From Prototype to Production: Building with Gemini Enterprise Agent Platform

Is your AI agent ready for the real world? Building a prototype is easy, but moving an agent into a production environment requires more than just a good prompt. Join Addy Osmani, Dave Elliott, and Jason Davenport live from Google Cloud Next '26 for an inside look at the Gemini Enterprise Agent Platform. In this session, the team breaks down the four pillars of a professional agentic workflow: Build, Scale, Govern, and Optimize. What You’ll Learn: The Agent Development Kit (ADK): How to use the framework (supporting Python, Go, TypeScript, and Java) to move from "zero to one" in minutes. Identity & Governance: Why giving your agents "cryptographically generated identities" is the secret to secure, auditable enterprise workflows. Memory Bank & Long-Running Agents: How to build agents that operate over days or weeks without losing context or "forgetting" their mission. Observability & Evaluation: Using agent tracing and sandboxes to limit the "blast radius" and ensure non-deterministic models stay on track. The Future of Devs: Why the rise of AI agents isn't replacing developers, but moving them toward higher levels of abstraction and problem-solving. Ready to build? Check out the new Agent Platform repo: https://cloud.google.com/products/gemini-enterprise-agent-platform?e=48754805 and explore the AI Agent Challenge for Startups: https://devpost.team/hackathon_guest_invites/4fb181b4-2722-415d-a442-285a57dcaba5 #googlecloudnext #Gemini #GeminiAI #AIAgents #SoftwareEngineering #GenerativeAI #GoogleCloud