Plan before you build: Deterministic planning patterns for AI agents

Details

Can your agents scale without bankrupting your API quota or your compute budget? Join this session to explore a production-ready architecture that uses the LLM Council pattern, Google’s Agent Development Kit (ADK) for Java and Spring AI, all deployed on Cloud Run. And discover how replacing probabilistic reason-and-act (ReAct) loops with LLM Council enable goal-oriented action planning (GOAP) creates agents that are not only smarter but also significantly lighter and faster – making them the perfect workload for serverless execution.

**About the venue** Free Times Cafe has bistro-style seating and a full food and drink menu. Please consider helping to support the venue by planning to have supper during the talk.

**Speaker Bio** Dan Dobrin is an App Architect in Google Cloud, helping customer teams adopt modern AI technologies, cloud-native patterns, practices and technologies in GCP.

Dan focuses on building GenAI applications, adopting AI coding assistance tooling with Gemini, Native Java Images, Spring AI, event-driven architectures, production-readiness and performance optimization.

Prior to joining Google Cloud, he has built highly scalable, low-latency, frameworks for technology, security and financial services organizations.