Iris-chan – Your Persistent Voice-First AI That Sees, Thinks & Acts

Iris-chan is a persistent, voice-first AI orchestrator that sees what you see, commands agent teams, and interacts the way humans naturally do — no typing, no switching, no recompiling.

✨ What It Does:
• Delegation: Commands specialized sub-agents (e.g., clawdbot) for complex tasks
• Continuity: Maintains memory, state & preferences across sessions
• Monitoring: Live task monitoring, cost tracking & decision visibility
• Autonomy: Intent-driven interaction, auto 2FA detection & autonomous web exploration

⚙️ How It Works:
Built on a persistent-state architecture (Convex) with a "hot-reload" capability — update logic without losing context. Iris-chan acts as the conductor while Focus compiles specs into tasks. The continuous-state model eliminates context loss on prompt changes.

🛠️ Built With: Gemini Live API, UI-TARS, Perplexity, Convex, TypeScript / Node.js

🏆 Built at the Gemini Live Agent Challenge

🔗 Project: https://devpost.com/software/i-r-i-s-bygm6c

#IrisChan #AI #VoiceAI #AgenticAI #Gemini #MultiAgent #HackathonProject #AIAssistant

Видео Iris-chan – Your Persistent Voice-First AI That Sees, Thinks & Acts канала Livio GamAI