// Medtech

Medtech: where a mistake costs more than release speed.

We build medical products at the intersection of AI, computer vision and applied analytics. What sets the domain apart: a product can't be judged by model metrics alone — it embeds into a specialist's work, falls under regulatory requirements and has to be explainable.

What we have already built

Clinical decision support

NovaMed: multimodal analysis of patient history, lab data and records, probabilistic diagnostic recommendations. The product supports the doctor's decision, it does not make it.

Computer vision on medical images

EyeMesearing: automated calculation of ophthalmic parameters from an eye photo (MRD1, MRD2 and others), segmentation, integration with clinics and telemedicine.

Software as a medical device

If the product falls under the medical-device perimeter, that changes engineering: you need traceability of decisions, model versioning and documented behaviour. We build that in rather than bolting it on later.

Clinic and telemedicine integrations

Integration with clinical systems and telemedicine flows with no manual gaps: the result has to land where people actually work with it.

// How we work in this domain
A safety protocol from day one: guardrails, model-quality control, transparent recommendations.
The product supports the specialist's decision and does not replace them. That's not a disclaimer, it's a design constraint.
Responsibility and regulation remain with the client; we build in the traceability that makes them passable.
Client names, metrics and screenshots are published only with written consent.
// FAQ

Will AI replace the doctor?

No. We design the product as an assistant: it speeds up the work and surfaces what's easy to miss, but the final decision stays with the human.

What changes if the software is a medical device?

Requirements appear for traceability, documentation and change control. In practice: model versioning, decision logs and reproducible behaviour. Registration is the client's process; we prepare the product for it technically.

How is medical data protected?

Architecture is chosen to fit privacy requirements: access control, logging and, where needed, self-hosted models so data never leaves the perimeter.

Can we see your medtech cases?

The public versions are described on the NovaMed and EyeMesearing pages. Details, metrics and screenshots are shown under NDA on a call.

A medtech task in a sensitive domain? Let's discuss.

Start a project