Why I Chose Swift Over Cross-Platform for Movedoro

When I started building Movedoro, I had to make a choice. Go native with Swift for macOS only, or use something cross-platform that works on Windows too.

I went native. Here's why.

Precision in Body Detection

I needed accurate body tracking. Really accurate.

Swift gives me direct access to Apple's Vision framework. It's built into macOS. It's optimized for the hardware. And it's precise enough to detect subtle movements.

Could I have used a cross-platform solution? Sure. But the body detection wouldn't be as good. And for an app that forces you to exercise based on what it sees, precision matters.

Right now, the tracking is pretty naive. I just count squats. That's it.

But I'm expecting real-world scenarios. People with half their body out of frame. Weird lighting. Different camera angles. I need the flexibility to handle all that.

Having native access means I can explore more complex detection later. Track form. Detect different exercises. Make it smarter about what counts as actual movement.

Developer workspace with multiple monitors

Local AI on Mac Hardware

The second reason is about where this could go.

Mac machines can run local LLM models. Not cloud-based. Not sending your data somewhere. Just local, on-device AI.

I'm thinking about adding a personal trainer you can talk to. It would run entirely offline. You ask it questions between sets. It adjusts your routine based on what you're doing.

This only works if I have full access to the hardware. Native Swift gives me that.

Build for Power First

My approach is simple. Take advantage of the best hardware available. Build something really powerful. See if it works.

If Movedoro takes off on Mac, I can port it to Windows later. The core idea stays the same. The implementation would be different, but that's fine.

Going cross-platform from day one means compromising on both platforms. Going native means building the best possible version for one platform first.

That's the bet I'm making.

I want body detection that's precise enough to handle messy real-world scenarios. I want the option to add AI features that run locally. I want to use the full power of the hardware.

That's why Movedoro is Swift-native on macOS. If you're on Windows, I hear you. Maybe later. For now, I'm focused on making the Mac version as good as it can be.