What Is Vertex AI? Google's AI Platform Explained

Vertex AI is the name many people still use for Google Cloud's platform for building with AI, and you'll see it in agency proposals, job ads and developer chatter. The name has moved on, though, which confuses a lot of people. Here's the plain version, with the rename explained, for anyone who wants to know what's behind the AI features in the software they use.

Vertex AI in one clear explanation

Vertex AI is Google Cloud's managed platform for building, running and managing AI and machine learning projects. Think of it as a workshop on Google's cloud. It holds the models, the tools to adapt them and the machinery to run them, so a technical team doesn't have to assemble all of that from scratch.

It's aimed at developers and data teams rather than at someone who wants to chat to an assistant. The chat assistants you may know, such as Gemini, are products. Vertex AI is the place where businesses build their own products using models like them.

Vertex AI and its new name

Google's own release notes say that Vertex AI is now part of Gemini Enterprise Agent Platform. They list the changes: Vertex AI Studio became Agent Studio, the Vertex AI API became the Agent Platform API, and Vertex AI Agent Engine became Agent Runtime.

So if a supplier says they build on Vertex AI, they usually mean this same platform under its older name. Google's documentation now uses the new name, but you'll still find the old one in plenty of tutorials, job descriptions and older articles. When you search the official docs, try both names.

What you can do on the Vertex AI platform

Going by Google's documentation, the platform covers several areas:

• Model Garden. A catalogue of models, including Google's own, partner models and open models.
• Gemini models. Google's family of models for text, images, audio and more.
• Agents. Tools to build AI helpers that carry out multi-step jobs.
• AutoML. Training models on your images, tables or forecasting data with less hand-coding.
• Custom training. For teams that want to build and train their own models.
• MLOps tools. Pipelines, a feature store, a model registry and monitoring to keep models running reliably.

That's a lot of surface area. Most small businesses will meet it indirectly, through an agency or an app that has been built on top of it. If you're weighing up an agent project, my guide to AI agents and automation for small businesses is a good place to get your bearings.

Vertex AI and your data

The question to ask before anything else is what happens to your information. Google's data governance page for the platform states that Google won't use your data to train or fine-tune any AI or machine learning models without your prior permission or instruction.

The same page covers a few details worth knowing:

• For supported Gemini models, cached data is deleted no later than 24 hours after it's written, and caching can be switched off.
• For Google models, Google may log prompts to detect abuse and policy violations. Eligible customers can ask for an exception to get zero data retention.
• Some features keep data for set periods, such as grounding with Google Search or Maps, and Google lists those on the page.

Rules differ by feature and by agreement, so read the data governance page for the feature you're using rather than relying on a summary.

How to get started with Vertex AI

You'll need a Google Cloud account. After that, the route Google documents is to open the platform in the Google Cloud console, pick a model from Model Garden or try a Gemini model in the studio, and follow the quickstarts and developer guides. Names in the console are shifting with the rebrand, so use the current documentation rather than older tutorials.

Honestly, a developer should do the real build. If that's not you, a short conversation with an AI consultant can save you from buying more platform than you need.

Is Vertex AI right for a small business?

It's a good fit if you have a specific, repeatable job and a technical person or agency to build it, for example a support assistant that answers from your product guides, or a tool that sorts incoming documents. It's overkill if you just want to write emails faster. For that, a ready-made assistant is simpler, and my list of the best AI tools for small business owners covers options that need no set-up.

Questions to ask a supplier who builds on Vertex AI

If a supplier proposes Vertex AI, or Gemini Enterprise Agent Platform as it's now called, these plain questions help you judge the plan:

• Which model will it use, and who decides when to switch?
• Whose Google Cloud account will hold the project, and who can see the data in it?
• Which features need prompts to be logged, and have you read Google's data governance page for them?
• How will answers be tested before customers see them?
• Who looks after it when the platform's names and menus change?

Vertex AI compared with Amazon Bedrock

Amazon's equivalent is Bedrock, a managed service that gives developers access to many foundation models. Both give businesses a managed way to use big AI models inside a cloud they already run on. The sensible deciding factor for most companies is simply which cloud their data and apps already live in.

Frequently asked questions

What is Vertex AI?

It's Google Cloud's managed platform for building and running AI and machine learning projects, with access to models, tools to adapt them and ways to deploy them.

Has Vertex AI been renamed?

Google's release notes say Vertex AI is now part of Gemini Enterprise Agent Platform, with several features renamed, such as Vertex AI Studio becoming Agent Studio.

Is Vertex AI the same as Gemini?

No. Gemini is Google's family of AI models. Vertex AI is the platform where businesses can use Gemini and other models to build their own applications.

Does Vertex AI use my data to train models?

Google's data governance page says it won't use your data to train or fine-tune any AI or machine learning models without your prior permission or instruction. Check the page for each feature you use.

Do I need to code to use Vertex AI?

For anything beyond experiments, usually yes. Some parts, such as AutoML, reduce the coding, but a technical person should plan and review the build.