Ultimate Guide: Deploy Google ADK Agents to Vertex AI & Cloud Run (Step-by-Step Tutorial)
Master the deployment of AI agents built with Google's Agent Development Kit (ADK) to scalable cloud environments with this ultimate step-by-step tutorial. We clarify documentation gaps and provide clear walkthroughs for:
Vertex AI Agent Engine (Fully Managed Google Cloud Service)
Google Cloud Run (Managed Container Platform)
What You'll Learn:
Prerequisites for Google Cloud deployment (APIs, Billing, Permissions, GCS).
Setting up Python environments and installing ADK.
Deploying to Vertex AI Agent Engine using a Colab notebook.
Defining agents with tools like Google Search (featuring Gemini 2.5 Flash Preview).
Local vs. Remote testing of agents.
Managing deployed resources on Agent Engine.
Deploying to Cloud Run with ADK CLI and gcloud.
Structuring projects (agent.py, requirements.txt, .env).
Handling authentication and IAM permissions.
Enabling web UI for seamless testing.
Testing agents on both platforms.
Perfect for developers and AI engineers using Google ADK to move agents from development to production on Google Cloud Platform.
Resources & Timestamps:
0:00 Introduction & Goal: Deploying ADK Agents
0:26 Overview of Deployment Options (Vertex AI Agent Engine, Cloud Run, Custom)
1:12 Why Focus on Vertex AI Agent Engine & Cloud Run (Documentation Gaps)
1:50 Deploying to Vertex AI Agent Engine
2:41 Prerequisites for Vertex AI Agent Engine (APIs, Billing, Permissions, GCS Bucket)
4:11 Colab Notebook Walkthrough: Setup & Installation
4:42 Colab: Authenticating and Initializing Vertex AI SDK
4:59 Colab: Defining the Agent (using Google Search Tool)
6:19 Colab: Preparing Agent for Deployment (AdkApp & Tracing)
6:41 Colab: Local Agent Testing (Create Session & Send Query)
7:25 Colab: Deploying the Agent to Agent Engine (agent_engines.create())
8:12 Colab: Remote Agent Testing (Create Remote Session & Send Query)
8:41 Colab: Retrieving an Existing Deployed Agent (Optional)
9:06 Colab: Cleaning Up Deployed Agent Engine Resources
9:38 Deploying to Google Cloud Run
10:15 Prerequisites for Cloud Run Deployment (Project Structure, Env Setup)
10:30 Cloud Run: Setting up Virtual Environment & Installing ADK
10:55 Cloud Run: Creating Project Directory and Files (agent.py, __init__.py, etc.)
12:24 Cloud Run: Setting Environment Variables
12:39 Cloud Run: Authentication (gcloud auth login & Project Config)
14:29 Cloud Run: Granting IAM Permissions to Service Account
15:01 Cloud Run: Running the Deployment Command (adk deploy cloud_run --with-ui)
15:37 Cloud Run: Testing the Deployed Agent via Web UI
17:14 Conclusion & Wrap-up
Drop your questions in the comments!
Links:
ADK:
https://google.github.io/adk-docs/dep...
Vertex AI Agent Engine quickstart:
https://cloud.google.com/vertex-ai/ge...
gcloud CLI:
https://cloud.google.com/sdk/docs/ins...
Notebook: https://github.com/LLM-Implementation...
Google Search Agent: https://github.com/LLM-Implementation...
A nice guide: https://github.com/GoogleCloudPlatfor...