A macOS Desktop Overlay app for voice-powered AI assistance using on-device speech recognition and xAI Grok.
- On-device speech-to-text with WhisperKit
- AI responses via xAI Grok API
- Global keyboard shortcuts
- Spotlight-style floating overlay
- macOS 14.0 or later
- xAI API key from https://console.x.ai
- Clone and open in Xcode
- Build and run
- Enter your API key in Settings
Shortcuts can be customized in Settings.
User speaks
|
v
AVAudioEngine (captures audio)
|
v
AudioProcessor (converts to 16kHz mono)
|
v
WhisperKit (transcribes to text)
|
v
GrokService (sends to xAI API)
|
v
Streaming response displayed in UI
The xAI API key can be configured in two ways:
- In-App Settings: Open the overlay, click the gear icon, and enter your API key.
- Environment Variable: Set
XAI_API_KEYin your environment (useful for development).
The app uses the grok-4-1-fast-reasoning model by default. To change this, modify the model constant in GrokService.swift.
- WhisperKit by Argmax for on-device speech recognition
- KeyboardShortcuts by Sindre Sorhus for global hotkey management
- xAI for the Grok API