GDGC Build with AI Workshop

Welcome to the GDGC Build with AI Workshop! This hands-on workshop will teach you how to use Google's Gemini API to create AI agents that can call custom functions based on natural language input.

Prerequisites

• Python 3.11 or higher
• A Google Cloud account with access to Vertex AI
• Basic understanding of Python and async programming
• Your Google Cloud API key

Setup Instructions

1. Setup the codebase

   Using Firebase Studio:

   1. Go to Firebase Studio
   2. Click "Import Repo"
   3. Enter the work repository URL:

   https://github.com/Developer-Student-Club-UoA/build-with-ai-workshop-2025

   4. Wait for the codebase to initialise and setup to complete. You will know once the README opens. While you wait for this, you can go to step 2 of the setup instructions (Get Gemini API key).

   Setting up the project locally:
   (if you setup with firebase, skip to step 2)

   1. Fork the Repository

   2. Clone your forked repository

      git clone https://github.com/Developer-Student-Club-UoA/build-with-ai-workshop-2025.git
      cd build-with-ai-workshop-2025

   3. Set Up Your Python Environment

      # Create a virtual environment
      python -m venv venv
      
      # Activate the virtual environment
      # On Windows:
      .\venv\Scripts\activate
      # On macOS/Linux:
      source venv/bin/activate
      
      # Install dependencies
      pip install -e .

   4. Get Gemini API key

      Note: university accounts do not support Google AI studio, we recommend signing up with your personal email.

      Head to the AI studio website to get setup.

Project Structure

.
├── README.md
└── workshop/
    ├── module1        # Basic setup and configuration of Gemini agent
    ├── module2        # Single function calling implementation
    └── module3        # Advanced function chaining and conversation flow
        ├── main.py        # Main application with Gemini agent
        └── tools.py       # Function declarations and implementations

Understanding the Code

1. Function Declarations (tools.py)

The tools.py file contains two key components for each function:

• A function declaration that tells Gemini what the function does and what parameters it expects
• The actual function implementation that executes when Gemini calls it

Example:

get_current_weather_declaration = {
    "name": "get_current_weather",
    "description": "Gets the current weather for a given location.",
    "parameters": {
        "type": "object",
        "properties": {
            "location": {
                "type": "string",
                "description": "The city name (e.g. 'San Francisco', 'New York', 'London')"
            }
        },
        "required": ["location"]
    }
}

def get_current_weather(location: str) -> dict:
    """
    Corresponds to the get_current_weather_declaration.
    Will fetch the weather given a location.
    """

2. Gemini Agent (main.py)

The main.py file contains the GeminiAgent class that:

• Initializes the Gemini client with system instructions
• Manages the conversation and function context
• Processes responses and chats

Workshop Modules

Module 1: Setting up the Gemini Agent

1. Set up your environment variables and authentication
2. Initialize the Gemini client with basic chat capabilities
3. Test basic conversations with the agent

Module 2: Adding New Functions

1. Create new function declarations
2. Implement the functions with context awareness
3. Add them to the agent's available tools

Module 3: Chaining functions

1. Learn how to chain multiple functions together to solve complex queries
2. Handle function dependencies and order of operations
3. Practice with example scenarios like:
   • Getting weather for current location (chains location + weather)
   • Converting temperatures between units (chains weather + conversion)
   • Comparing weather in different cities (chains multiple weather calls)

Running the Application

1. Configure Gemini API Key:

   # Create or edit your .env file
   echo "API_KEY=YOUR_GEMINI_API_KEY" > .env

2. Run the application:

   python src/main.py

3. Try some example queries:

   • "What's the current weather in the captial of Japan?"
   • "How hot is it in my location?"

Getting Help

• Gemini API documentation
• Project IDX documentation
• Firebase documentation
• Google Cloud Authentication guide

Next Steps

After completing the workshop:

1. Explore advanced Gemini features
2. Add more complex function capabilities
3. Add more complex function chains
4. Implement persistent conversation storage

License

This workshop material is licensed under the MIT License. See the LICENSE file for details.