TLDR: Google has officially announced expanded support for the Go programming language within its Agent Development Kit (ADK), enabling developers to build powerful and production-ready AI agents and multi-agent systems using their preferred languages. The ADK, an open-source, flexible, and model-agnostic framework, aims to streamline the creation, deployment, and orchestration of sophisticated AI applications.
Google has significantly advanced the landscape of artificial intelligence development with the announcement of enhanced support for the Go programming language within its Agent Development Kit (ADK). This move empowers a broader range of developers to leverage their existing Go expertise to construct sophisticated AI agents and multi-agent systems, making agent development feel more akin to traditional software engineering rather than experimental research.
The Agent Development Kit, initially introduced at Google Cloud NEXT 2025 in April, is an open-source, flexible, and modular framework designed to simplify the end-to-end development lifecycle of AI agents. It is built to facilitate the creation, deployment, and orchestration of agentic architectures, ranging from simple task automation to complex workflows.
ADK is distinguished by its model-agnostic nature, allowing developers to integrate various large language models (LLMs) beyond Google’s own Gemini. It supports models accessible via Vertex AI Model Garden and offers LiteLLM integration, providing access to a wide selection of models from providers like Anthropic, Meta, Mistral AI, and AI21 Labs. Furthermore, the kit is deployment-agnostic, enabling agents to be deployed across diverse environments, from local machines to Vertex AI Agent Engine and custom cloud infrastructures, though it is optimized for seamless integration within the Google Cloud ecosystem.
Key features of the ADK include its multi-agent by design approach, which allows for the creation of modular and scalable applications by composing multiple specialized agents in hierarchical structures, facilitating complex coordination and delegation. Developers can equip agents with a rich ecosystem of tools, including pre-built options like Google Search and code execution sandboxes, custom Python functions, and integrations with third-party libraries such as LangChain, LlamaIndex, and CrewAI. The framework also supports the Model Context Protocol (MCP) for standardized tool integration and even allows other agents to be used as tools, enabling powerful delegation patterns.
For enhanced user interaction, ADK incorporates built-in bidirectional audio and video streaming capabilities, enabling human-like conversations with agents. Recognizing the importance of robust applications, the ADK also addresses security and safety as first-class concerns, providing mechanisms for implementing controls such as tool confirmation flows, authentication systems for protected resources, and execution boundaries to limit agent actions.
Also Read:
- Google Cloud Enhances Vertex AI Agent Builder with New Observability and Deployment Tools
- Google’s Gemini Set to Replace Assistant Across Android Devices in Major AI Shift
Internally, Google has been utilizing the ADK to power agents within its own products, including Agentspace and the Google Customer Engagement Suite (CES), underscoring its proven capabilities and reliability. With the expanded Go support, alongside existing Python and Java capabilities, Google aims to provide developers with powerful, flexible tools to build the next generation of production-ready AI applications in the rapidly evolving agent landscape.
