Course Overview
In this course, you’ll learn to use the Google Agent Development Kit to build complex, Multi-Agent Systems. You will build agents equipped with tools, and connect them with parent-child relationships and flows to define how they interact. You’ll run your agents locally and deploy them to Vertex AI Agent Engine to run as a managed agentic flow, with infrastructure decisions and resource scaling handled by Agent Engine.
Who should attend
Technical roles from partner organizations:
- Machine learning engineers
- Gen AI engineers
Prerequisites
Python, gen AI prompt engineering, gen AI tool use
Course Objectives
- Build an agent with tools using the Google Agent Development Kit.
- Establish interaction patterns between multiple agents with parent-child relationships and flows.
- Utilize features such as session memory, artifact storage, and callbacks.
- Deploy a multi-agent app to Agent Engine.
- Query an agent app running on Agent Engine.
Outline: Deploy Multi-Agent Systems with Agent Development Kit and Agent Engine (DMASADKAE)
Module 1 - Get started with the Agent Development Kit
Topics
- Basics of building an agent in the Agent Development Kit.
Objectives
- Explain how the Agent Development Kit compares to other tools such as the Google Gen AI SDK or LangChain.
- Describe the parameters used to build an agent in Agent Development Kit.
Activities
- Lecture and lab demo
Module 2 - Empower Agent Development Kit agents with tools
Topics
- Enhance agents with tools and cover the growing breadth of available tools.
Objectives
- Discuss the importance of structured docstrings and typing when writing tool functions for agents.
- Demonstrate the ability to provide tools to an agent.
- List common and useful tools available for the Agent Development Kit agents, including LangChain tools.
Activities
- A lecture and a lab
Module 3 - Build Multi-Agent Systems with Agent Development Kit
Topics
- Manage communication and task-sharing between agents through parent-child relationships and flows to enable coordinated responses to queries.
Objectives
- Describe the directory structure and naming conventions encouraged by the Agent Development Kit.
- Demonstrate the ability to create multiple agents and relate them to one another with parent-child relationships.
- Describe the different flow options and when you might use them.
- Get responses that have passed through multiple agents.
- Control content at different points with callbacks.
Activities
- A lecture and a lab
Module 4 - Deploy Agent Development Kit agents to Agent Engine
Topics
- Deploying agent apps to Agent Engine and querying responses.
Objectives
- Describe the benefits of deploying agents, especially Multi-Agent Systems, to Agent Engine over self-hosting, such as in Vertex AI online predictions.
- Demonstrate deploying to Agent Engine.
- Demonstrate querying a deployed agent app.
Activities
- A lecture and a lab
Module 5 - Evaluate agent systems
Topics
- Evaluate agents within the Agent Development Kit.
Objectives
- Evaluate agents within the Agent Development Kit.
- Use the web interface to view evaluations.
Activities
- A lecture and a lab
Module 6 - Provide agents session memory and artifact storage
Topics
- Provide agents session memory to iterate on a current state. Grant agent access to documents to enable them to craft and refine documents.
Objectives
- Utilize sessions in an agent.
- View and debug sessions in the Agent Framework web UI.
- Utilize artifact storage.
Activities
- A lecture and a lab