CramPlan

• 

  Architecture

Inspiration

In today's fast-paced world, our generation often struggles to juggle multiple responsibilities. Balancing schoolwork with extracurricular activities can be challenging, and as a result, many students end up cramming their studies at the last minute. This is a common experience shared by school and university students alike. So, why not create a tool that can make the cramming process easier and more efficient?

What it does

CramPlan enables students to upload their course materials, which are then analyzed to identify key topics and subtopics. Based on this analysis, the tool generates a 10-question multiple-choice quiz. After the student completes the quiz, CramPlan evaluates their strengths and weaknesses. Using these insights, the AI agent creates a personalized learning roadmap tailored to the student’s specific needs.

How we built it

We used OpenAI’s Agent SDK to develop a multi-agent workflow integrated with FastAPI. For real-time voice chat functionality, we implemented WebRTC sockets in combination with LiveKit. The frontend of the application was built using Next.js.

Challenges we ran into

One of the key challenges we faced was establishing a smooth connection between the backend with both FastAPI and the vector database to ensure efficient data handling and retrieval. Additionally, integrating real-time voice functionality posed its complexities, particularly in achieving low-latency, stable communication.

Accomplishments that we're proud of

We successfully developed a fully functional MVP featuring a real-time voice agent capable of interacting with students and addressing their questions or doubts on the spot.

What we learned

On the technical side, we gained hands-on experience working with the latest version of OpenAI’s Agent SDK, including effectively utilizing VectorDB for knowledge storage and retrieval. We also explored implementing real-time voice chat using LiveKit and WebRTC, learning how to manage low-latency communication. Additionally, we developed skills in integrating these components seamlessly using FastAPI, ensuring smooth interaction between the backend and frontend.

From a soft skills perspective, we realized the importance of starting early, planning, and maintaining clear communication within the team. This approach helped us manage deadlines better and overcome unexpected challenges more efficiently.

What's next for CramPlan

We plan to make CramPlan multilingual to accommodate students from diverse ethnic backgrounds, ensuring accessibility for all. Additionally, we aim to integrate a mental health management agent that recommends timely breaks and strategies to help students avoid burnout. Finally, we intend to introduce visual-based learning features, offering visual representations of concepts to better support students who learn more effectively through visual aids.

Built With

• cursor
• fast-api
• livekit
• next-js
• openai-agent
• python
• vector-database