Cross-Platform Voice & Text AI Desktop Assistant
Wingman is an intelligent, privacy-conscious desktop companion engineered for hands-free workflow automation and open-domain conversational AI. Powered locally by Meta's BlenderBot-1B-Distill model, Wingman blends system-level desktop controls, voice synthesis, and real-time information retrieval into a unified assistant that runs unobtrusively in the background.
Pre-built binary packages and releases for Windows, Linux, and macOS are hosted on SourceForge:
👉 Download Wingman on SourceForge
- Core Capabilities
- System Architecture & Intelligence
- Supported Platforms
- Quickstart & Usage
- Command Reference
- Authors & Acknowledgments
Driven by the facebook/blenderbot-1B-distill transformer model, Wingman delivers empathetic, human-like dialog with contextual memory for natural conversational flow.
- Media Automation: Query and stream audio tracks directly on YouTube and popular streaming providers via voice commands.
- Web Navigation: Launch URLs, open system browsers, and navigate to target resources instantly.
- Query Resolution: Run automated Google searches and extract top result summaries without lifting a finger.
- Knowledge Retrieval: Fetch structured Wikipedia summaries and factual data directly into the assistant UI.
- Screen Capture: Capture high-resolution desktop screenshots instantly with simple hotkey triggers or voice inputs.
┌─────────────────────────────────────────────────────────┐
│ User Input │
│ (Voice Command / Speech-to-Text) │
└───────────────────────────┬─────────────────────────────┘
│
▼
┌─────────────────────────────────────────────────────────┐
│ Intent Router / Classifier │
└──────────────┬───────────────────────────┬──────────────┘
│ │
[Automation / System Tasks] [Open-Domain Chat]
│ │
▼ ▼
┌───────────────────────────────┐ ┌───────────────────────┐
│ Desktop Automation Suite │ │ BlenderBot-1B-Distill │
│ • Media Playback │ │ • Contextual Memory │
│ • Google & Wikipedia Queries │ │ • Conversational NLP │
│ • Screen Capture Utility │ │ • Natural TTS Voice │
└───────────────────────────────┘ └───────────────────────┘
| Intent | Sample Command | Action |
|---|---|---|
| Music Playback | "Play Starboy on YouTube" | Automates browser search and starts audio/video stream |
| Web Navigation | "Open GitHub" | Spawns system browser directly to the requested target |
| Quick Search | "Google latest space missions" | Executes Google search query and surfaces top results |
| Encyclopedia | "Wikipedia Quantum Computing" | Fetches and reads a concise structured entity summary |
| Screen Utility | "Take a screenshot" | Captures current desktop display and saves to local storage |
| Free Chat | "How's your day going?" | Routes conversation to BlenderBot NLP for dynamic dialogue |
- Vignesh S. — @Vignesh-72
- Built on top of Meta AI's open-source BlenderBot-1B-Distill model.