Agent Compute Engine and expanded agent governance (cloud.google.com)

Google Cloud has announced the launch of the Agent Compute Engine and expanded governance capabilities for autonomous agents during the AGNTCon + MCPCon event. The Agent Compute Engine is a specialized compute service designed to manage the unique demands of running large-scale AI agents, known as Gemini, enabling high-performance, secure, and efficient execution of tasks across multiple concurrent sessions. It features a microVM architecture that optimizes speed and resource use while enhancing security through hardware isolation and cryptographic identities, allowing businesses to deploy custom agents seamlessly. This development is significant for the AI/ML community as it addresses the increasing complexity and scale of autonomous agent operations. The expanded governance capabilities include tools like the Agent Gateway for managing traffic and security policies, as well as the Agent Registry for tracking agent identities and interactions. By implementing zero-trust protocols and advanced observability features, Google aims to mitigate risks associated with untrusted code and enhance the management of agent fleets across diverse environments—making it easier for organizations to integrate AI agents into their workflows while maintaining strict security and operational controls.