Google makes agents central

Google used its Cloud Next conference in Las Vegas on April 22 to make AI agents the centerpiece of its enterprise software pitch. (reuters.com) At the event, Alphabet chief executive Sundar Pichai told customers and investors that agents — software that can carry out tasks, not just answer prompts — are central to how Google plans to make money from artificial intelligence in business tools. Google also introduced the Gemini Enterprise Agent Platform at Next ’26. (reuters.com) (blog.google) Google said the new platform combines Vertex AI model tools with agent integration, security, DevOps and monitoring features, and gives developers access to Gemini 3.1 Pro and Gemini 3.1 Flash Image. Google’s Cloud Next recap described the broader goal as helping companies become an “agentic enterprise.” (blog.google 1) (blog.google 2) An AI agent is software that can take actions across business systems after receiving instructions and access to data. That is different from a chatbot that mainly returns text, because an agent can be connected to calendars, customer records, internal documents or ticketing systems and then complete a step in a workflow. (reuters.com) (blog.google) Google’s demos focused on that action layer. Reuters reported that the company showed agents tied to practical workplace jobs and partner products, while Google and Salesforce announced integrations meant to let agents work across Google Workspace, Slack, Salesforce data and other enterprise systems. (reuters.com) (salesforce.com) That changes the sales pitch for enterprise AI. Instead of charging only for model access or chat features, Google is packaging models, security controls, orchestration tools and partner connections into software that companies can use to automate work inside existing applications. (reuters.com) (blog.google) It also raises the burden on software vendors and information-technology teams. If agents are going to send messages, update records or trigger workflows, companies need permission controls, audit trails, testing and monitoring so the software can act inside business systems without creating security or compliance problems. (blog.google) (salesforce.com) Google has been building toward this for more than a year. In 2024 it introduced Agentspace as a way for employees to search company information and use custom agents, and in late 2025 Pichai described Gemini Enterprise as a platform that lets workers chat with company data and build agents. (blog.google 1) (blog.google 2) The competitive backdrop is crowded. Reuters said Google is trying to win enterprise customers as OpenAI, Anthropic and Salesforce all push their own workplace AI products, and Google used Next ’26 to argue that its advantage is a full stack that runs from chips and cloud infrastructure to models and office software. (reuters.com) (blog.google) The immediate test is whether customers buy agents as production software, not as demos. Google’s message in Las Vegas was that the next enterprise AI contract will be about software that can do the work, with controls tight enough for companies to trust it. (reuters.com)