Inspiration

We are a team that has previously worked with fitness apps, and we wanted to take this time to develop our ideas and knowledge base further, using different aspects of AI.

What it does

Our app prompts the user to find out what aspects of a workout regimen would be optimal for them. The app then transforms the given data in addition to pre-defined parameters, into a LLM prompt. The LLM will then return a full workout plan to the app. The app will then unpack the LLM-response in the front end, creating a checklist and links to the specific exercises. This allows the user to specify what they do/don't like when it comes to lifting, as well as what their goals are, and the app will customize and create a workout plan just for them. The app also utilizes an AI Model to determine how correct a recorder exercise is. The user can record themselves performing either deadlifts, squats or shoulder press, and our App will analyze their lifting form, and determine how correct they are.

How we built it

This is built in Expo using React Native on Expo, that way is available on both IOS and Android. There is a Fast API Backend. We also utilized YouTube's API, and Google's Gemini.

Challenges we ran into

We had a hard time configuring the front end to build, this resulted in our development being extremely delayed to start the hackathon. We also had a lot of version issues between our 3 developers, which ended up delaying our coding by quite a lot.

Accomplishments that we're proud of

We are proud of the advanced AI tools that are applied to the form-checking portion of our app. While other apps simply judge simple body points and how they move, our app tracks angles and specific metrics that go into performing good form.

What we learned

We learned a lot about Prompt Engineering, as well as Machine Learning, and for a lot of us, a good amount about front-end development.

What's next for WorkoutGenie

We hope to further develop the app we created over this weekend, and turn it into a fully operational app, for users to create workout plans for themselves using the power of AI. We also plan to develop the form-correction video software, to ensure users can properly learn with this app.

Built With

Share this project:

Updates