This agent sample acts as a helpful virtual assistant for an auto insurance company. It is able to register new members and perform several functions for existing members, including filing claims, providing roadside assistance, and returning information about reward offers from partner companies. The agent uses several tools to fulfill these tasks. The tools are custom APIs registered in Apigee API hub.

The agent has access to the following tools:

• membership: Registers new members.
• claims: Handles claims.
• roadsideAssistance: Provides roadside assistance, including towing services.
• rewards: Finds nearby reward offers from partners.

The tools are provided by custom APIs. The specifications are imported to API hub and then referenced in the agent code using ADK's built-in ApiHubToolset. This lets agent developers easily turn any existing API from their organization's API catalog into a tool with just a few lines of code. For this sample, the APIs themselves are served using Apigee.

• Python 3.12+
• uv for dependency management:

  curl -LsSf https://astral.sh/uv/install.sh | sh

• Google Cloud Project with the following roles assigned
  • Apigee Organization Admin
  • Secret Manager Admin
  • Storage Admin
  • Service Usage Consumer
  • Logs Viewer

Once you have created your project, install the Google Cloud SDK. Then run the following command to authenticate:

gcloud auth login

You also need to enable certain APIs. Run the following command to enable:

gcloud services enable aiplatform.googleapis.com

For this sample you must also provision Apigee and API hub in your project, to serve the APIs that act as the tools for the agent.

The API assets and additional prerequisite instructions are available in the Apigee Samples repo here: Auto Insurance Agent APIs.

If you already have Apigee and API hub provisioned in your project, you can simply deploy the assets by following the quickstart below.

Follow the instructions in this GCP Cloud Shell tutorial.

Use the Agent Starter Pack to create a production-ready version of this agent with additional deployment options. The easiest way is with uvx (no install needed):

uvx agent-starter-pack create my-auto-insurance-agent -a adk@auto-insurance-agent

# Create and activate a virtual environment
python -m venv .venv && source .venv/bin/activate # On Windows: .venv\Scripts\activate

# Install the starter pack and create your project
pip install --upgrade agent-starter-pack
agent-starter-pack create my-auto-insurance-agent -a adk@auto-insurance-agent

The starter pack will prompt you to select deployment options and provides additional production-ready features including automated CI/CD deployment scripts.

1. Clone the repository:

   git clone https://github.com/google/adk-samples.git
   cd python/agents/auto-insurance-agent

   For the rest of this tutorial ensure you remain in the python/agents/auto-insurance-agent directory.

2. Install the dependencies:

   uv sync --dev

3. Configure settings:

   • Set the following environment variables. You can set them in your .env file (modify and rename .env.example file to .env) or set them directly in your shell. For example:

   export GOOGLE_GENAI_USE_VERTEXAI=1
   export GOOGLE_CLOUD_PROJECT=my-project
   export GOOGLE_CLOUD_LOCATION=my-region
   export GOOGLE_CLOUD_STORAGE_BUCKET=my-storage-bucket  # Only required for deployment on Agent Engine

You can run the agent locally using the adk command in your terminal:

1. To run the agent from the CLI:

   adk run auto_insurance_agent

2. To run the agent from the ADK web UI:

   adk web

   Then select the auto-insurance-agent from the dropdown.

The agent can also be deployed to Vertex AI Agent Engine using the following commands:

uv sync --group deployment
uv run python3 deployment/deploy.py

When the deployment finishes, it will output the resource ID of the remote agent deployment, for example:

Created remote agent: projects/<PROJECT_NUMBER>/locations/<PROJECT_LOCATION>/reasoningEngines/<AGENT_ENGINE_ID>

For more information on deploying to Agent Engine, see here.

The deployment script adds the AGENT_ENGINE_ID to your .env file. To test the remote agent, simply run:

python3 deployment/test_deployment.py

You may then interact with the deployed agent from the shell. You can type quit at any point to exit.

Here's a quick example of how a user might interact with the agent:

User:

hi

Agent:

Welcome to Cymbal Auto Insurance! How can I help you today? Do you have a member ID?

User:

37184324

Agent:

Thanks, Tom! What can I help you with today?

User:

My car broke down and I need help

Agent:

What type of help do you need? Do you need a tow, a jump start, fuel, a tire change, or help unlocking your vehicle? Also, can you provide me with your location? You can give me an address or an approximate location by cross street.

User:

I ran out of gas. I'm on the shoulder of eastbound I-70 near the Kipling exit

Agent:

Okay, I'm creating a request for you now. I have found a company nearby who can help. They estimate they will be there in about 45 minutes. The company is called Roadside Rescue. They will give you a call shortly. Is there anything else I can help you with?

User:

No that's all thanks

Agent:

Thank you for contacting Cymbal Auto Insurance.