Voice assistant
I added push-to-talk capture, streaming transcription and synthesized speech.
A voice and text assistant that can inspect and operate apps.

I built Peaky as a macOS menu-bar assistant for voice and text tasks. It captures speech, reads screen context and connects to an AI agent that can operate apps, with provider access through a Cloudflare Worker or local chat shim.
I built a native app for macOS 14.2 and later with Swift, SwiftUI, AppKit, AVFoundation and Accessibility APIs.
I added push-to-talk voice capture, streaming transcription, multi-monitor screen capture and accessibility information.
I built a floating panel and a transparent full-screen overlay for the assistant.
I connected AI and transcription through a Cloudflare Worker proxy or local chat shim, using user-supplied API keys.
I added push-to-talk capture, streaming transcription and synthesized speech.
I connected multi-monitor capture with accessibility information.
I built a floating panel and a transparent full-screen overlay.
I connected AI and transcription through a Cloudflare Worker proxy or local chat shim.
