Build reliable multi-agent applications with ADK Go 2.0 (developers.googleblog.com)

The Agent Development Kit (ADK) for Go has now launched its 2.0 version, introducing a transformative graph-based workflow engine designed for building reliable multi-agent applications. This update allows developers to structure complex orchestration processes into graphs of connected nodes, simplifying the execution of multi-step workflows that involve branching, human approval, and dynamic data handling. Key features include built-in human-in-the-loop (HITL) functionality, the ability to persist states across executions, and support for various agent modes, all while maintaining Go's idiomatic style. The significance of ADK 2.0 lies in its capacity to streamline the development of advanced applications that require flexibility and resilience in their workflows. By enabling a unified execution model for both single agents and complete graphs, the toolkit enhances reliability and observability in AI-driven applications. Developers can use simplified constructs to create intricate control flows with built-in retry policies, state management, and customizable routing, all of which supports complex interactions in real-world scenarios. This evolution positions ADK 2.0 as a powerful resource for Go developers venturing into the AI/ML domain.