Consumer × Streaming

Swipe Watch

A swipe-based discovery experiment for Disney+ and Hulu that turns taste signals into a faster, more active recommendation loop.

Overview

Swipe Watch explores a simple question: what happens when streaming discovery feels active instead of passive? Rather than browsing endless shelves, the product uses a swipe interaction to let users react quickly, train their recommendations, and shape what the system learns.

It is a lightweight concept, but it is aimed at a real product problem: helping people express taste faster than a conventional streaming interface usually allows.

What the product does

  • Presents streaming choices through a rapid swipe-based interface.
  • Turns user reactions into recommendation signals.
  • Adds coins and watchlist building as lightweight feedback loops.
  • Tests whether a faster interaction model can make discovery feel less static.

Build notes

Swipe Watch was built over a weekend in vanilla JavaScript. The speed of the prototype matters: it is a deliberately lightweight way to test the interaction idea without the overhead of a larger stack.

Why it matters

Recommendation systems are often evaluated on model quality alone, but the interface that gathers preference signals matters too. Swipe Watch focuses on that front-end layer: the mechanics of how people tell a product what they want to watch next.