AI brand checker: does a language model know you exist?
We give a language model nothing but your company name, your sector and your city, and ask whether it knows you. If it doesn't recognise you, that's a signal: too little has been written about you in the places these models learn from. You get the weak points back, and what you can do about them.
Read this with your result: this score is the verdict of one language model with no live internet access. It isn't a measurement across ChatGPT, Gemini and Perplexity together; if you want to see those three side by side, there's a separate block further down for that. The score is an estimate, not a measured position. What it does show is whether your company has a footprint big enough to be recognised.
This was one question about your name. What this doesn't show you is whether you turn up in the questions your customers actually ask. That's what the AI Index NL runs for: a fixed set of questions per sector, repeated monthly, with the previous measurements alongside it. Running that list for you costs €75 once, or €75 a month if you want to keep tracking it.
The check above asks one model and gives you a score. This block puts the same question to three models side by side and gives a precise answer per model: does your company name appear among the ten names it mentions, yes or no. No score, and deliberately no list of the companies that are mentioned. Those names come from training data and can be outdated or invented; showing them as a measured result would be a gamble with another company's name in it.
We use the company name, sector and city from the form above. Only your sector and your city go to the three models: the question is which companies they'd recommend in that trade in that place. Your company name stays on our server and is only checked against the answer to see whether it appears in it. As with the check itself, we store those three fields and the result, along with the page you arrived from; how and for how long is set out in the privacy statement NL. Who those three providers are and where they're based is on the sub-processors page.
- The name is distinctive enough not to be confused with another business
- Sector and place are unambiguous, so a model can do something with them once there are sources
- No mentions outside its own website and social channels
- Nothing on the site worth citing: no original figures, no explanation that isn't found anywhere else
- No place in the lists or overviews a model comes across during training
Why this is different from Google
In Google you compete for a place in a list. There are ten results, and with work you can move up. With an AI assistant there is no list: one answer comes back, and in it you're either mentioned or you're not. There is no place eleven.
The second difference is where the answer comes from
Google draws on an index it refreshes continuously. A language model draws on what it saw during training, topped up with whatever it's allowed to look up. That means your own website counts for less than you're used to, and what others write about you weighs more heavily. A mention in a trade publication, a sector guide or a well-read forum counts for more here than an extra page on your own domain.
That's also exactly why this scores low for most local businesses, and why that needn't be a problem
The question is whether your customers search this way. For a bakery round the corner, barely yet; for a B2B service provider, increasingly.
We don't score well here ourselves either
It would be strange to offer a tool that measures whether you're mentioned without saying how we ourselves fare. Fair's fair: it isn't great. Beyond our own site, almost nothing is written about us, and that's exactly the weakness this check finds in others too.
What we're doing about it is the same as what will appear in your own list of steps: making things people reference of their own accord
The Claude skills we publish, the free tools on this site, the scanner. It's slower than buying mentions somewhere, but it's the only thing that lasts.
Our own measurement is public in the AI Index NL, including the parts where we score badly.
Frequently asked questions
Does this tool really ask ChatGPT, Gemini and Perplexity?
Not for the score. The 0 to 100 score comes from one language model, GPT-4o, which answers the question from its training data and doesn't look at the internet. The separate block further down this page does put the same question to three models side by side: gpt-4o-mini at OpenAI, gemini-2.0-flash at Google and sonar at Perplexity. That block gives just a yes or no per model, and Perplexity does search the web for it, the other two don't.
So what does the result actually tell you?
Whether your company has a footprint big enough for a language model to know you at all. That's a real signal: models are trained on what's publicly written about you. If you're mentioned nowhere, there's also nothing to train on, and that doesn't change on its own when a new version comes out tomorrow.
Why don't you show competitors?
Because the model dredges those names up from its memory without being able to check them. They can be outdated, based in the wrong city or not exist at all. Showing a list of companies we claim do better than you as the outcome of a measurement, when it's actually a guess, isn't something we do.
Is a low score bad?
For most local businesses a low score is normal. A language model knows the big brands and the players that a lot has been written about. If you operate locally and mainly show up in Google Maps, that isn't a failure, it's a starting point. The question is whether your customers search via AI, and that's growing.
What does it cost and what happens to my data?
The check is free and no account is needed. Your input is logged so we can see what the tool gets used for. Email address and phone number are optional: leave them, and you're giving us permission to contact you about the result.
Do you measure your own AI visibility too?
Yes, it's public in the AI Index, including the parts where we score badly. Right now our own weakest point is exactly what this tool measures: almost nothing is written about us beyond our own site. That's stated plainly.
Start with your own site
Being mentioned by an AI starts with a site that can actually be read. If your technical setup is broken, or it isn't clear what you do and where, no system has anything to work with. The free scan checks that in a few seconds, including whether your page is readable for AI systems.
Questions about this tool?
Email us or book a call. You'll get an answer from the person who built it, and if something doesn't work the way it says here, we'd like to hear about it.