Decision 1: head movement for navigation, lip pucker for confirmation. The alternative was blink- or gaze-based selection, the default of most hands-free experiments. Rejected, because blinking is involuntary and gaze-dwell is slow and exhausting — both produce false positives that erode trust in the interface within a minute. A pucker is deliberate, low-effort, and nearly impossible to do by accident.

Decision 2: build the real interaction instead of simulating it. The alternative was a click-through prototype that pretends to track a face. Rejected, because the entire claim lives in the latency and reliability of the real thing. Built working in Figma Make in one day — which required understanding, at least at a high level, how the code underneath behaves when it misbehaves.

Decision 3: frame it as an interaction layer, not an app. Pucker isn't a product with a feature list; it's a way of confirming intent that could sit on top of any app or platform. That framing points it at its most serious use case: accessibility — interfaces for people whose hands aren't available, temporarily or permanently.