Skip to content

ro1406/adk-report-generator

Repository files navigation

ADK Workshop on Multi-Agent Workflows With ADK

This repository contains the code and examples presented at the following workshops:

  • "It Starts with a Stock Bot: An Agentic AI Design Pattern for Scalable AI Workflows" at Google Developer Group DevFest 2025, held at Google HQ, Dubai.
  • "Creating Multi-Agent Workflows with Gemini Enterprise & ADK" at Build With AI, Google Developer Groups & Zaka AI, held at In5 Tech, Dubai.

Overview

This workshop demonstrates how to build scalable AI workflows using Google's Agent Development Kit (ADK) through practical examples. The repository showcases various AI agents that solve real-world problems, from financial analysis to creative tasks and research assistance.

🚀 Quick Start

To run any of the agents with a web interface, use:

adk web

This will start the ADK web interface at http://127.0.0.1:8000/dev-ui/ where you can interact with all the available agents.

📁 Repository Structure

Market Analysis Agent (Stock Bot) 🏦

Location: adk-report-generator/market-analysis-agent/

The main example from the workshop - a sophisticated financial analysis system that demonstrates the agentic AI design pattern for scalable AI workflows.

Features:

  • Stock Data Retrieval: Fetches real-time stock data using Yahoo Finance API
  • News Analysis: Analyzes market sentiment and news impact
  • Technical Analysis: Provides comprehensive technical indicators and trends
  • Report Generation: Creates detailed investment reports with actionable insights

Agent Architecture:

  • Interface Agent: Coordinates user interactions and workflow execution
  • Data Retriever Agent: Handles stock data collection and news gathering
    • API Agent: Specialized sub-agent for Yahoo Finance data retrieval
    • News Agent: Specialized sub-agent for news analysis and sentiment
  • Summarizer Agent: Processes and analyzes collected data into comprehensive reports

File Structure:

adk-report-generator/
├── market-analysis-agent/
│   ├── agent.py                    # Main interface agent
│   ├── prompt.py                   # Agent instructions and prompts
│   └── sub_agents/
│       ├── data_retriever_agent/
│       │   ├── agent.py            # Data collection coordinator
│       │   └── data_sub_agents/
│       │       ├── api_agent/      # Yahoo Finance API integration
│       │       └── news_agent/     # News analysis and sentiment
│       └── summarizer_agent/
│           ├── agent.py            # Report generation agent
│           └── prompt.py           # Analysis instructions
├── lol.ipynb                       # Jupyter notebook for testing
└── README.md

🛠️ Prerequisites

  • Python 3.8+
  • Google Cloud Project with Vertex AI enabled
  • ADK (Agent Development Kit) installed
  • Google Cloud Credentials properly configured

📋 Installation

  1. Clone the repository:
git clone <repository-url>
  1. Install ADK and dependencies:
pip install google-adk
  1. Set up Google Cloud credentials:
export GOOGLE_APPLICATION_CREDENTIALS="path/to/your/credentials.json"
  1. Configure environment variables: Each project contains an env.example file. Copy it to .env and configure:
cp env.example .env
# Edit .env with your Google Cloud configuration

🚀 Running the Market Analysis Agent

Method 1: ADK Web Interface (Recommended)

adk web

Then navigate to http://127.0.0.1:8000/dev-ui/ and select the market analysis agent.

Method 2: Direct Access

Navigate directly to the market analysis agent:

# Navigate to http://127.0.0.1:8000/dev-ui/?app=interface_agent

Method 3: Jupyter Notebook Testing

Use the provided Jupyter notebook for testing and experimentation:

cd adk-report-generator
jupyter notebook lol.ipynb

🏗️ Architecture Pattern

This repository demonstrates the Agentic AI Design Pattern with the following key concepts:

1. Hierarchical Agent Structure

  • Root Agent: Main interface and coordinator
  • Sub-agents: Specialized agents for specific tasks
  • Tool Integration: Each agent has access to relevant tools

2. Scalable Workflow Design

  • Sequential Processing: Agents work in coordinated sequences
  • Parallel Execution: Multiple agents can work simultaneously
  • State Management: Proper handling of conversation state and context

3. Modular Components

  • Agent Definitions: Clear separation of agent responsibilities
  • Tool Libraries: Reusable tools across different agents
  • Configuration Management: Environment-based configuration

📊 Example: Market Analysis Workflow

The stock bot demonstrates a complete agentic workflow:

  1. User Input: Stock symbol and date range
  2. Data Collection: API agent fetches stock data
  3. News Analysis: News agent gathers and analyzes market sentiment
  4. Technical Analysis: Processing of price trends and indicators
  5. Report Generation: Comprehensive analysis with recommendations

🔧 Development

Extending the Market Analysis Agent

  1. Adding New Data Sources: Create new sub-agents in data_sub_agents/
  2. Enhancing Analysis: Modify the summarizer agent prompts and logic
  3. New Tools: Add tools to the api_agent/tools.py for additional functionality
  4. Testing: Use the Jupyter notebook for rapid prototyping and testing

Agent Structure

  • Interface Agent: Main coordinator that handles user input and orchestrates workflow
  • Sequential Agent: Manages the execution flow between data retrieval and summarization
  • Sub-agents: Specialized agents for specific tasks (API data, news analysis, summarization)

📝 Workshop Takeaways

This repository demonstrates:

  • Practical AI Implementation: Real-world applications of AI agents
  • Scalable Architecture: Patterns that work for complex workflows
  • Google Cloud Integration: Leveraging Vertex AI and other GCP services
  • Best Practices: Code organization, error handling, and user experience

🤝 Contributing

This repository serves as educational material from the workshop. Feel free to:

  • Experiment with the market analysis agent
  • Extend functionality by adding new data sources or analysis methods
  • Improve the agent prompts and instructions
  • Share improvements and learnings

📄 License

This project is provided as educational material from the Google Developer Group DevFest 2025 workshop.

🙋‍♂️ Support

For questions about the workshop content or implementation:

  • Review the market analysis agent code and structure
  • Check the ADK documentation
  • Experiment with the provided Jupyter notebook
  • Examine the agent prompts and tool implementations

Workshop: "It Starts with a Stock Bot: An Agentic AI Design Pattern for Scalable AI Workflows"
Event: Google Developer Group DevFest 2025
Location: Google HQ Dubai
Repository: Practical examples and code from the workshop

About

Use ADK for data collation, analysis, and report generation

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors