Why subscribe — and why this price?

An honest answer, on one page.

In one sentence

We help you prove that the data training your AI is clean and compliant with the AI Act (Article 10) — and stays that way over time — without your data ever leaving your servers.

The problem you have (even if you don’t see it yet)

If your AI system — software that decides or predicts on its own from data (grant a loan, screen CVs, flag fraud…), as opposed to hand-written rules — is high-risk under the AI Act (credit, insurance, HR/recruitment, healthcare, graded education…), the law doesn’t only require that your AI performs well. It requires you to document and prove the quality of the data that trained it (the past examples it “learned” from): where it comes from, whether you have the right to use it, whether it is representative, accurate, complete, and whether it creates bias.

“Bias”, in plain words: a systematically unfavourable treatment of a group of people — e.g. denying credit more often to one gender, or to the residents of a neighbourhood, not because the actual risk warrants it, but because your historical data already leaned that way. The AI then reproduces the imbalance present in the data. The AI Act wants you to detect and correct it.

Most teams have polished the AI itself (the program making the decisions). Almost no one has the data dossier that goes with it. Yet that dossier is what an auditor, a large-account customer or a regulator will ask for — and its absence can block a market launch or lose you a contract.

What we actually do

1. We classify your risk first. Often, our first value is telling you ”you do NOT need all this”: your system isn’t high-risk, or some obligations don’t apply to you. We don’t sell fear.
2. We analyse your data where it lives. Our tool runs on your side (in-place): it reads your data, produces a report (duplicates, missing values, personal data, bias) and a structured Article 10 dossier.
3. We maintain it over time. Compliance isn’t a one-shot PDF. When your sources change, we detect it and update the dossier. It’s a subscription because it’s a state, not a deliverable.

Why you can trust us with this (without trusting us with your data)

This is what passes a security review (IT/CISO):

• The agent that runs on your side is open-source and auditable — your team reads what it does, line by line.
• It is technically incapable of sending your data elsewhere: we run it network-off, and a manifest proves it at runtime. Not a marketing promise — verifiable.
• The analysis engine stays our property, but it is fenced in by that agent: even it can’t exfiltrate anything.
• Operated in the EU. We keep only the dossier / metadata, never your data.

Why this price

Compare it to what the subscription replaces:

Our approach: software that does most of the work at near-zero marginal cost, + a thin layer of expert when your case needs it. That’s how we can be far cheaper than a consultancy while being continuously maintained.

Order of magnitude: a subscription in the tens of k€/year, depending on your scope (number of sources, vertical, support). The price follows value — unblocking a launch, securing a contract — not the volume of your data.

We prove it before you pay

1. A 2-minute questionnaire: are you concerned? (often the answer is “no”).
2. A demo on synthetic data: you see exactly what the dossier contains.
3. A free trial on a sample of YOUR data, locally.
4. Only then: an in-place pilot, then the subscription.

What we are not (to be clear)

• Not a law firm: our output is not legal advice (but it gives your lawyer/DPO something concrete to work with).
• We don’t magically “clean” your data: we reveal, document and maintain — the business decisions remain yours.
• We don’t want to sell you what you don’t need. That’s our first instinct — and it’s exactly why we’ll earn your subscription the day you truly need it.

Indicative, not legal advice. The AI Act is evolving (Digital Omnibus) — confirm for your specific case. Your data stays with you; processing operated in the EU.