A major learning was that visibility plays a significant role in effective interaction. Users responded positively to large, high-contrast alerts (e.g., "RIGHT HEAVY!"), but struggled when labels were ambiguous or indicators weren't clearly "live". This drove us to adopt conspicuous labeling and natural information arrangement.
Users expected dynamic elements. They tried to tap the fixed Posture Score and assumed the static spine illustration would move. This inspired us to design real-time interactivity—making elements hoverable and tappable to increase discoverability and transparency.
Navigation consistency emerged as a serious issue, with users reporting missing bars or hidden screens. We redesigned the app to have a common bottom navigation bar and permanent shortcuts across all core screens, minimizing cognitive load and preventing task breakdowns.
Users expect AI companions to provide efficiency, not just conversation. Participants wanted voice input and preset prompts. We aligned with these expectations by adding quick-access buttons like "Why did I slouch?" and "Explain my score" to make the AI more functional.
Safety-critical metrics like battery status and slouching statistics were initially buried. We restructured the dashboard to prioritize these metrics, using vivid icons and salient indicators to ensure long-term engagement and system reliability.