd rather have an agent keep your public facts consistent and tell you only when something drifts, built around your own site and systems by a dedicated engineer, see how our M2M and AEO work is set up, or talk to RubikSoft.

Want to see this in your company?

Schedule a free discovery call with RubikSoft and learn how an AI Agent can transform your operations in weeks.

Schedule Your Free Consultation