Google doubles down on agents

Google used its Cloud Next event on April 22 to make AI agents the center of its enterprise sales pitch. (reuters.com) At the conference in Las Vegas, Google launched Gemini Enterprise Agent Platform, a system for building, deploying and governing large numbers of AI agents inside companies. Google said the product is the evolution of Vertex AI and adds agent integration, DevOps, orchestration and security tools. (cloud.google.com) An AI agent is software that can carry out multi-step tasks with less human input than a chatbot, and Google is pitching this platform as the place where information technology teams manage those agents at scale. Sundar Pichai said the question for customers has shifted from whether they can build an agent to how they can manage “thousands of them.” (blog.google) Google tied that product launch to a broader message it called the “Agentic Enterprise,” a term it used repeatedly across Cloud Next materials on April 22. The company said nearly 75% of Google Cloud customers now use its artificial intelligence products, and that direct customer use of its first-party models has risen to more than 16 billion tokens per minute, up from 10 billion last quarter. (cloud.google.com) The company is also trying to show that the agent push is not just a demo-stage idea. Google and Merck announced a partnership on April 22 to use Gemini Enterprise across Merck’s operations, with Merck saying the effort is meant to support 75,000 employees worldwide. (merck.com) Other customer references point to the same strategy. Mars said on April 22 that it would use Gemini Enterprise as the primary artificial intelligence operating system for its global workforce, and Home Depot said in January that it was expanding work with Google Cloud on agentic tools for shoppers and store associates. (googlecloudpresscorner.com) (corporate.homedepot.com) SAP is part of that sales story too. Google has been pushing joint data and artificial intelligence products with SAP since 2025, including tools to analyze SAP data in BigQuery and build agents on top of enterprise systems that many large companies already run. (cloud.google.com) The backdrop is a cloud market where Amazon, Microsoft and Google are all spending heavily to turn generative artificial intelligence into recurring enterprise revenue. Google used Cloud Next to pair its agent software with new infrastructure pitches, including its Tensor Processing Units and custom Axion processors, so customers buy both the tools and the computing behind them. (reuters.com) (cloud.google.com) Google’s bet is that companies will not want one assistant on one screen. It wants them buying a managed system for fleets of agents, with Google Cloud selling the models, the controls and the infrastructure underneath. (cloud.google.com)