Google unveils Gemini platform and Agent Studio at Cloud Next, framing an 'agentic era'

Google used Cloud Next ’26 to unveil Gemini Enterprise Agent Platform and Agent Studio, its new centerpiece for building and managing AI agents in business software. (cloud.google.com) An AI agent is software that can plan steps, call tools, and hand work to other systems instead of just answering one prompt. Google said its new platform is the evolution of Vertex AI, with added orchestration, DevOps, security, and governance. (cloud.google.com) (blog.google) Agent Studio is the low-code front end for that system, while the Agent Development Kit is the code-first path for developers building larger workflows. Google’s documentation says Agent Studio covers model discovery, prompt engineering, and agent building inside the platform. (cloud.google.com) (docs.cloud.google.com) Google also tied the launch to its Agent2Agent, or A2A, protocol, which is meant to let agents built by different vendors discover each other and delegate work securely. Google’s docs describe A2A as a universal language for agents, and the company has been pushing version 0.3 of the protocol for enterprise adoption. (docs.cloud.google.com) (cloud.google.com) The company’s message at Next was that customers are no longer asking whether they can build one agent, but how to manage thousands of them. Sundar Pichai said Google built the new platform as “mission control” for that shift inside large organizations. (blog.google) Google said the platform gives customers access to Gemini 3.1 Pro, Gemini 3.1 Flash Image, also called Nano Banana 2, and Lyria 3. The company positioned that model lineup as part of a single managed stack rather than a set of separate tools. (blog.google) (cloud.google.com) Cloud Next ’26 was broader than software agents alone. Google also used the event to announce eighth-generation Tensor Processing Units and to frame what it called the “Agentic Enterprise” across data, security, and productivity products. (blog.google) (cloud.google.com) Google’s own examples leaned heavily on operations and security. Pichai said Security Operations Center agents already triage tens of thousands of unstructured threat reports each month and cut mitigation time by more than 90%. (blog.google) The immediate change for Google Cloud customers is structural: Vertex AI is no longer the whole story, and agent building now sits inside a broader Gemini Enterprise layer with registry, governance, and interoperability features. Google is betting companies want AI systems that can be assigned outcomes, not just prompts. (cloud.google.com 1) (cloud.google.com 2)