Jym.ai

The exercises generated by Google's Gemini
The calendar of the user's completed or incomplete days of the month
The main page where a user can input a prompt
The checklist that the user can knock off completed exercises

Inspiration

*One of our group members wanted to make a tangible difference in the world with our app. We settled on the idea of trying to improve the gym experience. From there, we thought about maybe making going to the gym and exercising more accessible. Some of our group members could relate to feeling uncomfortable at the gym because we might not know what to do. So we made this app to help. *

What it does

It uses Google's Gemini AI to generate an exercise plan based off the users' desired workout. From there, Gemini will provide a list of exercises it found appropriate. The user can then ask Gemini what each exercise is and how to do them step by step. An appropriate image is then fetched from Pexels to show what the exercise would look like. A checklist is also provided to tick off completed exercises. A calendar section allows the user to track their progress in the month with their daily generated exercise plans.

How we built it

** We used Flutter to make this an app on the phone. We used Google's Gemini API to get some of the generated exercises and responses. We used Pexels to find fitting images to provide for each exercise.**

Challenges we ran into

We tried implementing a database with our project, namely MongoDB Atlas. However, we ran into some difficulties actually implementing it. Because we were using Flutter and using Dart to write our programs, we had trouble connecting to our databases within VS Code. We spent a lot of time trying to make it work, but we inevitably just cut our losses.

Accomplishments that we're proud of

*Using the API for Google's Gemini was very cool. Toying around with the app and pushing the creativity of its responses was fun. Also most of us in the group do not have too much experience developing under such a intense time constraint. So having a respectable prototype done in time to present feels rewarding. *

What we learned

*Know when to cut your losses. Learning that some newer technologies are not as easily compatible a popular technologies. *

What's next for Jym.ai

*Instead of providing one image for one exercise, you would have more images per. Maybe one image per step to really walk them through it. Probably including a database would be beneficial if we had the time to. *