Vertex AI is Google Cloud’s enterprise platform for building, deploying and managing AI. At Google Cloud Next ’26 (April 22, 2026), Google folded it into the Gemini Enterprise Agent Platform, which Google describes as an evolution of Vertex AI. The capabilities in this guide still apply. Most of the product names have changed.
Is Vertex AI now called Gemini Enterprise Agent Platform?
In Google’s own documentation, yes. The April 22, 2026 release notes state that Vertex AI is now part of Gemini Enterprise Agent Platform, and a Google Cloud roundup says all Vertex AI services and roadmap evolutions will be delivered through the Agent Platform rather than as a standalone service. You will still see “Vertex AI” in older documentation, code samples and search results, so here is how the names map, based on Google’s name-changes page:
| Former name | Current name |
|---|---|
| Vertex AI | Gemini Enterprise Agent Platform (“Agent Platform”) |
| Vertex AI Studio | Agent Studio |
| Vertex AI Model Garden | Model Garden on Agent Platform |
| Vertex AI Pipelines | Pipelines on Agent Platform |
| Vertex AI Agent Engine | Agent Runtime |
| Vertex AI Search | Agent Search |
| Vertex AI Vector Search | Vector Search on Agent Platform |
| Vertex AI Feature Store | Feature Store on Agent Platform |
| Vertex AI Experiments | Experiments on Agent Platform |
The API has a new name too. Google’s release notes list the “Vertex AI API” becoming the “Agent Platform API,” and the Apps Script documentation now describes the Agent Platform API as “formerly the Vertex AI API,” even though Apps Script’s built-in service is still called the Vertex AI service. Expect old and new names side by side in docs and code for a while.
Google’s pages list these renames. In the pages we reviewed, they do not say that existing projects must migrate. If you run production workloads on Vertex AI, confirm with your Google account team or with us before you change anything.
One thing not to mix up: the Gemini Enterprise app is the AI front door for employees, while the Agent Platform is the developer platform behind custom agents and models. Our Gemini Enterprise explainer covers the app side.
What can you do with it? Build, scale, govern and optimize
Google organizes the platform around four pillars, according to its platform overview:
- Build: Agent Development Kit (a model-agnostic framework), Agent Studio (a low-code visual canvas), Agent Garden (prebuilt agents and templates), Model Garden (Google, third-party and open-source models), RAG Engine and Vector Search.
- Scale: Agent Runtime for deploying agents, plus Sessions, Memory Bank and sandboxed Code Execution.
- Govern: Agent Registry (a catalog of agents, tools and MCP servers), Agent Identity, Agent Gateway and governance policies.
- Optimize: agent evaluation and simulated testing, observability (monitoring, logging and tracing) and prompt optimization.
The platform also keeps the model building and tuning services that made up Vertex AI. At launch Google listed Gemini 3.1 Pro, Gemini 3.1 Flash Image and Lyria 3, and the platform also supports Anthropic’s Claude models. Model lineups change often, so check our Gemini models guide before you pick one.
Why should business leaders care?
Most companies get stuck in “AI pilot purgatory”: they test tools but cannot move them into production. The platform is built to close that gap with one governed environment for the full lifecycle, from data prep and model tuning to deployment, monitoring and access control. That means less time wrestling with infrastructure and more time on the use case.
It also helps with the other two blockers leaders raise most often. Shared tooling reduces the data and team silos between business, engineering and analytics. Built-in governance (identity, policies and logging for agents) gives security teams something concrete to approve.
How does it handle data security?
Security questions come first in most enterprise AI conversations. Here is what Google documents today:
- No training on your data without permission. Google’s data retention page says it won’t use your data to train or fine-tune AI or ML models without your prior permission or instruction.
- Caching is on by default for Google’s Gemini models to reduce latency. Customers who need zero data retention have to turn specific features off, so review those settings with your security team.
- Data residency. Regional and multi-regional endpoints keep cached data in the selected location. The global endpoint can serve requests from any region.
- Enterprise controls. Google’s security controls page lists data residency, customer-managed encryption keys (CMEK), VPC Service Controls and Access Transparency for the core platform. Support varies by service and preview status, so check the table for the features you plan to use.
- Compliance. Google lists its generative AI services on the platform as FedRAMP High authorized. Scope can differ by feature and region, so confirm against Google’s compliance pages for your use case.
What are real-world use cases?
AI only matters if it solves a business problem. These examples come from Google-published customer stories (not Premier Cloud clients) and from our own work:
- Retail and e-commerce: Google reports that Mercado Libre’s rollout of Vertex AI Search for product discovery is already delivering millions of dollars in incremental revenue (Google Cloud Next ’25 recap).
- MLOps: Wayfair used Vertex AI Pipelines to build a CI/CD process for model maintenance and cut hyperparameter tuning from two weeks to under an hour (Google Cloud blog).
- Customer service: teams pair the platform with Google’s contact center tools to build virtual agents that understand intent and escalate to people. See our guide to Google Contact Center AI.
- Gaming: Premier Cloud helped Kidas make its gaming text-chat processing generic, so a new game launch does not need code changes. Read the Kidas case study.
How can your organization take its first steps?
- Pick one high-value use case. Start with something measurable, like shorter support wait times or better demand forecasting.
- Run a proof of concept. Use Model Garden and Agent Studio to prototype quickly. Focus on outcomes, not perfection.
- Plan for scale early. Decide who owns access, logging and evaluation before the first agent goes live.
- Align the teams. AI adoption needs business and technical owners at the same table.
- Get expert support. Premier Cloud has guided organizations from workshops to production-grade solutions.
How Premier Cloud helps
Premier Cloud is a Google Cloud Premier Partner with deep AI expertise. We build alongside your team:
- Workshops and enablement: hands-on sessions that upskill your team quickly.
- Proofs of concept: tailored POCs that prove value in weeks, not months.
- Production deployments: secure, scalable rollouts that connect AI to your existing systems.
- Strategic guidance: help choosing the right use cases, managing risk and measuring ROI.
Frequently asked questions
Is Vertex AI discontinued?
Not as a set of capabilities. Google has made Vertex AI part of the Gemini Enterprise Agent Platform and says its services and roadmap now arrive through that platform. The Vertex AI name is what has been retired.
Do I need to migrate my Vertex AI projects?
Google’s release notes and name-changes page list renames but, in the pages we reviewed, do not describe a required migration. Test any change in a non-production project and check with your Google account team or Premier Cloud first.
What is the difference between Gemini Enterprise and the Agent Platform?
Gemini Enterprise is the app employees use to find information and run agents. The Agent Platform is where developers build, deploy and govern custom agents and models. They are designed to work together. Our Gemini Enterprise FAQ goes deeper on the app.
Is my data used to train Google’s models?
Google states it will not use your data to train or fine-tune models without your prior permission or instruction. Review caching and retention settings with your security team.
How much does it cost?
Pricing depends on the models and services you use. We can scope a pilot and estimate usage before you commit.
- Gemini Enterprise Agent Platform release notes (April 22, 2026)
- Gemini Enterprise Agent Platform name changes (updated Oct 26, 2026)
- Agent Platform overview
- Apps Script: Vertex AI service (Agent Platform API, formerly the Vertex AI API)
- Google: Gemini Enterprise Agent Platform announcement (April 22, 2026)
- Google Cloud: what Google Cloud announced in AI this month
- Agent Platform and zero data retention
