Case study · Oryn
Oryn: a voice-first AI companion you can actually talk to
A warm, voice-first AI companion that is ready whenever you want to talk, start in seconds with Google sign-in or as a guest.

The challenge
Make talking to an AI feel natural, warm and effortless.
Most AI assistants feel like typing into a search box. Oryn set out to be different: a companion people simply talk to, that feels calm and approachable, and that anyone can start using without friction.
Our approach
Voice first, friction last.
We designed the experience around conversation rather than text, built on Next.js with Supabase for authentication and realtime data. A friendly animated character gives the agent a face and a sense of presence, the interface stays minimal, and onboarding takes one tap: continue with Google, or try it instantly as a guest.
The solution
A voice-first conversational agent with a calm, character-led interface.
- Real-time, voice-first conversations powered by a large language model (LLM).
- Low-latency audio streaming in the browser using the Web Audio API (AudioWorklet) and WebSockets.
- Supabase Authentication with one-tap Google sign-in and anonymous guest sessions.
- Supabase Realtime for live session updates.
- An animated character that makes the agent feel present and approachable.
- A clean, responsive Next.js and React interface that works on mobile and desktop.
Technology
- Next.js
- React
- Supabase
- WWebSockets
- WWeb Audio API
- LLLM