Inspiration
We were inspired to build this project because, as students who struggle to stay still and maintain focus, we found the traditional format of long, static lecture videos nearly impossible to navigate. We wanted to solve the 'focus fatigue' we personally experience by creating a tool specifically to support users with ADHD, transforming passive, overwhelming content into an enjoyable, gamified journey that replaces the struggle to concentrate with interactive, high-stimulation 'Quests.'
What it does
Our platform transforms the exhausting experience of sitting through long lectures into a streamlined, interactive adventure. By stripping away the "filler" and "dumbing down" the content to the essential "need-to-know" information, we eliminate the cognitive overwhelm that often leads to distraction.
Instead of passive watching, the core info is delivered through a high-engagement interactive experience. Users test their knowledge through targeted questions that turn study material into a series of "Quests." To keep the momentum going, we’ve integrated a reward system: for every lecture URL posted and every question answered correctly, users earn XP (Experience Points). This turns a daunting academic chore into a rewarding game, providing the immediate feedback and clear milestones necessary to keep learners focused and motivated.
How we built it
We used the Gemini API as our central intelligence. It’s responsible for the heavy lifting, analyzing long form audio from lectures and "distilling" it. We chose Gemini specifically for its massive context window, which allows it to process a full lecture in one go and output structured JSON data that we use to generate our interactive questions. Our backend is built with FastAPI, served by Uvicorn. This setup was crucial because it supports asynchronous processing. This means while the AI is busy "thinking" and summarizing a long video, the rest of our app stays responsive and doesn't freeze for the user. To get the content from the web into our AI, we utilized yt-dlp. This tool allows our backend to take a simple YouTube URL and extract the audio stream directly, which is then fed into the Gemini API for analysis. The frontend is a React application designed with a "gamification-first" mindset. We used React’s component-based structure to build the interactive "Quest" interface and the real-time XP (Experience Points) reward system. To ensure that every bit of progress is tracked, we used MongoDB to store user profiles, earned XP, and the history of completed quests.
Challenges we ran into
Our greatest technical hurdle was the steep learning curve of Generative AI, specifically mastering prompt engineering to ensure the Gemini API returned structured, 'bite-sized' data without breaking our frontend. We also faced significant workflow challenges, from losing hours to case-sensitivity bugs to navigating complex GitHub merge conflicts while trying to keep track of our pushes and pulls. These obstacles forced us to move toward more organized communication, ensuring the frontend and backend teams stayed perfectly in sync. Finally, we struggled to balance our styling, aiming for a UI that felt high-stimulation and rewarding without becoming a distraction for our target users.
Accomplishments that we're proud of
We are incredibly proud of successfully completing a fully functional MVP within the tight hackathon deadline, moving from a rough concept to a polished, end-to-end application. A major highlight was seeing the Gemini API deliver highly accurate summaries, it consistently captured the core essence of long lectures and turned them into relevant, high-quality study material. Finally, we’re proud of how we balanced a complex tech stack to create a smooth reward system that effectively turns a video into an engaging, gamified experience.
What we learned
This weekend taught us that a great idea can quickly become overwhelming unless it is organized well through clear communication and structured planning. We gained a deep, hands-on understanding of what a tech stack really is, specifically learning how to build and route a FastAPI backend to handle our data logic. Most importantly, we learned how to bridge the gap between our backend and a React frontend, proving we can build a cohesive system that solves real-world problems.
What's next for Bookiverse
Our primary goal is to move away from static 'correct or incorrect' answers and toward a more dynamic AI interaction. We want the Gemini API to act as a Socratic tutor, if a user gets a question wrong, the AI won’t just give them the answer, but will instead ask probing follow-up questions to guide them toward the right conclusion. This creates a continuous feedback loop that keeps the user’s mind active and engaged in a conversation rather than a repetitive test. By adding this layer of interaction, along with personalized player settings and fun fact pop-ups, we can ensure that every study session feels like an evolving adventure tailored to the user's specific pace.
Log in or sign up for Devpost to join the conversation.