AI Pizza Delivery Agent

An AI-powered pizza ordering agent with a modern React frontend and a FastAPI backend, using Google Gemini (AI Studio) for natural dialog.

---

Features

• Step-by-step, conversational pizza ordering agent
• Menu, toppings, sides, and dietary preferences
• Order summary and transcript download
• Modern React frontend, FastAPI backend
• Uses Google Gemini (AI Studio) LLM via API

---

Requirements

• Python 3.8+
• Node.js 16+
• Google Gemini API key (from Google AI Studio)
• (macOS, Windows, or Linux)

---

Setup Instructions

1. Clone the Repository

git clone https://github.com/aishvayy/PizzaDeliveryAgent.git

---

2. Backend Setup

cd backend
python3 -m pip install -r requirements.txt

Set up your Gemini API key

• Create a .env file in the backend/ directory with this line:

  GEMINI_API_KEY=your_actual_gemini_api_key_here
  

• The backend uses python-dotenv to load this key automatically.

Start the Backend

uvicorn main:app --reload

---

3. Frontend Setup

cd ../frontend
npm install
npm run dev

• Open http://localhost:5173 in your browser.

---

Usage

• Click “Show Menu” to view the pizza menu.
• Chat with the agent to place your order.
• After completing your order, download the order summary (JSON) and transcript (TXT).

---

Project Structure

backend/    # FastAPI backend, menu, Gemini LLM integration
frontend/   # React frontend (Vite)

---

Troubleshooting

• CORS errors: Make sure the backend is running and CORS is enabled (already set up in main.py).
• Gemini errors: Ensure your API key is valid and you have access to the Gemini models.
• Port conflicts: Default backend is on 8000, frontend on 5173.

---

.gitignore Recommendations

# Python
__pycache__/
*.pyc

# Node
node_modules/
dist/

# Mac
.DS_Store

# Env
.env

---

Credits

• Google Gemini (AI Studio) for LLM
• FastAPI
• React

---

Working dialog and Transcript

Video of the working dialog:

Pizza.Agent.1.1.mp4

Transcript is available upon hitting the download button, attached transcript for the demo conversation can be found here: (https://github.com/user-attachments/files/21471049/transcript.txt)

Screenshot of the order saved in a structured JSON format, available on the backend on the /order/sessionId path: