Google Vertex AI is Google Cloud's machine learning and generative AI platform, now presented under the Agent Platform name. It gives developers one place to build, scale, govern and optimize AI agents grounded in company data. Agent Studio is used to test and tune prompts, Model Garden offers over 200 Google, third-party and open models, and custom training supports your preferred framework. MLOps tools such as Model Evaluation, Pipelines and Model Registry manage models over their lifecycle. New customers get up to $300 in free credits.
Data scientists and ML engineers who train, tune and deploy models are the primary users, and application developers building agents with the Agent Development Kit are another. Enterprise IT and platform teams use governance tools to register and manage agents across an organization. Companies already running on BigQuery gain native notebook integration through Colab Enterprise or Workbench, so data and AI work share a single surface. Individual hobbyists can start with an API key, though Agent Platform is aimed at technical teams.
Common projects include deploying a model for production, extracting, summarizing and classifying data from documents, and building a customer-facing agent that connects to enterprise systems. A team might compare models in Model Garden, evaluate them with objective metrics, then tune the best one and serve it at scale. Researchers train custom models with open source frameworks on optimized infrastructure, while product teams prototype prompts using text, code, images or video. Pipelines automate retraining, and a feature store helps teams share and reuse ML features across projects.
