Show HN: Nailed – Stop biting your nails with on-device ML and your Mac's camera

2 points

2 months ago

I've been a nail-biter for most of my life. At some point I realised it happened the most when I'm at my computer, doing resarch, shipping code or thinking. By the time I notice, it's already too late. I tried many things, bitter polish was helpful to an extend but didn't stop the habit entirely. Nothing stuck because the habit is unconscious.

So I built Nailed. It sits in your macOS menu bar, watches your camera feed using MediaPipe (running locally via WebAssembly), and detects when your hand moves toward your mouth. When it catches you, you get a red vignette animation and a gentle beep: just enough to snap you out of it.

The technical bit that might interest this crowd: all inference runs on-device. Camera frames are processed in memory and immediately discarded. There are no servers, no analytics, no data collection of any kind. The ML models are bundled with the app, so it works fully offline.

It requires macOS 12+ and an Apple Silicon Mac (M1 or later). It's a one-time $4.99 purchase on the Mac App Store, no subs or IAP.

I'd love feedback on the detection accuracy, false positive rate, and whether the alerting is too aggressive or not aggressive enough. Happy to answer questions about the ML pipeline or the architecture.

App Store: https://apps.apple.com/app/nailed-stop-biting-nails/id676173...

Site: https://nailedapp.io

2 comments