DriveLearn

Inspiration

On the road you can't read or access your phone without endangering other drivers. A friend and I were calling and we were both in a class together, and we were quizzing each other for our next midterm.

I realized that this was an extremely productive way to use commute time. Using spaced repetition helped me recall exam concepts better than before.

What it does

It teaches you concepts on the go with live TTS, and you can upload any of your notes to the app. The AI will ask you questions and based on your answers, adapt in real-time and ask its next questions based on your mastery of previous material.

How we built it

We utilized various AI agent softwares, such as lovable and claude code for the development. technical design was inspired by similar voice agent softwares like Vapi

Challenges we ran into

Supabase authentication proved difficult. Same with hosting an edge function and deploying it properly.

Accomplishments that we're proud of

We got the landing page to work, and successfully setup claude code and most of the codebase.

What we learned

We learned the limits of AI programming agents like claude code and how you should always do due diligence and question its approach, because it often goes down rabbit holes,

What's next for DriveLearn

A full working MVP and a launch to UCSC undergrads. Eventual expansion to other UCs and undergrad schools.

Built With

• claude
• node.js
• openai
• typescript