llms.txt: the 30-minute file that introduces your business to AI
What llms.txt actually is, what it can and can't do for AI visibility, and how to write one that ChatGPT and Perplexity can actually use.
Somewhere on your website there’s a robots.txt file quietly telling Google what it’s allowed to crawl. It’s been there for decades and you’ve probably never thought about it.
llms.txt is its new sibling — except instead of telling AI what it can’t do, it hands AI a menu of what you are. One plain-text file at yourdomain.com/llms.txt that says: here’s the business, here’s what we sell, here’s where we operate, here are the pages that matter. Thirty minutes of work, and most of your competitors haven’t heard of it.
What it actually is
llms.txt is a proposed standard: a markdown file at your site’s root written for language models rather than browsers. Where a human gets your homepage — hero animation, cookie banner, mega menu — an AI assistant gets a clean brief:
- Who the business is, in one paragraph
- Key facts: location, hours, contact, service area
- What makes you different (the stuff buried in your About page)
- Your core pages, linked and described in one line each
Think of it as the elevator pitch you’d give a new employee on day one — except the new employee is every AI assistant that might one day recommend you.
The honest part: what it won’t do
Let’s not oversell this. llms.txt is not a ranking hack, and adoption is uneven — some AI crawlers read it religiously, others ignore it entirely. Anyone promising “add llms.txt, appear in ChatGPT” is selling you a magic bean.
What it actually does is cheaper and more boring: it removes friction. When an AI system does crawl you — and they decide which brands to name based on what they can confidently extract — a clean llms.txt means there’s zero ambiguity about who you are, what you do and where. Ambiguity is what gets you skipped.
Low cost, real upside, no downside. In our book that’s not a gamble, that’s insurance.
What a good one looks like
The mistakes we see are always the same: companies paste their homepage copy (“We’re passionate about excellence!”) and give the AI nothing to work with. Write facts, not vibes:
- Lead with the category and the place. “Family-owned plumbing & septic company in Canton, Georgia” beats three paragraphs of brand story.
- List what you sell, one line each, with links. Every service, plainly named — this is not the place for creative naming.
- Include the differentiators as claims, not slogans. “One crew handles both plumbing and septic — rare locally” is quotable. “We go the extra mile” is noise.
- Keep it current. A stale llms.txt that contradicts your site is worse than none; generate it from your site data so it updates itself.
That last point is how we build them: on our client sites, llms.txt is generated from the same data that powers the pages, so every new service or location shows up automatically.
Where it fits in the bigger picture
llms.txt is one brick. AI visibility — the discipline now called GEO — is mostly about being citable: answer-first pages, structured data, real facts stated plainly, the searches that still convert handled better than anyone else in your market. If your content is mush, a beautiful llms.txt won’t save you. If your content is sharp, llms.txt makes sure nothing gets lost in translation.
That’s the whole GEO playbook in one sentence, honestly: make it effortless for machines to understand you and safe for them to recommend you. It stacks on top of SEO, not instead of it — the same clean structure that ranks on Google is what gets you quoted by Perplexity.
Do this today
- Open yourdomain.com/llms.txt. If it 404s, you have homework.
- Write the brief: one paragraph of who/what/where, key facts, services with links, differentiators as facts.
- Put it at the root, plain markdown, and keep it in sync with the site.
Or skip the homework — we set this up as part of every GEO engagement, wired to your site’s data so it never goes stale.
