🍷 Wine E-commerce Assistant

An intelligent sommelier assistant that listens, thinks, and responds. Designed to guide customers of Les Celliers de Sion through an interactive and multimodal shopping experience.

Project Highlights

This project transforms the classic e-commerce experience into a fluid conversation using an advanced RAG (Retrieval Augmented Generation) architecture.

• Contextual Intelligence (RAG): Combines the power of a LLM with a vector knowledge base (FAISS) built from PDF sheets and the official website.
• Voice Interactions: Input: Fast audio transcription via Whisper.
• Autonomous Agent (LangGraph): An agent capable of planning its actions using multiple tools before responding and giving the source of its information.
• Performance: Backend optimized with FastAPI and package management via uv.
• Interface: Responsive UI built with React and Vite.

---

Quick Start

Prerequisites

• Install Ollama

  • Follow instructions at https://ollama.ai to install it on your system

• For Mac and Linux users, install zbar (used for Qr code reading):

  # MacOS
  brew install zbar

  # Linux 
  sudo apt-get install libzbar0

  It should be already included in Windows installation.

Setup and Run

# Clone the repository
git clone https://github.com/anouillz/Ecommerce-assistant.git
# Navigate to project directory
cd Ecommerce-assistant

# Pull ollama moddel 
ollama pull ministral-3:14b

# Package management
uv sync
# Run backend server
uv run python backend/api.py

# navigate to frontend directory
cd frontend
# install npm and run server
npm install
npm run dev

---

Documentation

You can find more detailed information in the documentation:

• User Guide: Instructions for setting up, using, and troubleshooting the assistant.
• Developer Guide: Technical details about the architecture, code structure, and development workflow.