KABOOT!

• 

• 

  Kaboot!'s RAG pipeline

Inspiration

School and other commitments can get overwhelming, so here’s our take on gameifying studying: an AI powered version of a beloved game that encourages friendly competition.

What it does

You input study material from school -> Study material goes in a vector database -> LLM uses RAG to create tailored questions based on past academic performance.

How we built it

First, users upload files such as PDFs or Word documents through the React-Vite web app. These materials are then converted into markdown format, making the text easier to process. The markdown content is passed into an embedding model, which transforms the text into numerical vector representations (embeddings). These embeddings are stored in ChromaDB, a vector database that allows efficient semantic search and retrieval. When a quiz is requested, the system uses these stored embeddings to generate query embeddings and retrieves relevant content from ChromaDB. This retrieved information is then passed to the OpenAI language model (LLM), which creates quiz questions based on the user’s study material. The generated quizzes and their scores are finally stored in MongoDB, ensuring users can track progress and performance through the web interface.

Challenges we ran into

Connecting the frontend and backend, especially with the wifi issues was painful.

Accomplishments that we're proud of

We learned 2 new frameworks in 36 hours!

What we learned

We learned how to use React and Vite, we also learned how to use ChromaDB and RAG.

What's next for Kaboot!

We would love to add a social aspect to Kaboot! Unlike modern AI study tools that just generate notes and flashcard, users will be able to host games with their friends to battle it out on Kaboot! After the game, users could share which topics they found difficult and help each other out!

Built With

• css
• javascript
• node.js
• openai
• python
• react
• vite