Kinetic

Inspiration

Today, improving your workout technique often requires expensive personal trainers or guesswork in front of a mirror. We wanted to make high-quality, real-time feedback accessible to anyone—whether you're lifting at home, in the gym, or just getting started.

What excited us most is how broad the impact can be. Beyond performance gains, proper form is critical for injury prevention, and our app makes that guidance immediate and continuous. We also saw an opportunity to make fitness more inclusive: with real-time audio feedback, users—including those who are visually impaired—can receive actionable coaching through speakers or Bluetooth headphones without needing to look at a screen.

What it does

Our app turns your phone into a real-time AI fitness coach.

Users film themselves performing exercises in our iOS app, where we leverage Apple’s native computer vision capabilities to track body movement and posture. As the workout happens, the app analyzes form and delivers instant feedback and corrections, helping users adjust on the fly instead of after the fact.

How we built it

We approached development like building a real product, not just a prototype.

We started with a high-level system architecture to define how each component—from video capture to motion analysis—would interact. From there, we built a functional skeleton and iterated rapidly, layering in features and refinements.

To move fast without sacrificing structure, we used AI-assisted development tools like Cursor, which allowed us to accelerate implementation while keeping our codebase organized and adaptable.

Challenges we ran into

We took on a few ambitious challenges.

First, we built the app in Swift for iOS—an entirely new environment for all of us. That meant a steep learning curve and slower testing cycles, which forced us to be very intentional about where we spent time.

Second, we had to strike the right balance between speed, cost, and accuracy in our motion analysis. Instead of relying on slower, more expensive approaches like calling large language models for every analysis, we combined Apple Vision’s real-time video processing with our own custom logic for interpreting movement. This allowed us to deliver fast, responsive feedback, which is essential for a product focused on preventing injury during live workouts.

Accomplishments that we're proud of

We’re proud that this isn’t just a concept—it’s a working, usable product.

Our custom model for interpreting Apple Vision data, combined with an intuitive and responsive UI, creates an experience that feels like having a coach in your pocket. The real-time feedback loop is smooth, actionable, and immediately valuable to the user.

What we learned

We learned how to go from idea to product in a completely new technical environment.

Along the way, we:

Built a full iOS app from scratch in Swift Worked with Apple Vision for real-time motion analysis Designed and implemented a cohesive, user-focused experience

Just as importantly, we learned how to balance ambition with execution under tight time constraints.

What's next for Kinetic

We’re just getting started.

Right now, our motion analysis supports a limited set of exercises. Next, we plan to expand our exercise library and improve our proprietary joint-point detection algorithm to handle a wider range of movements with greater precision.

Longer term, we want to evolve Kinetic into a fully personalized coaching system—one that not only corrects form in real time, but adapts to each user’s goals, progress, and injury risk over time.

Built With

• apple-vision
• cursor
• fastapi
• openai
• supabase
• swift/ios
• xcode