Hands-On AI: Building Agents with the Google Agent Development Toolkit (ADK)

Course details

This course teaches the skills to build and test your own agents using the Google Agent Development Kit. Instructor Jigyasa Grover shows you how to define agents, equip them with various tools including custom functions and third-party integrations, and configure their underlying language models. The course also covers managing input, output, and context within agents to enhance their functionality. Jigyasa also explores the concept of workflow agents, including sequential, loop, and parallel agent systems, and how to leverage built-in and MCP tools within the ADK. By the end of this course, you should understand how to develop agents and multi-agent systems using the Google Agent Development Kit.

Learning objectives
Set up an ADK project and build your first agent.
Define Agents with persona to take on specialized tasks.
Configure your agent’s use of models and manage input and outputs.
Understand and use different workflow agents.
Use tools through functions, built-in utilities and third-party services.
Use tools through MCP.

Concepts

Getting Started

01 - Welcome

1. Introduction

02 - Multi-agent systems - Core concepts
03 - Introducing Agent Development Kit (ADK) - Your AI building blocks
04 - ADK quickstart - Building your first text-based agent

2. Diving Deeper into ADK Agents

05 - Core agent types - LLM, workflow, and custom agents
06 - LLM agents - The engine of intelligent conversation and reasoning
07 - Building LLM agents - From basic prompts to tool-enabled power
08 - Workflow agents - Orchestrating complex tasks with precision and order
09 - Custom agents - Ultimate flexibility with Python logic
10 - Strategic agent selection - Choosing the right ADK agent

3. Empowering Agents - Tools and Integrations

11 - ADK's built-in toolbox - Essential capabilities for your agents
12 - Function tools - Tailoring agent actions with Python
13 - Expanding horizons - Integrating third-party tools and live data
14 - Leveraging Google Cloud - Native ADK integrations and secure services
15 - Tool design - Best practices for reliable and secure ADK agents

4. Orchestrating Multi-Agent Systems and Context

16 - Hierarchical designs - Delegation, routing, and orchestration
17 - Parallel agents - Concurrent execution for enhanced efficiency
18 - Loop agents - Iterative processes and refinement
19 - Conversational blueprint - Understanding session, state, and memory
20 - Context-aware agents - Practical session and state management
21 - Putting it all together - Advanced multi-agent orchestration with a coordinator

5. Advanced Agents - Persistent Artifacts and Lifecycle Callbacks

22 - Artifacts - Giving agents lasting, versioned memory
23 - Data artifacts - Best practices for naming, versioning, and namespaces
24 - Callbacks - Customizing the agent lifecycle with precision hooks

6. Productionizing Agents - Deployment, Evaluation & Responsible AI

25 - Deploying agents - Agent Engine, Cloud Run, and GKE
26 - Operational excellence - Security, monitoring, and observability
27 - Agent quality - Trajectory analysis and response evaluation
28 - Responsible agents - Upholding safety, ethics, and trust

Conclusion

29 - Course recap - The road ahead