Gemini Enterprise Agent Platform: The Vertex AI Rebrand Explained

TL;DR

At Google Cloud Next 2026, Google renamed Vertex AI to Gemini Enterprise Agent Platform and folded Agentspace into the same product. The rename matters because it ends a three-product confusion (Vertex AI + Agentspace + Gemini API) and signals that Google is going all-in on agents, not just models. The platform now bundles agent building, deployment, data integration, security, and optimization under a single brand. This post unpacks what actually changed, what is new, what is just renamed, and what it means for buyers picking an agent platform in 2026.

Why A Rebrand Was The Right Move

Vertex AI was a 2021 brand designed to sell "AI Platform" to enterprises that mostly wanted MLOps. By 2025, Google had three overlapping things:

• Vertex AI — the model and MLOps platform.
• Agentspace — the agent layer launched in 2025.
• Gemini API — the developer-facing model API.

The three SKUs created exactly the confusion you would expect. Customers asked: "Do I build my agent in Vertex AI or Agentspace? Is the Gemini API different?" The rebrand collapses all three under Gemini Enterprise Agent Platform and signals one product, one billing surface, one developer experience.

What Is Actually New (Versus Renamed)

A useful split:

New:

• Workspace Studio (visual agent builder inside Google Workspace).
• Pre-integrated partner agents (Box, Workday, Salesforce, ServiceNow, Dun & Bradstreet, S&P Global).
• Gemini 3.1 Ultra (2-million token context) and Gemini 3.1 Flash-Lite.
• A2A protocol donation to the Linux Foundation.
• Tighter security and governance bundled into the platform.

Renamed / consolidated:

• Vertex AI Agent Builder → part of Gemini Enterprise Agent Platform.
• Agentspace UX → folded into the unified product.
• Model garden → Gemini Enterprise Models tab.

Most "new" features are genuinely new. The rebrand is doing real work, not just relabeling.

The Five Bundles Inside Gemini Enterprise

The platform is now sold as five bundled capabilities:

1. Agent building — visual + code, with reusable tools, evals, and prompt management.
2. Deployment — managed serving with autoscaling and observability.
3. Data integration — first-party connectors to enterprise data (BigQuery, Workspace, Drive) and third-party (Box, Workday, Salesforce).
4. Security — IAM, VPC-SC, KMS, audit, plus model-level safety filters.
5. Optimization — caching, distillation, model routing across Ultra/Pro/Flash/Flash-Lite.

The bundle is meaningful because the friction in enterprise AI in 2025 was almost never "the model was not smart enough." It was data integration, security review, and cost optimization. Google is signaling that those are first-class problems, not afterthoughts.

Gemini 3.1 Ultra And 3.1 Flash-Lite

The model layer also moved. Gemini 3.1 Ultra ships with a 2-million token context window and full multimodality (text, image, audio, video). Gemini 3.1 Flash-Lite is the cheap, fast workhorse for high-volume workloads. We deep-dive the implications in our 2M-context post.

The practical buyer move is to use Flash-Lite for the long tail (intake, classification, summarization), Pro for most reasoning, and Ultra for the cases where 2M context or strong multimodality actually changes the answer.

What This Means For Builders

If you were building on Vertex AI in 2025, your code keeps working — Google is doing the right kind of rebrand where the underlying APIs are stable. If you were on Agentspace, you migrate to the unified Gemini Enterprise UX. If you were on the raw Gemini API, you can stay there for individual model calls but you now have a real upgrade path to the platform when you need governance, partner agents, or evals.

What This Means For Buyers

Three buyer takeaways:

• One vendor, one bill, one onboarding. The rebrand removes the buying-committee confusion that Vertex AI + Agentspace + Gemini API generated. Procurement is materially easier.
• Partner agents are a real differentiator. Box, Workday, Salesforce, ServiceNow, Dun & Bradstreet, and S&P Global agents are pre-integrated. If your enterprise runs on those systems, the platform is months ahead at integration time.
• A2A donation is a trust signal. By giving A2A to the Linux Foundation, Google reduced lock-in concerns for buyers who otherwise worry about being captured.

Where CallSphere Lives Next To This

CallSphere is the AI voice and chat front-door — voice/chat/SMS/WhatsApp, 57+ languages, six verticals (healthcare, real estate, sales, salon, IT helpdesk, after-hours), HIPAA-friendly, $49/$99/$149 per month, 3–5 day launch. In a Gemini Enterprise stack, CallSphere captures the inbound conversation, qualifies and books, then hands off to whichever partner agent owns the workflow downstream (Workday for HR, Salesforce for sales, ServiceNow for IT, etc.). CallSphere is not trying to be a horizontal agent platform; it is the front-door that ships this week. See a demo.

Migration Notes For Vertex AI Customers

A pragmatic checklist:

• API stability. Existing Vertex AI SDK calls continue to work. New features are exposed through the Gemini Enterprise SDK, which is additive.
• Billing. Watch the rollover from Vertex AI line items to Gemini Enterprise line items in May–June 2026 billing.
• IAM. Some role names changed in the IAM console — review your service accounts.
• Agentspace assets. Existing Agentspace agents are migrated automatically; review them in the unified UX before relying on the migration.

FAQ

Is the rebrand just marketing? No. The rename is doing real work — consolidating three overlapping SKUs and unifying the developer experience. The new features (Workspace Studio, partner agents, Gemini 3.1) are independent of the rename and would have shipped either way.

Does my Vertex AI code break? Existing APIs are stable. New features ship through additive SDK surfaces.

Is this competitive with ServiceNow + NVIDIA? Different shape. Google is selling a horizontal agent platform; ServiceNow + NVIDIA is selling a workflow-anchored stack. Many enterprises will buy both — and put a focused front-door like CallSphere in front of them.