Back to work
AI/ML
Multi-Model AI Discussion Platform
Stadia product
A desktop application that lets a human moderator orchestrate structured discussions among multiple AI models — cloud and local — with real-time streaming and workspace-based project management.
The Challenge
Getting value from multiple AI models usually means juggling separate tools; there was no single place to run a moderated, multi-model conversation with full history and control.
What We Built
- A Tauri (Rust) desktop shell with a React + TypeScript frontend and a FastAPI (Python) backend.
- Moderator-controlled turn orchestration across configurable models (e.g. Claude, GPT, local Ollama), with real-time WebSocket streaming and token tracking.
- Workspace-first organization (projects, threads, AI personas), local-first SQLite storage, and OS-keychain credential storage.
Highlights
- Moderated multi-model AI discussions in one app
- Cloud + local (Ollama) model orchestration
- Real-time WebSocket streaming with token tracking
- Local-first, secure credential storage