Google Cloud has announced a set of new generally available capabilities for its Gemini Enterprise Agent Platform, including Agent Memory Bank, Agent Runtime, Agent Identity, Agent Gateway, Agent Registry, and integrated evaluation and observability tools. Additionally, CodeMender, a managed code security agent, was announced to advance from passive scanning to automated code remediation and reduce zero-day risk.

Agent Memory Bank enables low-latency personalization by defining structured schemas that automatically extract and maintain conversation context, allowing agents to retain user preferences, past decisions, and account history across long-running tasks. Agent Runtime supports complex multi-day agents that can run continuously for up to seven days, enabling asynchronous workflows like week-long sales sequences or multi-stage onboarding without constant human intervention.

On the security and governance side, Agent Identity is a new IAM type built on open standards that enforces least-privilege permissions, mitigates token theft by binding access to the agent runtime, provides non-repudiable auditing, and automatically manages the identity lifecycle to eliminate dormant credentials. Agent Gateway provides a central control point to secure and govern interactions across the agent ecosystem, with granular access controls via IAM conditions and natural language rules, and inline protection against prompt injection, tool poisoning, and data leakage. Agent Registry serves as a single library for all AI agents, servers, and connections across an organization, helping teams find and reuse agents and allowing administrators to monitor agent sprawl.

For performance optimization, the platform now offers agent evaluation with online evaluation monitors that proactively detect performance degradation and behavioral drift, supporting pre-built metrics, custom Python, LLM-as-a-judge, and adaptive rubrics from Google DeepMind. Agent observability provides end-to-end visibility into agent reasoning, tool utilization, and execution performance through comprehensive tracing and real-time dashboards.

Customer quotes from AT&T, Best Buy, Commerzbank, Liberty Global, and WellSky highlight production use of these features. AT&T uses Agent Memory Bank for long-term memory in sales agents. Best Buy applies Agent Identity for accountability. Commerzbank evaluates Agent Registry and Gateway for governance. Liberty Global uses Gateway and Registry for multi-entity security. WellSky implements Agent Registry for compliance and lifecycle management.

The features are available now. Google Cloud provides documentation and a livestream walkthrough for the agentic lifecycle.