This notebook demonstrates how to use the Financial Advisor agent with full observability through Arize AX tracing. The agent provides comprehensive financial advice through a team of specialized AI sub-agents:

1. Data Analyst Agent: Creates market analysis reports for specific stock tickers
2. Trading Analyst Agent: Develops trading strategies based on risk tolerance and investment duration
3. Execution Agent: Creates implementation plans for trading strategies
4. Risk Evaluation Agent: Analyzes risks and provides mitigation strategies

🔍 Arize Tracing: All agent interactions, tool usage, and LLM calls are automatically traced and sent to Arize for observability, performance monitoring, and evaluation.

⚠️ Important Disclaimer: This is for educational purposes only and does not constitute financial advice.

✅ You will need a Google Cloud account with Vertex AI enabled and an Arize account to run this notebook.

Set Up Keys & Dependencies

Install Dependencies

Google Cloud Setup

The Financial Advisor agent requires Google Cloud setup because it uses:

• Vertex AI for Gemini 2.5 Pro model
• Google Search tools for market data analysis

1. Create a Google Cloud Project (if you don't have one)

2. Enable required APIs:

   • Vertex AI API
   • Generative AI API

3. Set up authentication:

   gcloud auth application-default login
   gcloud auth application-default set-quota-project YOUR-PROJECT-ID
   

Environment Variables

Create a .env file in this directory with the following variables:

# Required for Google ADK
GOOGLE_GENAI_USE_VERTEXAI=1
GOOGLE_CLOUD_PROJECT=your-project-id
GOOGLE_CLOUD_LOCATION=your-project-location  # e.g., us-central1

# Required: Arize Configuration
ARIZE_SPACE_ID=your-arize-space-id       # Found in Space Settings page
ARIZE_API_KEY=your-arize-api-key         # Found in Space Settings page

Configure Tracing

Build Financial Advisor System

Here we define all the prompts and agents inline to make this notebook completely standalone.

Test the Financial Advisor Agent

The following helper functions allow you to interact with the financial advisor agent from within a Jupyter notebook. They handle agent initialization, session management, and provide async wrappers for querying the agent and displaying responses.

📊 Arize Tracing Dashboard

Once you start running the examples below, you can view the traces in Arize:

Arize Platform:

• Visit your Arize space dashboard at https://app.arize.com
• Navigate to your project: "google-adk-financial-advisor"
• View traces in real-time as agents execute
• Access advanced analytics and collaboration features

What You'll See in Arize:

• 🔗 Trace Trees: Complete execution flows from user query to final response
• ⏱️ Timing Data: Latency for each agent and tool call
• 💰 Token Usage: Input/output tokens and cost tracking
• 🔍 Tool Calls: Detailed logs of Google Search and agent interactions
• 📊 Performance Metrics: Aggregate statistics across all runs
• 🎯 Model Performance: Drift detection and performance monitoring
• 📈 Analytics: Advanced filtering, search, and analytics capabilities

The traces provide invaluable insights for debugging, optimization, and understanding how your multi-agent system performs in production.