Plan before you build: Deterministic planning patterns for AI agents

This session covers a production-ready architecture for scalable AI agents utilizing the LLM Council pattern, Google’s Agent Development Kit (ADK) for Java, and Spring AI on Cloud Run. The discussion compares probabilistic reason-and-act (ReAct) loops with goal-oriented action planning (GOAP) via LLM Council, demonstrating how this approach creates more efficient, performant agents optimized for serverless execution.