Gemini Enterprise: Google's Unified AI Agent Platform

The Gemini Enterprise Agent Platform is Google’s new, unified home for building, deploying, and governing AI agents. Google introduced it at Cloud Next 2026 by rebranding Vertex AI and folding in Agentspace, so the tools you used to find scattered across separate products now live in one place. As a result, searching for “Vertex AI” in the Google Cloud Console today redirects you straight to Agent Platform. This guide covers what changed and how to get started, though it’s worth noting upfront: this is a developer and enterprise-tier platform, not a five-minute consumer tool.

Quick Summary: The Gemini Enterprise Agent Platform replaces Vertex AI as Google’s entry point for AI development. It gives you 200+ models to choose from, including Anthropic’s Claude family, plus a no-code builder (Agent Studio) and a full-code framework (Agent Development Kit) depending on how hands-on you want to be.

What Changed in the Gemini Enterprise Agent Platform

Because this is a consolidation rather than a simple rename, a few things moved around. First, Vertex AI’s model training, Model Garden, and agent-building tools all now live inside Agent Platform. Next, Agentspace — Google’s older enterprise search and agent hub — was renamed Gemini Enterprise and now works alongside Agent Platform rather than as a separate product. Finally, the Agent2Agent (A2A) protocol, which lets agents on different platforms talk to each other, moved to production and joined the Linux Foundation’s Agentic AI Foundation alongside Anthropic’s MCP.

Before	Now
Vertex AI (training, Model Garden, agent building)	Folded into Gemini Enterprise Agent Platform
Agentspace	Renamed Gemini Enterprise, runs alongside Agent Platform
Vertex AI Agent Builder	Now the “Agents” section inside Agent Platform
A2A protocol	In production, governed under the Linux Foundation’s AAIF

Warning: If you’re already a Vertex AI customer, you don’t need to migrate anything manually — the underlying services work the same way. However, bookmarks and scripts that reference old Vertex AI URLs will redirect rather than break outright, so it’s worth updating them when convenient.

What You’ll Need to Start

A Google Cloud account — new customers get up to $300 in free credits
Basic cloud platform familiarity if you’re building with code
No coding experience required if you’re sticking to the no-code builder

Step 1: Open the Gemini Enterprise Agent Platform

Step 1Log into the Google Cloud Console and search for Agent Platform, or search “Vertex AI” and let it redirect you. New accounts can also try Express Mode first, which skips the billing setup entirely.

Step 2: Choose Your Build Path

Step 2Two paths exist here. Agent Studio offers a visual, low-code way to design agents and test prompts. The Agent Development Kit (ADK), on the other hand, is the code-first framework for building agents with deeper reasoning and tool use. Many builders start in Agent Studio, then export their logic into ADK once they need finer control.

Step 3: Pick a Model from Model Garden

Step 3Model Garden gives you access to over 200 models. This includes Google’s own Gemini 3.5 family, open models like Gemma, and — notably — Anthropic’s Claude Opus, Sonnet, and Haiku as first-class options rather than a bolted-on integration.

Step 4: Connect Real Data

Step 4An agent without real data is just a demo. So, use Native Ecosystem Integrations to connect to internal systems without custom code, or set up an MCP server if you need a more specific connection to a database, API, or internal tool.

Step 5: Deploy with Agent Runtime

Step 5Once your agent works the way you want, Agent Runtime handles production deployment. It supports sub-second cold starts and long-running agents that can hold state for days, which matters for workflows like an unattended sales-prospecting sequence.

Step 6: Set Up Governance Before You Scale

Step 6Before moving past a prototype, set up Agent Identity so every agent gets a unique, auditable ID. Then use Agent Registry to keep a single source of truth for which agents and tools your organization actually approves.

Tip: If your goal is simply using agents day-to-day rather than building them, the separate Gemini Enterprise app is the easier front door — it’s designed for employees to discover and use approved agents, not to build new ones from scratch.

The Gemini Enterprise Agent Platform Workflow at a Glance

Open Agent Platform → Choose Agent Studio or ADK → Pick a Model from Model Garden → Connect Real Data → Deploy with Agent Runtime → Set Up Governance

How Pricing Works

There’s no flat subscription here. Instead, you pay for what you use across several categories: compute and storage, pipeline execution, Vector Search, and token usage per model. New customers get up to $300 in free credits, which is enough to explore the platform seriously before committing real budget.

Warning: This is genuinely enterprise-tier pricing and complexity. There’s no simple monthly plan like a consumer AI tool offers, so if you’re testing this for a small personal project, budget time to understand the pricing calculator before scaling usage up.

Frequently Asked Questions

Do I need to migrate my existing Vertex AI projects? No. Existing services continue working as before; you’re just accessing them through the renamed platform instead of standalone Vertex AI.

Can I use Claude models here? Yes — Claude Opus, Sonnet, and Haiku are available through Model Garden alongside Gemini and other models.

Is the Gemini Enterprise Agent Platform free? New customers get up to $300 in free credits and can try Express Mode without billing, but ongoing usage is pay-as-you-go rather than free.

What’s the difference between Agent Platform and the Gemini Enterprise app? Agent Platform is where developers build and govern agents. The Gemini Enterprise app, by contrast, is the front end employees use to find and interact with agents that are already built and approved.

Gemini Enterprise: Google’s Unified AI Agent Platform