AI knows your name. It just doesn't recommend you.

Most businesses check their AI visibility by asking ChatGPT about themselves. If it describes them correctly, they relax. That's the wrong test. AI can know exactly who you are and still never mention you when a customer asks the question you exist to answer.

Aigentic is a UK AI advisory that checks whether businesses come up when their buyers ask AI who to use. Here's what we found for one client, a European medtech company with a new heart test, in September 2026.

Asked by name, every AI got it

We asked ChatGPT, Claude, Gemini and Perplexity about the product by name, in English and German. All four described it, in all 8 answers. Perplexity drew almost entirely on independent sources, such as health insurers and clinics, rather than the company's own site.

On that test alone, the company looked fine.

Asked the customer's question, no AI named it

Then we asked the four questions the company's own team said customers ask, such as how to find out most easily whether you're at risk of a heart attack, and whether AI can help avoid one. Same four assistants, same two languages.

Not one answer. Not on any assistant, in either language.

So who did AI recommend instead?

Mostly not a competitor. The assistants pointed people to risk calculators, standard heart scans, their GP and published guidelines from heart and health bodies. A few rival products did get named, in 4 of the 32 answers: Gemini named the same rival AI heart product in both languages, and Perplexity named three others, one of them with its price.

The striking part is that AI often described the exact gap this product is designed for. One ChatGPT answer said: "A risk calculator… doesn't definitively tell you whether you already have a significant coronary blockage." It named the problem, then recommended something else. It didn't know the solution existed.

Why does AI know you and still not recommend you?

Because the two questions are answered from different places.

Asked your name, AI looks for pages about you. There are plenty: your site, your press, your listings. So it finds you.

Asked the customer's question, AI looks for pages about the problem. For a health question that means guidelines, health foundations, studies and medical sites. If none of them mention you, you don't exist for that question, however well known your name is.

Two of the four assistants didn't even search the web for the customer's questions. Claude answered them from what it already knew, and only searched when asked about the product by name. Gemini showed no sources at all. For those answers, the only way in is to be in the material the AI has already learned from.

Is what AI says about you accurate?

Not always, even when it knows you. Asked by name, one assistant got several details about the product wrong, and another quoted a technical figure the company keeps out of its own patient-facing material. If you only check whether AI mentions you, you'll miss what it says about you.

What would we change?

This is our view, drawn from what the assistants cited:

1. Test the customer's question, not your name. Ask the questions your buyers ask, on more than one assistant, in each language you sell in. Checking your name gives false comfort.
2. Write the page that answers that question, in the customer's words, not just the product page. "How can I find out if I'm at risk of a heart attack?" is a question. A product page isn't an answer to it.
3. Get into the sources AI trusts for your topic. In health, that's guidelines, health foundations, published studies and the media that cover them. In other sectors, it's the trade bodies, directories and press your buyers' questions lead to.
4. Check what AI says about you when it does name you, and correct the sources it's drawing from.

Try it now: a prompt to check your own business

Copy this into ChatGPT, then into Perplexity. Fill in the brackets first. Use a fresh chat, logged out if you can, so it isn't answering from what it already knows about you.

You are an analyst checking whether AI recommends my company. Be brutally honest. Do not flatter.

Company: [NAME]
What we do: [ONE LINE]
Main competitor: [COMPETITOR]
Market: [SECTOR AND COUNTRY]

Answer each question the way you would for a real customer:
1. "[A QUESTION YOUR CUSTOMERS ASK, IN THEIR WORDS]" Do you name [NAME]? Who or what do you recommend instead?
2. "[A SECOND CUSTOMER QUESTION]" Same again.
3. "What are the best [CATEGORY] companies?" Do you name [NAME]?
4. "What is [NAME] and what do they do?" Is anything wrong or confused?
5. For each answer, list the sources you would rely on.

Then score [NAME]'s AI visibility from 0 to 10 and give the three biggest reasons a customer would not find or trust them.

Treat the result as a smoke alarm, not a full inspection. If questions 1 to 3 come back without your name while question 4 knows you well, you have the problem this article describes.

Key facts

• One European medtech company, four AI assistants (ChatGPT, Claude, Gemini, Perplexity), English and German, September 2026.
• Asked about the product by name: described in 8 of 8 answers.
• Asked the four questions the company's customers ask: named in 0 of 32 answers.
• Instead, the assistants recommended risk calculators, standard heart scans, GPs and guidelines; rival products appeared in 4 of 32 answers.
• Claude answered the customer questions without searching the web; Gemini showed no sources at all.

FAQ

If ChatGPT knows my company, will it recommend me? Not necessarily. In Aigentic's checks for a European medtech company, all four major AI assistants described its product correctly when asked by name, in 8 of 8 answers, but named it in none of 32 answers to the questions its customers actually ask.

How should I check whether AI recommends my business? Ask the questions your customers ask, not your company name, on more than one assistant and in every language you sell in. In Aigentic's checks, a company that looked fine when searched by name was absent from every answer to its customers' real questions.

Why does AI recommend other options instead of my product? Because it answers a customer's question from sources about the problem, such as guidelines, health bodies and trade media, not from pages about you. If those sources don't mention you, AI recommends what they do mention, even when it describes the gap your product fills.

Does a page on my website help with every AI assistant? Not on its own. In Aigentic's checks, Claude answered general customer questions without searching the web, and Gemini showed no sources. For those answers, you need to be in the wider material AI has learned from, such as studies, guidelines and press.