Inside the Google Agent Stack: A Deep Dive into Google agent Visual Builder , Agent Garden, and Web ADK

Audience exposure level: Level 2: Sometimes use AI

This course introduces participants to Google’s modern agent stack through a blend of conceptual learning and hands-on practice. Using the Agent Development Kit (ADK), participants explore how contemporary AI agents reason, plan, use tools, and coordinate within multi-agent systems. The first half of the course uses Colab demonstrations to unpack core architectural patterns - including memory, orchestration, and agent-as-tool design - while the second half provides a technical walk through of deploying agents locally using the Web ADK. By the end, participants will understand both the theoretical foundations and practical workflows needed to build, run, and refine advanced agentic interfaces powered by Gemini models.

• Understand the core components of Google’s Agent Development Kit (ADK) and how agents, sessions, and runners function conceptually.
• Recognise common multi-agent patterns—sequential, loop-based, and parallel—and understand when and why they are used.
• Identify how tools extend an agent’s capabilities, including how simple Python functions become callable actions within an agent workflow.
• Understand the high-level process for deploying agents locally using the Web ADK, including environment setup, API keys, and the browser interface.