EduCopilot 🎓🤖

EduCopilot is a prototype Agentic AI system designed to streamline and enhance higher education curriculum management and the learning process. Powered by LangGraph, FastAPI, and Next.js, it serves as an intelligent assistant for lecturers to automate administrative tasks, validate curriculum standards, and generate actionable insights from student data.

---

🌟 Key Features

EduCopilot provides an end-to-end ecosystem for course management through specialized AI agents:

1. RPS (Rencana Pembelajaran Semester) Builder Automatically draft structured syllabuses and course plans based on course descriptions, Learning Outcomes (CPL/CPMK), and weekly topics.

2. Compliance & Review Engine Validates the generated or uploaded RPS against 24 strict national higher education standards (SN-Dikti) criteria to ensure administrative and academic compliance.

3. Knowledge Base (RAG-based) Upload reference materials (PDFs, docs). The system chunks, embeds, and stores them in ChromaDB to provide precise, context-aware AI responses using Retrieval-Augmented Generation (RAG).

4. Quiz Generator & Editor Generate contextual quizzes (Multiple Choice and Essay) based on RPS materials and Bloom's Taxonomy. Includes a built-in draft editor and dynamic CSV template generation for student submissions.

5. AI Auto-Grading Automatically grade student submissions (especially essays) based on predefined rubrics. It provides detailed, personalized feedback for each student.

6. Remedial & Auto-Material Generation Analyzes grading results to identify weak topics across the class and automatically generates targeted supplementary learning materials and remedial recommendations.

7. Dashboard & Analytics Get a bird's-eye view of course performance, student progress, quiz statistics, and RPS compliance ratios.

---

🏗️ Architecture

The system follows a modern decoupled architecture:

• Frontend (Presentation Layer): Built with Next.js, TypeScript, and Tailwind CSS. Provides a responsive, dashboard-driven UI for lecturers.
• Backend API Layer: Powered by FastAPI (Python). Handles routing, business logic, and communication between the frontend and the AI layer.
• Agentic AI Layer: Uses LangGraph to orchestrate goal-driven agents (RPS Agent, Quiz Agent, Grading Agent, etc.) that perform multi-step reasoning. Uses Anthropic's LLM API.
• Data & Tools Layer:
  • Relational Data: SQLite (educopilot.db) via SQLAlchemy ORM.
  • Vector Database: ChromaDB (for document embeddings).
  • Search Tool: Tavily Search integration for external information gathering.

---

🚀 Getting Started

Prerequisites

• Node.js (v18+ recommended)
• Python (v3.10+ recommended)
• uv (ast Python package installer and resolver)
• Git

1. Clone the Repository

git clone https://github.com/szyxxx/Agentic-Educator-Copilot.git
cd Agentic-Educator-Copilot/app

2. Backend Setup

The backend relies on FastAPI and uv for dependency management.

cd backend

# Create a virtual environment and install dependencies
uv venv
uv pip install -r requirements.txt

# Configure Environment Variables
# Create a .env file based on the provided .env.example (if available)
# Make sure to add your ANTHROPIC_API_KEY
echo "ANTHROPIC_API_KEY=your_api_key_here" > .env

# Run the development server
uv run uvicorn app.main:app --reload

The backend API will be available at http://localhost:8000

3. Frontend Setup

The frontend is a Next.js application.

cd frontend

# Install dependencies
npm install

# Run the development server
npm run dev

The frontend application will be available at http://localhost:3000

---

🧠 AI Agents in EduCopilot

EduCopilot utilizes a Multi-Agent architecture where each agent has a specific domain of responsibility:

• RPS Agent: Formulates initial drafts of the semester learning plan.
• RPS Review Agent: Runs the compliance engine to find missing elements in the RPS.
• Quiz Agent: Uses RAG to generate relevant quiz questions and rubrics.
• Grading Agent: Evaluates student answers against rubrics and provides personalized feedback.
• Remedial Agent: Analyzes overall scores to recommend actions for underperforming students.
• Auto Material Agent: Searches or generates new supplementary materials based on class weaknesses.

---

🔒 Privacy & Limitations

• Decision Support System: EduCopilot is designed to assist lecturers, not replace them. All generated RPS, quizzes, and grades should be reviewed by a human before final publication.
• Data Privacy: Ensure that sensitive student data is anonymized before using the AI auto-grading features, as data is processed by external LLMs (Anthropic).

---

📝 License

This project was developed as a Final Project for the II5003 Applied Artificial Intelligence course. All rights reserved.