MetaNotes AI

Turn raw study notes into one unified explanation and a short quiz using an Agent-of-Agents pipeline.

Quick start

1. Backend (Python)

cd backend
python -m venv venv
venv\Scripts\activate   # Windows
# source venv/bin/activate  # macOS/Linux
pip install -r requirements.txt

Create backend/.env with your OpenAI API key:

OPENAI_API_KEY=sk-your-key

Run the API:

cd backend
uvicorn main:app --reload --host 0.0.0.0 --port 8000

2. Frontend (React)

cd frontend
npm install
npm run dev

Open http://localhost:5173. The app proxies /api to the backend.

Flow

1. Input – Paste text or upload a PDF, then click Generate Understanding.
2. Processing – Steps: Generating perspectives → Synthesizing understanding → Creating quiz.
3. Learning – Read the unified explanation (card-based, ~2 min), then Start Quiz.
4. Quiz – 5 questions (2 MCQ, 2 short-answer, 1 application). One at a time.
5. Results – Score, strengths, weak areas, feedback. Option to go back to explanation or start over.

Agents (backend)

• Agent 1 – Exam Perspective (definitions, keywords, bullets).
• Agent 2 – Concept Understanding (simple “why”, analogies).
• Agent 3 – Cheat Sheet (ultra-short revision bullets).
• Agent 4 – Example & Intuition (1–2 strong examples).
• Agent 5 – Meta Understanding (merge 1–4 into one coherent output).
• Agent 6 – Quiz Generation (5 questions from the explanation).
• Agent 7 – Quiz Evaluation (score, feedback, weak concepts).

Tech stack

• Frontend: React (Vite), minimal CSS, card-based layout.
• Backend: FastAPI, Python 3.10+.
• AI: Single LLM (OpenAI-compatible) with different prompts per agent.

Demo tips

• Use a short paragraph or a 1–2 page PDF for fast runs.
• Keep the backend running on port 8000 while demoing the frontend.