llms.txt for local business

llms.txt is a small Markdown file at the root of your domain that gives AI agents a clean, machine-readable map of your site: what you do, where, and where the pages that matter are. To understand whether it’s worth your time, you have to ask the right question.

Ask the right question

There are two very different questions people lump together. The first is “will an llms.txt file help me show up in AI answers?” — a ranking and citation question. The second is “when a customer points an agent at my site to do something, does it help the agent succeed?” — a usability question. For a local business, the second one is the whole game.

On the first question, be skeptical. No major AI provider has committed to using llms.txt as a ranking or citation signal, and large studies have found no correlation between having the file and being cited in AI answers. Don’t add it expecting to rank.

What it does for an agent taking action

This is where it earns its place. Agentic browsers — ChatGPT Atlas, Perplexity Comet, and the agent features now in Chrome and Firefox — take an instruction like “book me a plumber for tomorrow” and navigate, fill forms, and complete the task on the customer’s behalf. An agent dropped onto your homepage has to figure out where to go. An llms.txt hands the ones that read it a direct map: here’s /book, here’s /services, here’s where the pricing lives. That’s a shortcut straight to the action you want it to take.

The agents that already consume it routinely are coding assistants and agent frameworks — Cursor, Claude Code, and the tooling behind autonomous workflows fetch llms.txt to find the right pages fast. Agentic-commerce systems shopping on a user’s behalf increasingly want the same machine-readable surface for catalog, pricing, and availability. It’s worth being ready for that wave rather than behind it.

The honest caveat

Most consumer agentic browsers today don’t start with your llms.txt. They read the rendered page the same way the rest of this tool measures it — through the accessibility tree, in an observe-plan-act loop. So whether the booking actually completes is decided first by your page being semantic and your forms labeled. llms.txt is a map that speeds up the agents which read it; it is not a substitute for a site an agent can navigate. The heavier-duty action signal arriving next is WebMCP, where your site registers tools — search, book, add-to-cart — that an agent can call directly instead of clicking through the page.

So should a local business add one?

Yes, as cheap, forward-compatible agent-readiness. It takes under 30 minutes, it gives action-taking agents a clean map to your key pages, and Chrome’s Lighthouse now audits for it. Just add it for that reason — helping agents act — not for a ranking bump that isn’t there.

What to put in it

# Acme Plumbing
> Licensed plumber serving Buffalo, NY and the Southtowns. 24/7 emergency service.

## Key pages
- /book - schedule an appointment
- /services - full service list and pricing
- /contact - phone, address, hours

How it differs from robots.txt

robots.txt controls what crawlers are allowed to access, and every major crawler respects it. llms.txt maps your site for the agents that read it. More on robots.txt for AI crawlers.

Scan your site to see whether you have one — and make sure the pages it would point to are ones an agent can actually use.