Google Cloud launches Gemini Enterprise Agent Platform at Next 2026
Brad Calder describes the platform as a tool for building proactive autonomous agents
Google Cloud has unveiled the Gemini Enterprise Agent Platform, a comprehensive end-to-end solution designed to build, deploy, and govern autonomous AI agents at an enterprise scale.
Introduced at Google Cloud Next 2026, the platform marks a significant move to transition AI agents from experimental prototypes into robust production environments.
Brad Calder, Google Cloud President of Site Reliability Engineering, stated the platform is engineered to help developers create agents that "proactively help users and complete tasks independently."
The architecture covers the entire development lifecycle, featuring an Agent Development Kit (ADK) and a serverless runtime that allows for seamless scaling without manual infrastructure management.
To ensure enterprise-grade control, Google introduced structural governance tools, including unique Agent Identities, an Agent Registry for discovery, and an Agent Gateway to enforce strict access policies.
While anchored by Gemini 3.1 Pro and Flash, the platform remains model-agnostic, supporting third-party options such as Anthropic’s Claude via the Model Garden.
A highlight of the launch was a complex multi-agent simulation for planning a Las Vegas marathon, which Google has fully open-sourced.
This demonstration featured coordinated agents managing logistics and safety, while integration with Wiz showcased the platform's security prowess.
The "Wiz Red Agent" identified vulnerabilities, while the "Wiz Green Agent" applied automated fixes through Claude Code.
This seamless remediation illustrates a future where AI handles both task execution and active security within a single, governed stack.
Google has made the entire codebase available to lower the barrier for engineering teams.