FormFit TV

Inspiration

It started with a simple frustration. I kept seeing people (including myself) put on a workout video on the TV, follow along for a few minutes, then either stop or keep going with shaky form because there was no way to know if we were doing it right. The TV is the biggest screen in the house, yet most fitness experiences on it still feel like watching YouTube. I wanted something that actually coached you while you moved, without needing extra gadgets or a phone in your hand the whole time.

The Pain

Working out at home on the TV usually means playing a video and hoping you’re doing it right. You can’t see your form, you lose count of reps, and there’s no one to tell you when you’re rushing or letting your knees cave in. After a few sessions most people either quit or keep going with bad form and risk getting hurt. The big screen that should make exercise easier ends up making it passive and lonely.

What it does

FormFit TV turns that same TV into an active coach. You pick an exercise from large, focusable cards designed for the living room. A clear timer starts, you tap each time you finish a rep, and the app keeps the count. When you want feedback you tap “Ask AI Coach.” It gives you one short, practical tip and speaks it out loud so you don’t have to read while you’re moving. The experience is built for the 10-foot UI: big text, high contrast, simple navigation with the remote or taps, and voice so your eyes can stay on your movement.

How I built it

I used .NET MAUI and C# so the same codebase runs on Fire OS and Android TV. The interface follows TV patterns—large targets, strong focus states, and generous spacing. For the coaching, I integrated Amazon Bedrock (Claude) through the Converse API. When the model is available, it returns a real tip; if the call can’t complete, the app falls back to proven coaching lines so the session never breaks. Text-to-speech handles the voice output.

Challenges I ran into

Getting a reliable TV layout and focus behavior took iteration. Emulator performance and audio limitations made testing slower than expected. The biggest external hurdle was Amazon Bedrock model access on a new account—Anthropic enablement took longer than the usual window—so i designed the service to degrade gracefully while still keeping the real integration in the code. That forced me to think carefully about resilience, which made the final app stronger.

Accomplishments that I'm proud of

We got a clean 10-foot UI working with large focusable cards, a live timer, and rep tracking that feels natural from the couch. Adding voice so the coach speaks tips without forcing the user to read mid-rep was especially satisfying. We also kept the Amazon Bedrock integration in the code while making sure the app never breaks if the model isn’t available yet — that resilience mattered. Shipping something that actually runs on Fire OS / Android TV and solves a real “I quit home workouts” problem feels like a solid result for the time we had.

What I learnt

Building for the living-room distance is different from phone UI. Voice + short text works better than long instructions when someone is mid-rep. And planning for cloud services that aren’t instantly available taught us to separate the “happy path” from a solid fallback so the user experience stays intact. FormFit TV doesn’t replace a human trainer, but it removes the two biggest reasons people abandon home workouts: not knowing if their form is okay, and having no feedback or accountability on the big screen.

What's next for FormFit TV

Phone companion app that uses the camera for real pose estimation and sends form data to the TV. More exercises and short guided programs (beginner, strength, mobility). Family profiles so different people can track their own sessions. Alexa voice commands (“Alexa, next exercise”, “how’s my form?”). Progress history and simple streaks so the accountability sticks after the first week. Deeper Bedrock agent flow that remembers the user’s common form issues across sessions.

These keep the same core idea — making the TV an active coach — while adding the computer-vision and multi-device pieces that would make it even more useful.

Built With

• .net-maui
• amazon-bedrock
• c#
• fireos