Google launches Gemini Enterprise Agent Platform with 200+ models, MCP support for multi-agent workflows

Google released its Gemini Enterprise Agent Platform on Wednesday at Cloud Next, the company's annual enterprise conference. The platform, evolved from Vertex AI, consolidates model selection and tuning services with new capabilities for agent integration, security, DevOps, and orchestration.

Platform specifications

The Agent Platform provides access to over 200 models, including:

• Gemini 3.1 Pro
• Nano Banana 2
• Gemma open models
• Third-party models from Anthropic, including the newly released Opus 4.7

According to Google CEO Thomas Kurian, the platform "brings together the model selection, model building, and tuning services of Vertex AI that customers love, along with new features for agent integration, security, DevOps, orchestration, and more." Google confirmed that Vertex AI services will now flow through Agent Platform exclusively.

Agent development capabilities

The platform includes MCP (Model Context Protocol) support and an upgraded Agent Development Kit designed to structure agents into sub-networks for handling complex tasks. Google claims new features like faster runtime and Memory Bank enable agents to delegate to each other more efficiently and operate with extended context.

Developers can use Agent Simulation to test agents against real-world scenarios before deployment. Once built, agents can be published to the Gemini Enterprise app, where employees can run them or build their own using no-code tools like Agent Studio and Agent Designer.

Security and governance

The platform implements Agent Identity, assigning each agent a cryptographic ID for tracking and security. Google states the system "provides the same level of oversight and auditability found in essential business applications like payroll or quarterly financial reporting."

According to the company, Agent Platform standardizes governance and security through "a single control plane for governance," ensuring "every employee can use and share agents with full IT visibility." Both no-code and pro-code agents are managed through a consistent model for identity, security, and auditing.

Additional announcements

Google also launched Agentic Data Cloud, a data architecture designed to enable cross-cloud queries (AWS, Azure) without data movement, and Workspace Intelligence, which uses Gemini reasoning to understand semantic relationships across Docs, Slides, Gmail, and other Workspace apps. The company claims Workspace Intelligence can generate content that mimics a user's "exact voice, brand, style, and company templates."

What this means

Google is positioning itself as the enterprise platform for multi-agent workflows, directly competing with Microsoft's Copilot ecosystem and Anthropic's enterprise offerings. The inclusion of third-party models like Anthropic's Opus 4.7 signals a strategic shift toward platform openness, though the consolidation of Vertex AI into Agent Platform forces existing customers onto the new system. The emphasis on cryptographic agent identities and governance addresses legitimate enterprise concerns about autonomous agent security, though real-world effectiveness remains to be demonstrated. Pricing for the Agent Platform was not disclosed.