Inspiration
We wanted to make an app that helps the user focus on studying. The goal was something that locks the browser until the user finishes their work. It become complicated when deciding how to format it and what exactly would be blocked. Therefore, we chose a browser fullscreen that tracks when the user visits other tabs, how long they were gone, makes sure their not on their phone or away from their desktop, and provided a study buddy to help them along the way.
What it does
Deep Dive locks your browser into an underwater fullscreen session until your homework is done. Gemini verifies your submission before it releases, Coral (a jellyfish voice buddy) asks Socratic questions as you work, and an optional webcam proctor flags phones and off-screen distractions in real time. The app also tracks how many times you visit other tabs, how long you were gone, and how long it took to complete the assingment.
How we built it
We first started by developing our landing page and other pages with html and used react+tailwind css to complete the front end. We then use vite to build and host and used supabase to store the computed data like distraction or tab visits. Gemini does alot of the work for outputing static data like summarys, analyzing the user input and assingment, using the voice feature included with the gemini api, and analyzing the camera details to detect no-faces or a phone.
Challenges we ran into
The main chalenge we experienced was being able to efficiently use gemini without getting information all over the place. We needed summarys, correct answers, data gathered, and many other things from gemini. For example, the user can see their dashboard if gemini decideds if their assingment is good enough and in the correct format. This was very dificult as we had to establish rules on gemini. Setting up the voice was difficult as well.
Accomplishments that we're proud of
Me and my friend jared who wasnt able to stay the entire time put alot of effort on day 2. This was his first competition and my second one, and we learned alot about architecture, systems, and collaborating.
What we learned
We learned alot about the gemini api and much value it brings. It gives access to many things llke nanobanana, voice, and the api which were all used in this program. The assets like background or fishes were generated using it.
What's next for Deep Dive
I am currently working on hosting it, which is partly why I am rushing my submission sorry lollll. Hosting is defintly the goal.

Log in or sign up for Devpost to join the conversation.