How to Get Your Business Cited by ChatGPT, Claude, and Gemini

Someone asks ChatGPT "what's the best accounting software for freelancers?" or "find me a dentist in Portland that takes new patients" or "compare the top three project management tools for remote teams."

Your business should be in that answer. But it's not. Or it is, and the AI got your pricing wrong, your hours wrong, or described you as "a mid-range option" when you're actually premium.

This isn't a ranking problem. It's a readability problem. Here's exactly how AI assistants source their answers, and what you can do to make sure your business is the one they cite.

1. How AI Assistants Actually Source Business Information

Before optimizing for AI citations, understand where the information comes from. There are three sources. You control two of them.

Source 1: Training Data (You Cannot Control This)

Every major AI model is trained on a snapshot of the internet. If your business existed and had a web presence when the training cutoff happened, some version of your business facts is baked into the model.

The problem: training data is frozen. If you changed your pricing, added a service, moved locations, or updated your hours since the cutoff, the model still has the old version. It has no way to know it's outdated unless it finds a fresher source.

Training data is a snapshot. Your website is the live version. The AI will prefer the live version, if it can read it.

Source 2: Real-Time Web Search (You Partially Control This)

When a user asks a question that requires current information, AI assistants search the web in real time. They fetch multiple URLs, extract content, and synthesize an answer.

This is where traditional SEO still matters. If you rank on page one for relevant queries, you're in the candidate set. But being in the candidate set is not the same as being cited.

Here's what happens after you rank:

1. The AI fetches your page along with 5 to 15 others
2. It extracts text content from each page
3. It converts the text to tokens
4. It evaluates which sources have the most relevant, verifiable facts
5. It synthesizes an answer, citing the best sources

Steps 2 through 4 are where most businesses lose. If your page is 5MB of JavaScript, images, and marketing fluff, the AI extracts very few usable facts. If your competitor's page is clean, structured, and fact-dense, they win the citation.

Source 3: Structured Profiles (llms.txt, You Fully Control This)

llms.txt is a plain markdown file at your domain root that gives AI agents a structured summary of your business. It's built for AI consumption.

When an AI agent finds an llms.txt file, it treats it as a preferred, authoritative source. The file is owner-controlled: you wrote it, you host it, you keep it updated. It's structured for AI: plain text, no HTML, no JavaScript, no design. Every token carries meaning with no marketing fluff. And it's predictable: agents know exactly where to find it at yoursite.com/llms.txt.

llms.txt is a direct, owner-controlled channel between your business and every AI agent that will ever ask about you.

2. The Citation Decision Framework

AI agents don't browse your website. They run a structured evaluation. Based on our analysis of thousands of AI responses, here's the framework they follow:

The pattern is clear. Accessibility of facts is the competitive moat. A small business with clear, structured data will outperform a large competitor whose information is locked in PDFs, images, and JavaScript.

3. The Step-by-Step Playbook

Step 1: Audit Your Current AI Readability

Before you fix anything, understand what an AI agent actually sees. Run a Site Scan on your domain. It checks 8 critical signals across 4 categories: discovery (llms.txt, agents.md, sitemap.xml), metadata (meta description, robots.txt rules), trust (JSON-LD schema, Open Graph tags), and efficiency (homepage token weight, estimated agent research cost).

Most businesses score below 40 out of 100 on their first scan. The good news: the fixes are usually simple.

Step 2: Deploy an AI Website Profile (llms.txt)

This is the single highest-impact action. An AI Website Profile is a structured markdown file that contains everything an AI agent needs to understand and recommend your business: name, description, category, products and services with pricing, locations, hours, contact information, key differentiators, proof points, and links to deeper pages for specific topics.

Deploy it at yoursite.com/llms.txt. ChatGPT, Claude, Gemini, Perplexity: all crawl for this file.

Step 3: Add Schema.org Structured Data

Schema markup gives AI agents typed, explicit facts instead of forcing them to infer meaning from prose. This reduces hallucination rates by 40 to 60%.

Priority schema types: Organization or LocalBusiness, Product plus Offer with price, FAQPage for answers agents can cite directly, Review for aggregate rating and individual reviews.

Step 4: Fix Your Content Structure

AI agents weight the beginning of pages heavily. 44.2% of AI citations come from the first 30% of content. Put your most important facts first.

• Lead with concrete claims, not marketing preamble
• Use semantic HTML: h1 through h6, tables, article tags
• Convert PDF menus and image-based pricing to HTML text
• Make sure your Google Business Profile matches your website facts

Step 5: Monitor and Maintain

AI platforms evolve. New standards emerge. Your business changes. Treat your AI Website Profile as a living document. Update it when your pricing, services, or hours change. Run periodic scans to catch regressions.

4. What Success Looks Like

When your AI readiness is in place, here's what changes:

• ChatGPT cites your business with accurate facts instead of guessing
• Claude extracts your services, pricing, and differentiators in a single pass
• Gemini finds your business for local queries with correct hours and contact info
• Perplexity surfaces your content when it's fresh and relevant

You don't need to rank #1 on Google for this to work. You need to be the easiest business for an AI agent to understand.

5. The Businesses Already Winning

Over 844,000 websites have deployed llms.txt, including Stripe, Cloudflare, and Vercel. Restaurants are replacing PDF menus with structured HTML plus MenuItem schema. SaaS companies are publishing pricing in machine-readable formats instead of "Contact Sales" buttons. Local service businesses are making sure their Google Business Profile, website, and llms.txt all tell the same story.

The common thread: they made their business facts accessible. Not more persuasive. Not better designed. Just readable.

Run a free Site Scan at platinum.ai. You'll get a score, a breakdown of what's working and what's not, and a prioritized list of fixes. Most businesses go from AI-invisible to AI-readable in under an hour.