The One File Your Website Is Missing That Tells AI You Exist

Roughly 90% of small business websites are missing a file that takes about an hour to create and tells every major AI system on the planet what the business is about. It's called llms.txt. Combined with the schema markup that should already be on your site, it's the closest thing to a one-time fix that exists for AI search visibility.

This post is going to be unusually concrete, because the topic gets more abstract than it needs to. The actual steps are short. The reason they matter is what's worth understanding.

What llms.txt actually is

Think of llms.txt like robots.txt's smarter sibling. Robots.txt tells search engine crawlers which pages they can and can't read. llms.txt tells AI systems — ChatGPT, Claude, Perplexity, Gemini — what your business is, what's on your site, and where the useful information lives.

It's a plain text file. It sits at the root of your domain, at /llms.txt. It's structured as Markdown, which means it's readable by humans and parseable by machines. A small business llms.txt might look like:

• A one-paragraph description of who you are and what you do
• A short list of the most important pages on your site, each with a one-line description
• A list of FAQ topics with the URLs that answer them
• Optional: contact info, service area, hours

That's it. Most small business llms.txt files run 30-60 lines. The whole thing should take an hour to write the first time.

Why an AI cares about a text file

Modern AI systems don't read your website the way Google does. Google's crawler has been parsing HTML for twenty-five years and is excellent at it. Most AI systems are less excellent. When ChatGPT or Perplexity tries to understand your business from your website, it's reading rendered HTML through a tool less sophisticated than a 2010-era Google crawler. It misses content behind interactions. It misses content in single-page apps. It gets confused by complicated layouts.

llms.txt skips all of that. It hands the AI a clean summary in a format designed for machines to read. The result: when someone asks ChatGPT about businesses like yours, your business has dramatically better odds of being the one it can describe accurately.

Schema markup is the other half

llms.txt is one signal. Schema markup is the other big one, and it's been around longer.

Schema markup is invisible code in your website that labels every part of the page. "This block is a phone number. This block is a service. This block is a customer review with a 5-star rating." Without it, the AI has to guess. With it, every part of your site is labeled and confidently quotable.

For a local business, the schemas that matter most:

• LocalBusiness (or a more specific subtype like Plumber, Dentist, AutoRepair, Attorney). This is the foundation. Name, address, phone, hours, service area, geo coordinates.
• Service. One per service you offer. What it is, what it costs (a range is fine), who it's for.
• Review and AggregateRating. Your reviews, summarized in a structured way the AI can reference.
• FAQPage. For pages with frequently asked questions. Google retired the FAQ rich result in May 2026, so this no longer earns you a SERP feature — but the markup still hands AI answer engines a clean question-and-answer graph about your business.
• Person. If you have a personal brand — a doctor, a lawyer, the named partner — markup that establishes who the human authorities are.

What you have to do, in order

Step 1: Audit what's already there

Run your homepage through Google's Rich Results Test (free, no signup). It tells you what schema is currently on the page and what's missing or broken. For most small business sites built on templates, the answer is "very little, and what's there is generic."

Step 2: Add or fix the schemas listed above

This is a developer task in most cases. It's about a half-day of work for a site that doesn't have it yet. The work pays for itself in months, not years, because every AI tool reading your site benefits from it.

Step 3: Write llms.txt

A template that works for most small businesses:

• One-paragraph business description with city/region and what you do
• A "Key Pages" section listing the URLs for: homepage, services overview, about, contact, top three service pages
• A "FAQ" section listing the URLs for any pages that answer common questions
• Hours and contact info in plain text at the bottom

Save as llms.txt. Upload to the root of your domain. Confirm it's accessible at yourdomain.com/llms.txt by visiting that URL.

Step 4: Make sure your sitemap.xml exists and is updated

This is the boring one nobody talks about because it's old news. Your sitemap.xml is still the canonical list of what's on your site. AI tools use it. Many small business sites have sitemaps that are months out of date, missing pages, or missing entirely. Check yours.

Step 5: Test

After a few weeks, try the same query you ran before — ask ChatGPT or Perplexity about businesses like yours. Compare. If your business has moved into the answer, you're winning. If not, the next move is to look at the signals these tools weight most heavily after the technical foundation: reviews, mentions in third-party sources, and content that directly answers customer questions.

The mistake people make

The mistake is treating llms.txt and schema as the answer. They're not. They're the foundation. A perfect llms.txt on a site with no reviews, no real content, and no third-party mentions still won't get cited, because the AI is doing more checks than just reading your text file.

Schema and llms.txt make you legible to the AI. Reviews, content, and third-party mentions make you worth recommending. Skip either half and you're not in the answer.

What this looks like done well

A dental practice we know in Florida did exactly the steps above in late 2025. By March 2026 they were appearing in Perplexity answers for "best dentist in [their city]" — the only practice in their market that consistently did. Their organic traffic from AI tools was small in absolute terms (maybe 40-50 sessions a month) but it converted at three times the rate of their Google traffic. The customers arriving from AI tools were further along in their decision and ready to book.

The total time investment to get there: about six hours of work spread across two weeks. The cost: under a thousand dollars including the developer time to add schema. The window is open right now because most competitors haven't done any of this yet.