Phyn

by Belkin

IoT ArchitectureReact NativeMachine Learning IntegrationServerless Infrastructure

Scroll

Overview & Architecture

Belkin brought us in in 2016, before there was a product. Phyn monitors a home's plumbing 240 times per second and flags anomalies, and the bet was that the pressure signature alone would be enough to do it. We built the MVP that proved the bet: the device pairing, the sensor data coming off the hardware, the service behind it, and the app someone actually used to see their own water. That is the version that showed the idea held up, and Phyn went to market from there.

Pre-Market Prototype

Built the initial proof-of-concept that validated Phyn's core technology: Bluetooth device pairing, pressure data visualization, and leak detection alerts.

Production Architecture

Designed the MVP's architecture end to end: how the device talks to the service, how sensor data moves, where the ML models plug in, and the React Native app on top.

CES Launch

Shipped the consumer-ready product for its CES 2019 debut.

Scale & Team Transition

Scaled the platform for public consumer release while helping Belkin build their internal engineering team to own the product long-term.

“

ClearSummit has a rare talent of understanding what they're building and what user context it needs to function with. This fosters a collaborative environment where both designers and developers can push boundaries.

Director of Product Design

Technical Challenges

Consumer IoT at Scale

Getting hardware, ML, cloud infrastructure and a mobile app to behave as one product in someone's house was harder than any of them on its own.

Bluetooth Provisioning

Device pairing in consumer IoT is notoriously fragile. Network timeouts, firmware states, user error. Dozens of failure modes that all need graceful recovery. We built a provisioning flow that handles edge cases users never see, with clear guidance when they need to act.

Real-Time Sensor Integration

Phyn's hardware generates continuous pressure data that feeds ML models running in the cloud. The mobile app needed to reflect that intelligence instantly, showing users what Phyn learned about their home as it happened, not minutes later.

Serverless Event Architecture

Water emergencies don't wait for servers to spin up. The alert path was designed as event-driven functions rather than a running service, so an alert does not wait on a server to start.

Platform Design

Intelligence Made Tangible

The challenge with smart home products is making invisible data feel real. Phyn knows things about your plumbing that you don't, and the app's job is to surface that intelligence in ways that feel intuitive, not overwhelming.

Learning Mode UX

Phyn's ML models need time to learn a home's water signatures. We designed an onboarding experience that turns this waiting period into engagement, showing users what Phyn is discovering about their plumbing in real-time.

Alert Hierarchy

Not all water events are emergencies. We built a notification system that distinguishes between "your sprinklers ran longer than usual" and "shut off your water main now", with appropriate urgency in the UI.

Cross-Platform Consistency

The same intelligence, the same interactions, the same reliability across iOS, Android, and web. React Native for mobile, React for web, shared component libraries, unified design system.

Offline Resilience

IoT apps can't assume connectivity. We implemented offline-first architecture with sync, so users can review their water data and configure settings even when their network is down.

From Prototype to Platform

Phyn debuted at CES 2019 and shipped to consumers across iOS, Android, and web, from the MVP we built. The product line drew industry recognition of its own: Phyn Plus took a Red Dot Award: Product Design in 2019, and the Phyn Smart Water Assistant was named a CES 2020 Innovation Award Honoree. Beyond the product, we helped Belkin build their internal capability. Our architecture documentation, codebase standards, and direct involvement in hiring enabled them to assemble an in-house team that could own the platform long-term. That's the goal: build something excellent, then make ourselves unnecessary.

Build Your IoT Platform

If you are building something like this, tell us what you are trying to do.