Agent Development Kit on Google Cloud Platform
A Compact Field Guide for Builders, Beginners & the Curious
PREFACE
Before We Begin
Every great technology starts as something that sounds like science fiction and ends up feeling like common sense. This book is your guide to building AI agents that reason, plan, and act—using Google’s ADK and GCP.
📖 HOW TO READ THIS BOOK
Each chapter is self-contained. Jump to the chapter that answers your current question, or read front-to-back for a complete understanding.
CONTENTS
What's Inside
Chapter 01: The Big Picture — What Is ADK and Why GCP? Understand the vision behind ADK and how GCP supercharges your agents with scalability, reliability, and integration. Chapter 02: Core Concepts — Agents, Tools, and Memory Dive into the fundamental building blocks: agents as decision-makers, tools as their hands, and memory as their notebook. Chapter 03: Architecture — How ADK Is Wired Together Explore ADK’s architecture through the analogy of a newsroom, with runners, agents, and event buses. Chapter 04: Getting Started — Your First Agent Build and run your first agent in under 50 lines of code. A hands-on introduction to ADK. Chapter 05: Multi-Agent Systems — Agents Talking to Agents Learn how to design systems where multiple agents collaborate, specialize, and delegate tasks. Chapter 06: GCP Services — ADK's Superpower Unlocked Unlock the power of Vertex AI, Cloud Run, Firestore, and more to deploy and scale your agents. Chapter 07: Production Use Case — Real-Time Claims Assistant A deep dive into building a real-world, production-ready AI agent for processing insurance claims. Chapter 08: Best Practices & Pitfalls Avoid common mistakes and adopt best practices for building reliable, cost-effective agents. Chapter 09: What's Next — The Road Ahead Explore the future of agentic AI, including A2A protocols, MCP integration, and long-running agents.