Hush
AI-Powered 2MB macOS App, Hidden from Screen Sharing
Start new thread
Kaizo Konpaku

Hush - AI-Powered 2MB macOS App, Hidden from Screen Sharing

by•
Hush is a 2MB AI-powered macOS app for real-time audio transcription, screenshot analysis, and text processing. Hidden from screen sharing, it runs fast with <2MB file and <50MB RAM usage and integrates seamlessly using Google Gemini and SwiftUI.

Add a comment

Replies

Best
Kaizo Konpaku
Maker
đź“Ś
Hey everyone! I’m Kon, I created Hush because I wanted to make a truly open-source version of Cluely, a privacy-focused macOS app that brings AI-powered transcription, screenshot analysis, and text processing right to your desktop. Unlike many closed-source tools, Hush is 100% open source, so you can see exactly how it works, contribute to it, or customize it for your needs. It’s built to be fast, lightweight (under 50MB RAM and under 2MB space), and tightly integrated with macOS using SwiftUI and Google Gemini AI. Whether you’re a developer, student, content creator, or just someone who loves productivity tools, Hush is designed to give you powerful AI features without sacrificing control or transparency. NOTE: On “Hidden” Interview Assistance Tools Apps like Cluely are not completely hidden during screen sharing. Why is that? From what I’ve discovered, Apple does not currently offer any reliable or officially supported APIs that allow applications to remain fully hidden during screen sharing sessions. This means that even tools marketed as discreet or undetectable (like Cluely) can still be exposed with minimal effort—such as using features like Presenter Overlay on macOS. Cluely and similar apps often avoid clearly communicating this limitation. They don’t actively show or warn users that their presence can be easily revealed with just a single click by an interviewer. That’s one of the key reasons I built Hush, an open source alternative focused on transparency and user control, without paywalls or misleading promises. You shouldn’t have to spend money on tools that claim invisibility while still being detectable with basic OS features. Learn more: https://github.com/KaizoKonpaku/...