A Hierarchical Multi-Agent System for unified e-commerce customer support using Generative AI.
# Install dependencies
pip install -r requirements.txt
# Setup environment
cp .env.example .env
# Edit .env with your GITHUB_TOKEN
# Run migrations
python manage.py migrate
# Generate sample data
python scripts/generate_data.py
# Start server
python manage.py runserver 8000Open: http://localhost:8000/
| Component | Technology |
|---|---|
| Framework | Django 4.2 + LangGraph |
| LLM | GPT-4o via GitHub Models API |
| Database | SQLite (4 virtual databases) |
| Frontend | HTML/CSS/JavaScript |
| State Machine | LangGraph StateGraph |
"What are my recent orders?"
"I ordered a Gaming Monitor but it hasn't arrived. Where is my package?"
| Agents | Example |
|---|---|
| 1 Agent | "What are my recent orders?" |
| 2 Agents | "I ordered a Laptop. Where is my package?" |
| 3 Agents | "I ordered a Gaming Monitor but it hasn't arrived. I opened a ticket. Where is my package and has my ticket been assigned?" |
| Document | Description |
|---|---|
| 📋 Problem Statement | Original requirements |
| 🏗️ Architecture | System design & diagrams |
| 📊 Data Flow | Query processing flow |
| 🔌 API Reference | REST API endpoints |
| 📐 Diagrams | Mermaid visualizations |
| ✅ Implementation Report | Deliverables checklist |
├── apps/
│ ├── orchestrator/ # Super Agent (LangGraph)
│ ├── shopcore/ # E-commerce Agent
│ ├── shipstream/ # Logistics Agent
│ ├── payguard/ # Payments Agent
│ └── caredesk/ # Support Agent
├── api/ # REST API endpoints
├── docs/ # Documentation
├── scripts/ # Data generation
└── templates/ # Web UI
- ✅ 4 Specialized Agents - Text-to-SQL for each domain
- ✅ Super Agent Orchestration - LangGraph state machine
- ✅ Parallel Execution - ThreadPoolExecutor for agents
- ✅ Intent Caching - 40% latency reduction
- ✅ Pattern Matching - 60% queries skip LLM
- ✅ Multi-Intent Detection - Complex 3-agent queries
- ✅ Conversation Memory - MemorySaver checkpointing
- ✅ Web UI - Real-time chat interface