Taking over and reworking a health insurer's mobile app

The context

A health insurance mutual already had a mobile application for its members — their personal account area, a mobile port of a responsive web extranet. This was no early-stage product: more than fifteen major versions had shipped. We came in on an existing codebase, with its own decisions, history and constraints.

The challenge

Taking over someone else’s code is a particular exercise, especially on an application already in the hands of thousands of members. You cannot rewrite everything: every change has to be justifiable, and none may degrade what already works.

The sensitive point was authentication. It is the first screen, the one everyone sees, the one that decides whether the app gets used or deleted. A member who has to retype a complex password every time they check a reimbursement eventually stops opening the app at all.

What we did

• A native rebuild of the authentication journey. The sign-in screens were lifted out of the webview and rebuilt natively — this is the part where the difference in smoothness is felt immediately.
• Full integration with the Keycloak flow, the identity solution already in place on the client side. We adapted to their infrastructure rather than imposing ours.
• Biometric sign-in: fingerprint and face recognition, removing the password friction on every launch.
• Keeping the webview for the functional core of the member area. That business content was maintained on the web side by the insurer’s own teams: duplicating it natively would have created two sources of truth to keep in sync. A native layer exactly where it earns its keep, and nowhere else.

The outcome

The application is published on the App Store and Google Play, with a native, biometric sign-in journey on both platforms.

The defining decision on this engagement was not to rebuild everything. Redoing the member area natively would have been more spectacular, taken far longer, and forced the insurer to maintain the same business logic twice. The effort went into the 20% of the application where native genuinely changes the experience.