Google Cloud Next 2026: agent registry, gateway and identity become cloud services

Google Cloud Next 2026 ended with a list of 260 announcements, most of them about AI agents. For integration teams, the most important ones are not the new models but a set of governance services: a registry of agents and tools, a gateway for agent traffic and separate identities for agents.

These are the same building blocks the iPaaS vendors have been adding to their platforms over the past year. Now one of the big cloud providers offers them as part of its own platform.

What Google launched

Vertex AI has become the Gemini Enterprise Agent Platform, which Google describes as a platform for building, scaling, governing and optimising agents. Several of the new services are about governance:

• Agent Registry indexes the organisation’s agents, tools and skills, and keeps a library of approved tools so that only governed assets are available.
• Agent Gateway is one control point for connections between agents and between agents and tools. According to SiliconANGLE, it understands both MCP and A2A, and it applies security policies and Model Armor protection against prompt injection and data leakage.
• Agent Identity gives each agent its own identity with limited, delegated rights.
• Partnerships with security and identity vendors, including Okta, Ping Identity, Palo Alto Networks and Zscaler, let customers connect their own tools to Agent Gateway. This is in preview.

Google also announced a number of managed MCP servers, including for databases, Cloud Storage and Workspace, and support for connecting third-party MCP servers to Gemini Enterprise.

Why it matters

A registry, a gateway and identity for agents are the core of agent governance. They answer the questions that come up as soon as an organisation has more than a handful of agents: what exists, who owns it, what may it access and what did it do?

When a cloud provider offers these as standard services, the question for iPaaS buyers changes. It is no longer only which integration platform has the best agent governance, but where governance should live when agents run both in the cloud provider’s platform and in the integration platform.

How it compares

The iPaaS and API management vendors are building their own registries and gateways for agents and MCP servers, and the other cloud providers are doing the same. The protocols are the same, MCP and A2A, but each platform has its own registry and its own rules.

For most large organisations, the realistic outcome is several registries and several gateways. That means the ability to exchange information between them, or to scan other platforms for agents, will matter more than which platform has the most features.

What to ask

1. Where is the master registry? Decide which registry is the main source of truth for agents and MCP servers, and how others are kept in sync.
2. Identity across platforms. Check whether an agent’s identity in Google Cloud can be understood and limited when it calls an agent or API in another platform.
3. Gateway placement. Map which traffic goes through which gateway, so that no agent traffic bypasses policies and logging.