Build a Recipe Assistant with Function Calling

#promptengineering #functioncalling #openai #python

Building a Recipe Assistant with Function Calling and Nutritional APIs

What You'll Learn

In this tutorial, you will build an intelligent Recipe Assistant that leverages Large Language Models (LLMs) to interpret natural language queries. You will learn how to connect these models to external data sources using Function Calling. This technique allows AI agents to perform real-world actions, such as fetching nutritional data or searching for specific ingredients dynamically.

This skill is critical for modern application development because static prompts are limited. By integrating APIs, your applications become dynamic, accurate, and capable of handling complex user requests without hallucinating facts. You will gain practical experience in prompt engineering, API integration, and structured output handling.

Prerequisites

Before starting, ensure you have the following tools and knowledge ready:

Python 3.8+: Installed on your local machine or accessible via a cloud environment like Google Colab.
OpenAI API Key: A valid key from OpenAI Platform to access GPT-4 or GPT-3.5 Turbo models.
Basic Python Knowledge: Familiarity with functions, dictionaries, and asynchronous programming concepts.
Requests Library: Install via pip install openai requests to handle HTTP communications.

Having these prerequisites in place ensures a smooth workflow. You can focus on logic rather than debugging environment issues later.

Setting Up Your Environment

Start by creating a clean project directory. This keeps your code organized and separates dependencies effectively. Create a new folder named recipe-assistant and navigate into it using your terminal. Inside this folder, create a virtual environment to isolate your project packages.

Activate the virtual environment and install the necessary libraries. Run the command pip install openai requests python-dotenv. The python-dotenv library helps manage sensitive information like API keys securely. Never hardcode secrets directly into your source code for security reasons.

Create a file named .env in your root directory. Add your OpenAI API key here as OPENAI_API_KEY=your_key_here. This approach protects your credentials if you share your code on platforms like GitHub. Always add .env to your .gitignore file immediately.

# .env file content
OPENAI_API_KEY=sk-your-actual-key-here

Next, create a main script file called app.py. Import the required modules at the top. You will need os for environment variables, openai for model interaction, and requests for API calls. Load the dotenv configuration at the start of your script to make the API key available globally.

import os
import json
import requests
from openai import OpenAI
from dotenv import load_dotenv

load_dotenv()
client = OpenAI(api_key=os.getenv("OPENAI_API_KEY"))