Google ADK 2.0 pushes agents toward workflow and task runtimes
Google's Agent Development Kit 2.0 frames production agent apps around graph workflows, structured task delegation, sessions, human-in-the-loop checkpoints, and deployable code-first agents.
Why this signal matters
Google's ADK 2.0 README describes an open-source, code-first framework for building, evaluating, and deploying AI agents. The 2.0 framing is notable because it moves beyond a single Agent class into a graph-based workflow runtime with routing, fan-out/fan-in, loops, retry, state management, dynamic nodes, human-in-the-loop, and nested workflows. It also introduces a Task API for structured agent-to-agent delegation, including multi-turn task mode, controlled single-turn output, mixed delegation patterns, task agents as workflow nodes, and session-schema compatibility concerns. For agent builders, the practical signal is to evaluate frameworks by how explicitly they model workflows, delegation, sessions, retries, local inspection, and deployment, not only by how quickly they can run a chatbot demo.
Actionable summary
When comparing agent frameworks, inspect workflow graphs, task delegation semantics, session compatibility, retry/state controls, human-in-the-loop support, local tooling, and deployment paths.
- Agent usefulness
- 91/100
- Confidence
- 83%
- Canonical data
- JSON + Markdown