Google bets on enterprise agents
- At Cloud Next, Google unveiled a Gemini Enterprise Agent Platform built mainly for IT and technical users. - TechCrunch frames the product as an enterprise-first agent builder, not a consumer-facing tool. - The announcement positions agentic AI as infrastructure to be integrated and governed inside organisations. (techcrunch.com)
Google used Cloud Next on April 22 to launch Gemini Enterprise Agent Platform, a new Google Cloud product for building and governing artificial intelligence agents inside companies. (cloud.google.com) An AI agent is software that can plan and carry out multi-step work, like checking systems, calling tools, and handing tasks to other agents instead of just answering one prompt. Google said the new platform is the evolution of Vertex AI, combining model selection, tuning, agent building, orchestration, DevOps, and security in one system. (cloud.google.com) Google aimed the new platform at developers, information technology teams, and other technical staff, not at rank-and-file office workers. TechCrunch reported that business users are being steered instead to the Gemini Enterprise app, where they can run agents built by IT teams or use simpler no-code tools. (techcrunch.com) Google said the platform supports its own Gemini 3.1 Pro, Gemini 3.1 Flash Image, and Lyria 3 models, and also Anthropic’s Claude Opus, Sonnet, and Haiku models. ZDNET reported that Google is offering more than 200 models through the revamped platform. (blog.google) (zdnet.com) The company’s pitch at Cloud Next was that agents now need the same controls companies already expect for databases, networks, and employee software. Google said the platform is built to “build, scale, govern, and optimize” agents and tied the launch to what Chief Executive Thomas Kurian called the “Agentic Enterprise.” (cloud.google.com) (blog.google) That framing reflects where large companies have gotten stuck with generative artificial intelligence over the past year: not on demos, but on security, testing, identity, and deployment. ZDNET reported that Google added tools including Agent Identity, which gives each agent a cryptographic ID, and Agent Simulation, which lets teams test agents before release. (zdnet.com) Google paired the technical platform with a separate employee-facing layer. In keynote materials, the company said the Gemini Enterprise app now includes Agent Designer, an Inbox for agent activity, long-running agents, Skills, and Projects. (cloud.google.com) The launch also folded Google’s older enterprise artificial intelligence pieces into a clearer stack. Google said Gemini Enterprise is now the end-to-end system for the “agentic era,” while outside reporting described Agentspace as having been rebranded into Gemini Enterprise and Vertex AI as being rolled into the new Agent Platform. (cloud.google.com) (thenewstack.io) Google made the announcement as it pushed broader adoption numbers for its cloud artificial intelligence business. The company said nearly 75% of Google Cloud customers now use its AI products, and 330 customers processed more than 1 trillion tokens over the past 12 months. (cloud.google.com) The immediate contest is with Amazon and Microsoft, which are also turning agent software into managed enterprise platforms. TechCrunch described Google’s new product as its answer to Amazon Bedrock AgentCore and Microsoft Foundry, with the bet that companies will buy agent systems the way they buy other cloud infrastructure: centrally, with controls attached. (techcrunch.com)