a native macOS menu bar app for push-to-talk voice dictation. Hold a hotkey, speak, and your words stream into any app — email, Slack, VS Code, Notes, anywhere. Record your stand-ups and meetings without inviting any weird agent to your call, get meetings and reminders of who said what.
Built in Swift/SwiftUI with AVAudioEngine, Yakki runs Parakeet and Whisper entirely on-device for sub-200ms latency (0.1 Real-Time Factor). No audio leaves your Mac if you do not want to.
What makes it different: a screen OCR pipeline reads your current app context and adapts formatting automatically — properly formatted for code editors, professional tone for email, concise for Slack. Hot-swappable transcription engines let you choose speed vs. accuracy vs. language support.
Comments, support and feedback
About this launch
Yakki Ai by Wu Will be launched April 28th 2026.



