The six applications were not designed as a family; they were inherited through acquisitions and kept alive as-is. Each carried its own login, its own device-data pipeline, its own alert configuration and its own hosting stack. Under the surface, engineering teams were maintaining six versions of essentially the same plumbing, and every fix, security patch or improvement had to be repeated across all of them.

Hospitals bore the integration burden. A customer using several of the products had to integrate the same vendor six times: six connections into their clinical systems, six sets of credentials for staff, six alerting behaviours for clinicians to learn. Onboarding a new hospital took a full quarter of bespoke work, which throttled how fast the company could grow and how quickly monitoring reached patients.

The duplication also set the ceiling on product velocity. With every team anchored to its own stack, a new clinical-facing module took an 11-month cycle to deliver, and hosting and tooling costs multiplied across six parallel infrastructures that all did the same job. The company was effectively paying six times over just to stand still.

We started with the architecture, not the applications. The design pulled the common concerns into shared services: a single ingestion tier for device telemetry regardless of which product the data feeds, one identity and access layer covering every user, and a common FHIR-based integration gateway so a hospital connects its clinical systems once and reaches everything the vendor offers.

The six products did not need six user interfaces rebuilt from scratch. We delivered a modular application shell, and each product migrated into it as a feature module, keeping its clinical functionality while shedding its private login, pipeline and hosting stack. Migration ran product by product over 14 months, so hospitals saw a steady sequence of small transitions rather than one disruptive event.

Retiring the legacy stacks safely mattered as much as building the new platform. Each old application was withdrawn behind an API compatibility layer, so hospital integrations built against the legacy interfaces kept working while the systems beneath them were replaced. Nothing was switched off until the shared platform had fully taken over its load.