Google Cloud has revealed a solution for the creation and deployment of autonomous AI agents on an enterprise scale, and it has made all demo code presented at the event available as open source.
The Gemini Enterprise Agent Platform, which was launched at Google Cloud Next 2026, represents the company’s most organised effort to transition AI agents from experimental prototypes to operational environments.
This platform includes the full lifecycle of agent development, featuring an Agent Development Kit (ADK) for constructing agents and a serverless agent runtime for deploying and scaling them without the need for infrastructure management.
Brad Calder, President of Site Reliability Engineering at Google Cloud, stated that the platform is designed to assist developers in creating agents that proactively assist users and autonomously complete tasks.
In addition to development and runtime capabilities, the platform introduces governance infrastructure, which includes a distinct agent identity for each agent instance, an agent gateway for enforcing IAM policies, an agent registry for discovery, and an A2A protocol for collaboration between agents.
Notably, the platform helps models beyond Gemini Pro and Flash, including Anthropic’s Claude, via Google Cloud’s Model Garden.
Wiz Co-Founder Yinon Costica showcased how the platform integrates with cloud security tools through a pair of specialized agents. A “Wiz Red Agent” detected an authentication bypass vulnerability that extended from an internet-exposed entry point to sensitive data.
Later on, a “Wiz Green Agent” proposed prioritised fixes, which included downgrading IAM privileges, patching the bypass, and implementing AI guardrails. Claude Code executed the changes, and Wiz performed a rescan to verify the resolution.
The move highlighted that AI agents can not only manage task execution but also can perform security remediation within the same platform.
Read more: Google launches open source Gemma 4 AI model for developers
