The WinVoice Engine

Explore the autonomous neural routing architecture. A single local model dynamically parses complex voice commands and orchestrates up to 15 specialized agents.

🎙️
Electron Frontend
Captures global hotkeys and voice streams. Real-time VAD detects when you speak, securely piping audio via IPC.
Local AI Router & Memory
FastAPI intercepts the transcript. A pre-warmed Ollama model strictly parses your intent, cross-referencing your Vector Database to perfectly route your task.
Workspace & Productivity
📝
Docs Agent
Creates, reads, and dynamically types into Google Docs based on your instructions.
📊
Sheets Agent
Updates rows, queries columns, and generates spreadsheets natively.
📧
Gmail Agent
Drafts emails, summarizes unread threads, and manages your inbox contextually.
📅
Calendar Agent
Schedules meetings, checks availability, and resolves calendar conflicts.
Tasks Agent
Manages Google Tasks, organizes sub-tasks, and tracks your daily goals.
System & Local Control
💻
OS Agent
Executes raw system commands, manages window states, and adjusts system hardware natively.
⚙️
System Agent
Handles core application configurations, user state, and fallback queries.
🧠
Memory Agent
Specialized in explicitly memorizing, retrieving, and updating user preferences into pgvector.
⏱️
Timer Agent
Sets system alarms, manages deep-work timers, and handles time-based macros.
Web & Communication
💬
WhatsApp Agent
Spawns a headless browser to silently send and read WhatsApp messages.
🎵
Spotify Agent
Controls playback, queues songs, and analyzes your musical preferences.
🌤️
Weather Agent
Fetches real-time hyper-local weather conditions and multi-day forecasts.
👥
Contacts Agent
Resolves names to phone numbers or emails across all your interconnected apps.
🌐
General Agent
Your ultimate fallback. Performs real-time web searches to answer abstract or complex queries.