The patient record does not leave the building. The AI comes to it.

Health data is sensitive data under the revised FADP, and medical secrecy falls under art. 321 SCC. Any new form of processing calls for an impact assessment — hence an arrangement that can be described precisely.

Talk to an engineer Configure

01 — THE CONSTRAINT

Art. 321 SCC, revFADP art. 5(c) and art. 22

Sensitive data: explicit consent or a legal basis, and a mandatory impact assessment where processing presents a high risk to personal rights.

See every text

01 · USEFind a prior condition in a record running to hundreds of pages.

02 · USEPrepare a discharge letter from the stay, reviewed by the physician.

03 · USESummarise a ward protocol and answer nursing staff questions.

04 · USEKeep an access log usable for the impact assessment.

02 — THE OBJECTION

Does the machine open new access to records?

No. It applies the rights of the existing system: through it, a clinician sees only what they already see in the electronic patient record. The access perimeter does not change; what changes is the time it takes to find.

03 — THE SIZING

For a typical profile in this profession.

Computed, not chosen: headcount × peak ratio × weight of the work, against machine capacity. Your actual situation goes through the configurator.

Profile used	300 peopleFind, cite, summarise.
Concurrent sessions to absorb	45
Recommended machine	Numezis 43 machines — beyond a single unit
Required circuit	Dedicated T23 socket · 16 A circuit
List price	123'000 CHFservice 27'100 CHF / year

A case in this profession

A private clinic brings AI to patient records without the records leaving the building.

Private clinic

Before the machine

If the question is “where do we start” rather than “which machine”, the group's Advisory and Engineering practice handles exactly that.

The group