Agentic AI With Google Cloud: Notes From Build with AI

I attended Google's Build with AI Roadshow event last week and it was truly an immersive experience centered on multi-agent systems on Google Cloud. The event focused on building, exploring, and collaborating on next-generation intelligent applications.

Workshop Highlights

The cornerstone of the event was the hands-on workshop, "Agentic AI with Google Cloud," which guided participants through developing a multi-agent system. We gained practical experience building an AI-powered teaching assistant using Google Cloud's comprehensive toolkit. The exercise deck is embedded below.

Key Learnings

• Orchestrating Multi-Agent Interactions: We learned techniques for managing complex interactions between multiple AI agents working together.
• Development Workflow Optimization: The lab provided valuable insights into streamlining AI application development processes.
• Agent Customization: We practiced building and tailoring AI agents for specific functionalities and requirements.
• Collaborative AI Systems: The workshop demonstrated how to create systems where multiple AI agents work together to generate lesson plans, quizzes, and audio recaps.

Technology Stack

We leveraged various Google Cloud services, including:

• Vertex AI for accessing Gemini models
• Cloud Run for deployment
• Cloud SQL for data management
• Pub/Sub and Eventarc for inter-agent communication

The LangGraph framework was particularly useful for defining the logic and flow between different AI agents, enabling intelligent collaboration.

Special thanks to the Google Cloud and Google Developers Group teams as well as Christina Lin and Karl Weinmeister.

Download the workshop deck (Build with AI Demo), PDF, 45 pages

---

Originally posted on LinkedIn.