Google launches Gemini Enterprise Agent Platform to run autonomous, multi‑step enterprise workflows

Google Cloud on April 22 launched Gemini Enterprise Agent Platform, a new system for building and running autonomous business agents. (cloud.google.com) The company described it as the evolution of Vertex AI, bundling model selection, model building and agent development with new tools for integration, DevOps, orchestration and security. (cloud.google.com) (blog.google) Google announced the platform at Cloud Next ’26 and said it is aimed at teams that need agents to execute complex business workflows, not just answer prompts. (blog.google 1) (blog.google 2) In practice, that means software that can take a task, call other systems, hand work to sub-agents and keep running for hours or days. Google said enterprises now need agents with their own identity, registry and gateway so they can be traced and managed. (cloud.google.com) Google’s new stack includes Agent Studio for low-code building, Agent Development Kit for coding multi-agent systems, Agent Engine for deployment, and built-in monitoring and evaluation tools. (blog.google) (cloud.google.com) The platform also gives developers access to Gemini 3.1 Pro, Gemini 3.1 Flash Image and Lyria 3 through the same environment, according to Google’s launch materials. (blog.google) Google tied the product to a broader portfolio shift. A 2025 Google Agentspace post was later updated to say its agent creation and orchestration technology now powers core Gemini Enterprise functions. (blog.google) Google also used Cloud Next to showcase customers including Lowe’s, Target, Verizon, Wayfair and Wendy’s building agentic systems on Google Cloud. (blog.google) The pitch is less about a single chatbot and more about a managed fleet of software workers. Google’s own wording at Cloud Next framed the problem as moving from “Can we build an agent?” to “How do we manage thousands of them?” (blog.google)