Ai+ Training

This session explores how to design, build, and deploy multi-agent systems (MAS) using Google's Agent Development Kit (ADK) and the Agent-to-Agent Protocol (A2A).

We'll begin with foundational concepts of MAS, including how agents coordinate, communicate, and collaborate using different architectural patterns—sequential, parallel, loop, and hierarchical routing. Then, we'll dive into practical implementation strategies using ADK, showcasing how to compose workflows, build memory-aware agents, and define custom tools. Finally, we’ll unpack the Agent-to-Agent (A2A) protocol, a powerful standard that enables agents to discover, invoke, and collaborate with external agents securely using a standardized metadata format and JSON-RPC.

This session is ideal for AI developers, ML engineers, and technical architects building intelligent applications. 

Attendees will leave with a clear understanding of how to construct agentic workflows and leverage multi-agent orchestration to solve real-world problems—from research assistance to developer tooling.

Code Notebooks:

• Main Notebooks: goo.gle/agenticlab0

• Lab #1: goo.gle/multiagent1

• Lab #2: goo.gle/parallelagent

• Lab #3: goo.gle/loopagent

• Lab #4: goo.gle/agentastool

• Lab #5: goo.gle/a2a-tutorial

• Demo #1: colab.research.google.com/drive/1G6eMTfn35bgiitkMHdpCilV_Po3MvdlY