Skip to main content

Module 11: Enterprise Integration with OpenAPI Tools

Theory​

Connecting to the Enterprise Ecosystem​

Modern enterprises run on RESTful APIs. Whether it's a CRM, an ERP, or a custom internal service, these systems often expose their functionality through OpenAPI specifications. To build truly useful AI agents, you must know how to connect them to these existing enterprise services.

From Specification to Node​

In ADK 2.0, an OpenAPI tool is often used as a specialized Node in your graph or as a tool attached to an Agent.

1. Generating the Client​

You typically use a tool like openapi-python-client to generate a robust, type-safe Python client from your .json or .yaml specification.

2. Wrapping in a Function Tool​

Once you have the client, you write a small wrapper function that initializes the client and makes the API call. This wrapper follows the standard ADK 2.0 tool pattern (Pydantic models and ToolContext).

from google.adk import Agent
from my_api_client import Client, AuthenticatedClient
from my_api_client.api.default import get_data

def fetch_external_data(query: str) -> dict:
    """Fetches data from the production API."""
    client = AuthenticatedClient(base_url="https://api.example.com", token="...")
    response = get_data.sync(client=client, q=query)
    return response.to_dict()

# Register with Agent
agent = Agent(name="api_expert", tools=[fetch_external_data], ...)

Key Takeaways​

  • OpenAPI is the industry standard for describing REST APIs.
  • Use Code Generation to create type-safe Python clients from specs.
  • Wrap the generated client in an ADK Function Tool for easy integration.
  • This pattern allows your agent node to securely and reliably communicate with any enterprise service.