In nineteenth-century novels, when a family wanted to know what was being said about them at a gathering they hadn't been invited to, they did what families always did: found a distant relative with a good memory, a discreet household servant, or a well-placed acquaintance willing to report back faithfully. A confidant. Someone who could sit in that room without being noticed, then tell you, word for word, what they said about you. Something like that exists today. Considerably less dramatic, and considerably more revealing.
The meeting nobody invited you to
Every time someone asks ChatGPT, Perplexity, Google AI Overviews, or any other AI assistant «what do you recommend for X in Y city?», a conversation happens. A conversation about your industry, your market, what you sell. And in that conversation, someone else is taking the spot that could be yours.
Before these tools existed, the process was more visible. The customer searched on Google, saw the results, clicked. You could measure that. You could see where you ranked, at what position, how many times you'd been seen. There were data points — imperfect ones, but data.
Today, a growing share of that process happens in conversations that leave no trace in your Google Search Console. The customer asks. The AI responds. The customer decides. And you weren't there.
What the AI knows — and doesn't know — about your business
There are three questions I find fundamental when I talk to business owners or marketing directors about AI visibility. Each has a different answer for almost every company I know, and almost none of them have all three resolved at the same time.
The first: what does the AI know about your business when nobody asks about you directly? Not when someone searches your name. But when someone searches for what you sell, in your city, without knowing yet that you exist.
The second: when someone does ask for what you sell, who does the AI recommend? Do you appear? Does your competition? Does someone who isn't even your direct competitor but has better presence in the AI ecosystem?
The third: if you don't appear, what could you actually do about it? Not in six months. Not in theory. Concretely.
These three questions are what guide the AI visibility test we built at AEON42. The mechanics are simple: we ask an AI assistant — one that also searches the web in real time — exactly the same questions your potential customer would ask. Without mentioning you. Without giving it hints. Exactly the way someone who doesn't know you yet would ask. And we record, faithfully, what it responds.
The confidant who doesn't judge
What makes this kind of tool interesting isn't the technology. It's what it lets you see: the transcript of a conversation that used to happen without you, and about you.
Think about it for a second. For years, the only way to know what was being said about your company in word-of-mouth conversations was to ask someone who'd been there. A satisfied customer. An industry contact. Someone who'd tell you «hey, I recommended you to so-and-so» or, in the worst case, «they asked about you and nobody knew what to say.»
Conversations with AI are the modern version of that word of mouth. The difference is that now you can sit down and read exactly what the confidant said. What information it had about you. Who it recommended instead. And why.
That's not a small thing. It's probably the most honest competitive intelligence that exists today: no bias from whoever is relaying it, no politeness filter, no awkwardness of nobody wanting to tell you your company doesn't show up in the recommendations.
Here's where some people are going to stop reading:
There are companies with technically impeccable SEO. Fast site, clean structure, well-written metadata, decent rankings on Google. And they don't appear in any AI recommendation. None.
Not because they did something wrong. But because traditional SEO and AI visibility are two different things that feed on different sources. SEO talks to Google's crawlers. AI visibility — what we also call GEO, Generative Engine Optimization — talks to language models. And language models don't read your website the same way Googlebot does.
Put another way: you can have your house perfectly arranged for an inspector to approve it, and the neighbor who makes recommendations in the neighborhood doesn't even know you exist.
That's what I see happening all the time. Companies with solid digital presence, with years of SEO work, that don't appear when a customer asks ChatGPT which option it recommends. And smaller companies, with less overall presence, that do appear — because they have the type of content, structure, and authority that AI models recognize as trustworthy.
AI visibility isn't the same as Google ranking
GEO — Generative Engine Optimization — is the practice of optimizing your content so AI assistants cite and recommend you, not just so Google ranks you. They're complementary, but they're not the same thing. You can have both working well, just one, or neither. Most companies I know have the SEO side worked on and the GEO side untouched.
How the test works — no magic, no runaround
The mechanics of the AI visibility test are deliberately simple. You leave your name and email. You receive a personal link. You describe your company: industry, main product or service, city where it operates. That's enough.
You describe your company in customer terms
Not your company name. Not your corporate pitch. But how someone who doesn't know you yet would search for you: what you sell, where, for whom. That's what we ask the AI.
The AI responds without knowing you're the one asking
We ask the assistant the question your real customer would ask. Without mentioning you. Without giving it hints. The assistant responds the same way it would respond to any user: with what it knows, what it found, what it considers trustworthy.
You get a concrete result in a minute
Not a 40-page report. A result on screen and in your email: what the AI knows about your company, who it recommends when someone asks for what you sell, and three concrete actions you could take.
It's worth being clear about what the test is and isn't. It's a snapshot of a specific moment: what the AI knows about you today, in this query, with this question. It's not continuous monitoring — that's what AEON42, the SEO/GEO/AEO platform does, with 16 months of data from your Google Search Console. The test is the first look. The platform is the sustained work.
It's also worth saying the result can be uncomfortable. A company with years of digital marketing work can discover that the AI recommends a smaller competitor, or that it simply doesn't have enough information to recommend them at all. That's not failure. That's information. And uncomfortable information is the most valuable kind.
What to do when the result isn't what you expected
If the test shows the AI doesn't know much about you, or recommends someone else, there's work to do. It's not mysterious work or ultra-technical niche stuff. It's content work, structure work, building the kind of signals that language models recognize as authority on a topic.
Some things that generally help, and that I see working:
- Content that answers real customer questions — not corporate copy, but direct answers to what someone would ask an advisor.
- Consistent presence in sources AI models consider trustworthy: industry media, specialized directories, mentions on authoritative sites.
- Technical structure that AI crawlers can read without friction: structured data, schema, llms.txt.
- Clarity about what you do, for whom, and where — without jargon or euphemisms, without the AI having to guess.
None of this happens overnight. The SEO and GEO work that builds AI visibility typically takes between 3 and 6 months of consistent effort to show results in competitive positioning — and the starting point is always knowing where you stand today.
Which is exactly what the test gives you: the starting point. The diagnosis, not the treatment.
The persuasion you never hear
The newsletter this post came from was called «Persuasion.» Not because it sounds good. Because that's exactly what's happening when an AI recommends someone in your industry.
The AI is persuading your potential customer. It's telling them, with the confidence of something that appears to know everything, who to hire, where to go, which product to buy. And it does this before that customer reaches your site, before they see your ads, before anyone on your team has a chance to talk to them.
That persuasion happens with or without you. The question is whether the AI has enough information about your company to include you in that conversation — or whether it's recommending your competition because you simply don't show up on its radar.
Not knowing the answer doesn't protect you. It just leaves you out.
The test is at [aeon42.com/test-ia](https://aeon42.com/test-ia). Free. Result in a minute. Nothing to install, no site access required. You leave your name and email, describe your company, and we tell you what the AI knows about you and who it recommends when someone searches for what you sell.
If you'd like someone to walk you through improving that result, the Play.Interactive team can help. If you'd rather handle it yourself with your own tool, AEON42 is the option.
But first, the snapshot. First, knowing.
Frequently asked questions
No. The test gives you a quick snapshot of how the AI sees you right now: what it knows about you, who it recommends instead of you, and three concrete actions. A full SEO audit analyzes your technical structure, content, domain authority, and search rankings. They're complementary tools, not equivalents. The test is the starting point; the audit is the full picture.
Because traditional SEO and AI visibility are built differently. SEO talks to Google's crawlers. Language models like ChatGPT learn from different sources: text across the web, mentions in media, content structure that answers direct questions. You can have a technically impeccable site for Google and still be invisible to AI. It's not a contradiction — it's two games with different rules.
GEO and AEO — the disciplines that build presence in AI assistants — can show results on a timeline similar to competitive SEO: generally 3 to 6 months of consistent work, depending on existing domain authority. There are no real shortcuts, but there is a logical sequence: first know where you stand, then build from there.
No. The test runs the query through one AI assistant with real-time web search. It's a snapshot at a specific moment with that tool — not a multi-engine check and not something that runs automatically every day. For continuous monitoring there's AEON42 as a platform (ChatGPT and Perplexity coverage comes in its higher plans).
The main difference is perspective. If you ask ChatGPT about your own company, there are unavoidable biases: you know exactly what to search for, how to frame it, and it's hard to genuinely approach it with the eyes of someone who doesn't know you yet. The AEON42 test reproduces the question a potential customer would ask — without naming you, without giving the AI hints. That changes the result considerably.