LuggageToShip AI Agent

AI-powered customer service agent for LuggageToShip.com using RAG (Retrieval-Augmented Generation) and agentic tool calling.

This project implements a complete AI agent system that combines:

• RAG for knowledge retrieval from a comprehensive knowledge base
• Agentic Tool Calling for executing shipping operations (rates, bookings, tracking, etc.)
• Conversation Memory for multi-turn dialogues with context retention
• Mock API simulating real backend services

• Natural Language Interface - Chat with the agent in plain English
• Knowledge Base - 16 comprehensive documents covering all services
• 6 Tool Functions - Rate quotes, bookings, tracking, locations, rewards, storage
• Conversation Memory - Maintains context across multiple turns
• Modern UI - Clean Streamlit interface with chat history
• Fully Dockerized - One-command deployment

• Docker and Docker Compose installed
• At least 8GB RAM available
• 20GB free disk space

cd luggagetoship-agent
chmod +x start.sh stop.sh
./start.sh

The startup script will:

1. Start all Docker services
2. Pull required AI models (Mistral-Nemo 12B, Nomic-Embed)
3. Ingest knowledge base into ChromaDB
4. Launch Streamlit UI

Access the application at: http://localhost:8501

./stop.sh

User Input → Streamlit UI → Agent Orchestrator
                                    ↓
                    ┌───────────────┴───────────────┐
                    ↓                               ↓
            RAG Retriever                   Tool Executor
                    ↓                               ↓
              ChromaDB                         Mock API
                    ↓                               ↓
            Knowledge Context              Tool Results
                    ↓                               ↓
                    └───────────────┬───────────────┘
                                    ↓
                            LLM (Ollama)
                                    ↓
                          Agent Response
                                    ↓
                        Conversation Memory
                                    ↓
                                 Redis

• Docker Desktop installed and running
• NVIDIA drivers (optional, for GPU acceleration)
• ~20 GB free disk space

See LuggageToShip_Windsurf_Implementation_Plan.md for detailed implementation guide.