1. The Challenge
Most AI interfaces are restricted to basic conversational chat boxes. Omni explores spatial AI collaboration: a dynamic infinite canvas where agents generate code, diagrams, and voice responses in real time.
2. Design System & Wireframes
An infinite spatial grid with smooth panning, node zooming, and glassmorphic inspector drawers. Color palettes adapt dynamically based on the active agent persona.
3. Technical Architecture & Code Highlights
- **Google Gemini 2.0 Streaming**: Low-latency token and audio chunk streaming via Server-Sent Events.
- **WebRTC Voice Synthesis**: Bidirectional voice conversation with real-time audio visualization canvas.
- **Supabase SSR Authentication**: Protected workspace instances with Row Level Security.
4. Measurable Results & Outcomes
- **Under 250ms** first-token generation latency.
- **5-Star** product rating from early enterprise beta testers.
- **Zero** client-side memory leaks during 8-hour continuous canvas sessions.
