Google AI Releases ADK Go: A New Open Source Tool Designed to Empower Developers to Build Powerful AI Agents

How do you build reliable AI Agents that connect to your existing GO services without breaking the bank in another language? Google recently released Agent Development Kit for travel. GO developers can now build AI agents with the same framework that already supports Python and Java, while keeping everything within GO's standard toolset model.

For AI Devs and backend developers already using the services, this closes the gap. You no longer need a high-level Python stack for agents. You can demonstrate Agent Logic, Tool Usage, and tool implementation directly in GO code, then deploy the same agents in Vertex AI Agent Builder and Agent Engine when you're ready for production.

Which agent upgrade you offer?

The Agent Development Kit, or ADK, is an open source framework for developing and deploying AI agents. It is designed for Gemini and Cloud Cloud, but the architecture is model agnostic and deployment agnostic.

In practical terms, ADK offers you:

• An early programming model where behaviors, tools, and orchestrations are included in standard source files
• Workflow agents with secure, parallel, and loop-style control within the agent system
• MARKETING TOOLT A rich ecosystem of built-in tools, customization tools, Openpi tools, Google Cloud tools, and Ecosystem tools
• Deployment methods including local runs, containers, Cloud Run, and Vertex AI Agent Engine
• Built with testing and security patterns, integrated with Vertex AI Agent Builder

For developers, Adk turns the agent into a standard service. You run it locally, check fingerprints, and send it to a managed runtime, instead of running it as a single script that calls LLM.

What does adk for go add?

The GO release maintains the same basic feature set as the Python and Java SDKs but exposes it through the Idiomatic Go API. The Google AI team describes ADK as an idiomatic approach to building agents that use turonyrency and strong typing.

Here are some key points:

• ADK for travel is included go get google.golang.org/adk
• The project is open source and managed at github.com/google/adk-go
• It supports building, testing, and deploying amazing AI agents with flexibility and control
• It uses the same evolution of agents, tools, and workflows as other ADK languages

This means that the GO service can embed the functionality of agents without changing languages. You can create a multi-agent architecture where each agent in the GO Complent component integrates with others using the same framework.

A2A support

ADK for Go ships with native support for the Agent2Agent Protocol, or A2A.

The A2A protocol defines a way for agents to call other agents over a standard display. In the release of the move, Google highlights that the main agent can orchestrate and delegate special tasks to the lower level. Those sub-agents can work locally or as a remote supply. A2A keeps this interface secure and opaque, so the agent does not need to expose internal memory or proprietary logic to participate.

Google also contributed the A2A GO SDK to the A2A core project. That gives GO Protocol developers an entry level if they want agents that interact with other times and frameworks that also support A2A.

MCP Toolbox of data and tools

Key details in Google's official announcement of native integration with the MCP Toolbox for details. It says that ADK GO has out of the box support for more than 30 apps with this toolbox.

MCP Toolbox for Databases is an open source MCP database server. It handles connectivity, authentication, and other concerns, and exposes data operations as tools that use the Model Control System.

Inside ADK, that means:

• You register MCP Toolbox information as an MCP tool provider
• Agent functions call functions through MCP tools rather than creating raw SQL
• The toolbox enforces a safe, predefined set of actions that the agent can perform

This fits the ADK model of tools in general, where agents use a mix of built-in tools, Google cloud tools, ecosystem tools, and MCP tools, all of which are described in the vertex AI Agent build tools.

Integration with Vertex AI Agent Builder and Agent Engine

ADK is the Core Framework supported in Vertex AI Agent Builder for building multi-agent systems.

The latest update koment explains how to build where you:

1. Upgraded local agent using ADK, now including ADK for travel
2. Use adk QuickStart and Dev UI to test the agent with multiple tools
3. Use the agent in the vertex ai agent engine as a managed run time

For team mobility, this means that the language used in services and infrastructure is available throughout the Agent Lifecycle, from local development to managed production deployment.

This is presented in the position of an agent to develop an active bridge between AI agents and existing GO services, using the same open source, original code of tools that run under Python and Java. It brings the A2A Protocol Support and MCP Toolbox of DOPAbases into a native environment, compatible with Vertex AI Agent Builder and Vertex AI Agent Engine for deployment, testing, and monitoring. Overall, this release enables the first-of-its-kind language for building production-ready, interactive ai agents that run on Google's ecosystem.

Look Repo, samples and Technical details. Feel free to take a look at ours GitHub page for tutorials, code and notebooks. Also, feel free to follow us Kind of stubborn and don't forget to join ours 100K + ML Subreddit and sign up Our newsletter. Wait! Do you telegraph? Now you can join us by telegraph.