About the client
The manufacturer sells energy-efficient heating and HVAC systems in several international markets. Its heat pumps and controllers are connected, and three groups depend on that connection: homeowners who want to control heating remotely, field technicians who install and commission systems, and service partners and administrators who manage installations at scale. Each group needs the same device data presented in a way it can act on, in its own language.
Real-time control for a large device fleet, on every platform
- Consistency across platforms. IOS, Android and web had to behave and perform the same way.
- Data people can act on. Complex heating and device data had to make sense to a homeowner, a technician and an administrator.
- Testing without the hardware. Apps had to work with several device brands while physical units were scarce.
- Scale. Real-time updates had to stay reliable across a large, growing fleet.
- Third-party integration. Smart home platforms and external APIs had to fit without fragmenting the experience.
- Release speed. Manual testing was slowing releases and needed to be reduced without lowering quality.
What FiftyFive built
A dedicated FiftyFive team of eleven, working in Agile sprints, develops and runs the ecosystem: native mobile apps in Swift (iOS) and Kotlin (Android) for homeowners and technicians; web portals in React and Redux for service partners and administrators; backend services in .NET Core, Node.js and C#; cloud infrastructure on Azure IoT Hub, Azure Service Bus, Cosmos DB, PostgreSQL and Redis; automated UI testing with Java, Appium and TestNG, run through GitHub Actions CI/CD.
How the solution works
Device connectivity
- Connected heating units communicate through Azure IoT Hub, which handles device connectivity and message ingestion at fleet scale.
- Azure Service Bus decouples incoming device messages from the services that process them, so bursts of telemetry do not slow the apps.
Data and caching
- Device state and telemetry are held in Cosmos DB, structured application data in PostgreSQL, and frequently read state in Redis.
- This split keeps real-time reads fast for the apps while preserving relational data for portals and administration.
Backend services
- Services in .NET Core, Node.js and C# expose device data and commands to every client and handle integrations with third-party APIs and smart home platforms.
Apps for each audience
- Homeowners and technicians use native Swift and Kotlin apps; service partners and administrators use React and Redux web portals.
- Each presents the same device data in a form that suits the user's role, in multiple languages.
Automated quality
- UI tests written in Java with Appium and TestNG run in GitHub Actions on every change.
- Automated testing reduces dependence on scarce physical hardware and keeps release quality steady as the apps change.
Technology stack
What the solution enabled
Business transformation
Lower manual testing costs
Faster installation and commissioning
Latency across a large connected-device fleet
Sub-two-second latency at fleet scale
The architecture supports a large connected-device fleet with latency under two seconds, so monitoring and control stay real-time.
40% lower manual testing costs
Automated UI testing in CI/CD cut manual testing effort and its cost by 40%.
30% faster installation and commissioning
Technicians complete installation and commissioning 30% faster, reducing time on site.
Multilingual rollout across international markets
With remote monitoring for homeowners and technicians.
FiftyFive Technologies develops the connected-device apps of a European heating and HVAC manufacturer: native iOS and Android apps for homeowners and technicians, React web portals for service partners and administrators, .NET Core and Node.js services, and an Azure IoT Hub, Service Bus and Cosmos DB backbone. The architecture serves a large connected-device fleet with sub-two-second latency. Automated UI testing in CI/CD cut manual testing costs by 40%, and technicians complete installation and commissioning 30% faster.
Technical questions
How do you keep heating-app latency low across a large device fleet?
In this system FiftyFive Technologies routes device traffic through Azure IoT Hub, uses Azure Service Bus to decouple message processing from the apps, and serves frequently read device state from Redis, with telemetry in Cosmos DB. The architecture supports a large connected fleet with latency under two seconds.
How do you test IoT apps when physical devices are scarce?
FiftyFive automated UI testing with Java, Appium and TestNG and ran it in GitHub Actions CI/CD, so most regression testing does not depend on access to physical heating units. For this manufacturer, automation cut manual testing costs by 40%.