Vertex AI Agent Builder overview

Vertex AI Agent Builder is a suite of products that help developers build, scale, and govern AI agents in production. Vertex AI Agent Builder provides a full-stack, secure foundation that supports the entire agent lifecycle:

  • Build: Build agents using frameworks such as Agent Development Kit (ADK) or any other open-source framework of your choice.

    • Agent Development Kit (ADK) (Supported in preview) is an open-source framework that simplifies the process of building sophisticated multi-agent systems while maintaining precise control over agent behavior.

    • Agent Garden (Supported in preview) is a library in the Google Cloud console where you can find and explore sample agents and tools that are designed to accelerate your development. In the Agent Garden, you can find the following:

      • Agents: These are prebuilt, end-to-end solutions for specific use cases. For example, you might find an agent designed for customer service, data analysis, or creative writing. These agents are ready to be used and can be customized to fit your needs. Only Google can publish agents to the Agent Garden. To use an agent from the garden, you can select it and deploy it to your project.

      • Tools: These are individual components that you can add to your own agents. Tools provide specific functionalities, like interacting with a database, calling an external API, or performing a search. You can browse the available tools in the Agent Garden and integrate them into your agents built with the Agent Development Kit (ADK). This lets you extend your agent's capabilities without having to build these functionalities from scratch.

      Go to Agent Garden

  • Scale: Scale your agents into production with built-in testing, release management, and reliability at a global and secure scale.

  • Govern: Monitor what your agents are doing with an audit trail for end-to-end observability.

The following diagram shows the components of Vertex AI Agent Builder: Agent Builder Overview

To learn more about AI agents in general, see the following:

Workflow for building and deploying agents

  1. Discover agent samples and tools specific to your use cases in the Agent Garden.

  2. Build and test your agent using the Agent Development Kit.

  3. Deploy your agent to Vertex AI Agent Engine.

What's next