FeatherNet

An AI bird-watching platform built top to bottom: smart feeders with cameras, live video streaming, real-time species identification, a community for bird watchers, and the storefront that sells the hardware.

What it is

FeatherNet is a complete backyard-birding platform. Smart feeders stream live video to the cloud, an AI vision model identifies each species as it arrives, and members watch their own feeder cams and other people’s. A social layer adds follows, comments, leaderboards, rare-sighting alerts and shareable clips. The storefront sells the feeders and accessories that power the whole loop.

What we built

• Feeder integration. Feeders wake on motion, push compressed video into the streaming pipeline, and report telemetry back to the platform.
• Live video streaming. A low-latency RTMP/HLS pipeline so members can watch their feeder, and follow other backyard cams, in near real time on mobile and web.
• AI bird identification. A vision model runs against the live frames, tags species with confidence scores, and saves highlight clips automatically.
• Community. Profiles, follows, comment threads on sightings, weekly species leaderboards, rare-bird alerts and clip sharing.
• Storefront. The front door at feathernet.store, with an animated forest scene, light and dark themes, and a catalog tied directly to the platform’s hardware and subscriptions.
• Design system. One set of design tokens covers every surface, from the cam viewer to checkout.

The hard parts

• Real-time video over home internet connections, with encoding, buffering and AI inference kept within a per-feeder budget that makes sense.
• Identification that holds up across weather, lighting and partial views, with an interface that shows confidence honestly so a “possible cardinal” is still useful.
• A social feed organized by interests, location and rarity, so it stays useful as it grows.
• An animated scene that runs smoothly on phones alongside a real catalog and checkout.

Outcomes

• One product experience, with the store, the app and the community working together.
• Live streaming and identification at price points that suit a consumer audience.
• A daily reason for feeder owners to come back.
• Storefront, app and admin built from one codebase and one design language, so new features ship across all of them at once.

Where Think New helped

Think New designed and built FeatherNet end to end: the streaming pipeline, the identification model integration, the community features, the storefront and the design system, from first sketch to live platform.