Skip to content
ceeme

About

Who is behind ceeme, and why is that reliable?

ceeme is a Belgian product of DB Events, built by Njusja Orban, that measures whether AI assistants mention, cite or choose a company — and keeps the measurement and the source behind every figure. It runs in four languages, measures on the engines buyers actually use, and its quality gate runs on a different AI provider from the writer, so no model approves its own work.

By Njusja Orban, founder of ceeme5 min read

Measure your own company → Rather talk it through first

In short

Where
Belgium. Four languages, and the examples follow the reader's market — not their language.
Who
Njusja Orban, founder. One name under every piece on this site, with a date next to it.
What to hold us to
Every figure carries its measurement. No figure on this site without origin and date.

Who built ceeme, and from which country?

ceeme is made by Njusja Orban and published by DB Events, a Belgian company. That is not atmosphere but liability: one name sits under every written piece on this site, with the day it was written, and that name answers for what it says.

The size of the team is a fact, not a selling point. What it does determine is the shape of the product: everything that has to happen twice has become code, and everything that requires judgement stays human work with a name under it. Where that could go wrong — an AI writing text and the same AI approving it — a hard separation is built in, and it is set out below.

Who checks that a ceeme text holds up?

Research from 2026 into self-preference bias shows that a model judging its own text approves its own mistakes over half again as often, and that a more advanced model is no less prone to it. What does help is a rubric with several axes plus a jury from a different model family. Both are in place: six axes each with their own floor, and a single red axis is enough to reject.

And the jury gets no hint about who made the text. A separate 2025 study put the same texts up four times, with and without a name and twice with a false one: the name alone shifted the verdict, by up to fifty percentage points in preference, and not always upwards — one name lifted the score, another pushed it down. Blind judging costs us nothing extra and removes exactly that part. Self-preference itself remains, and that is what the other model family is for.

What happensWhy that choice
the measurement itselfHundreds of times per scan. An expensive model would make the measurement unaffordable and no more accurate.
writing the textWrites in your voice, follows instructions closely. A better model exists and costs three times as much.
rejecting the textA different family from the writer, and blind: the jury does not know who wrote the text.
ordering the fix listThe only step where real reasoning happens: which fix first, and why that one and not the other.

At ceeme, what is allowed to be promised, and what is not?

A guaranteed outcome is not allowed, and that is not caution. Nobody owns the answers of an AI assistant: the provider swaps its model, reweights its sources and changes its behaviour without notice. Guaranteeing a position in something you do not own is selling a bet with an invoice attached.

What is allowed: asking the same questions again on the same engines, making the difference visible, and leaving it visible when it went the wrong way. A badge we do not have is not shown here — no audit certificate, no enterprise sign-in over SAML, no availability guarantee. The day there is one, it will be here with its date.

Which markets is ceeme built for?

Four languages, and an example follows the market and not the language. A Dutch-speaking reader sees a business in Ghent, a French-speaking one sees Liège, a German-speaking one sees Hamburg — because German is not automatically Belgian German. The English version names no city, because an English-speaking reader is not located by default.

That is not editorial courtesy: the market you enter determines which engines carry weight and which questions get asked. An example that follows the language instead of the market shows you something that would never be measured that way for you.

Frequently asked questions

Is ceeme an agency or a tool?

Both, and that is a choice rather than a hesitation. You can measure only and execute yourself, or have the work done. What is not possible is execution without measurement: then there is no way to see afterwards whether it did anything.

How many clients do you have?

That number is not here, because it cannot be checked on this site and it changes weekly. What is here is what was measured, when, and by which method — that can be recalculated and a client count cannot.

Why are there no client logos?

Using a name requires a signature, and we do not sign it ourselves. Until there is a signed reference, a real case is called "a chain of photo booths in the Netherlands" here — recognisable enough to mean something, without putting anyone on our page unasked.

Read on

Free, no account and no card.