Google Reorganizes Vertex AI

Google has reorganized Vertex AI into the Gemini Enterprise Agent Platform, unveiling a platform for building and operating agents for enterprise use.

The platform brings together model selection, model building, and agent building capabilities, adding integration, DevOps, orchestration, and security features on top. Going forward, Vertex AI's services and roadmap will be delivered through this platform rather than as a standalone product.

The core components are as follows.

• Agent Studio / ADK: Supports both low-code and code-first development
• Agent Runtime: Supports long-running agents and state persistence
• Memory Bank: Maintains long-term context and memory
• Agent Identity / Registry / Gateway: Identity, approved tool management, and access control
• Simulation / Evaluation / Observability: Execution tracing and quality verification

On the model side, the platform offers 200+ models through Model Garden, supporting Google's Gemini 3.1 Pro, Gemini 3.1 Flash Image, Lyria 3, and Gemma 4, as well as third-party models such as Claude Opus, Sonnet, and Haiku.

Through this platform, Google emphasized the direction of connecting agents to employee-facing apps and scaling them into production environments within enterprise systems and security/governance frameworks.