Google launches agentic AI for Gemini Enterprise with subagents and workplace identity

At a Google Cloud event, the company introduced a unified agent for Gemini Enterprise designed to perform tasks beyond answering questions, including planning, generating code, scheduling meetings, and booking travel. The agent can operate across business applications such as Google Workspace, Microsoft 365, Slack, Jira, Confluence, Git, BigQuery, Databricks, Postgres, and Snowflake, as well as any Model Context Protocol (MCP) server within or outside a company’s network.

The agent is capable of delegating work to subagents and selecting models to complete tasks, with users able to choose models including third-party options like Anthropic’s Claude. Google plans to expand model selection to include open source and private models in the future.

Users can monitor the agent’s progress through a 'tasks inbox' interface that displays its reasoning, task delegation, skill loading, and code execution. The agent is assigned its own Workspace identity, including an email address and contextual awareness of team structures, time zones, approval workflows, and calendars.

Google will initially focus on business deployments before expanding to consumers. Early testers included On, Shopify, and PayPal, while enterprise customers already using Gemini Enterprise include BNP Paribas, Bradesco, Merck, Orange Spain, Santee Cooper, SOMPO, Ulta Beauty, and Wesfarmers.

The agent will be accessible via iOS and Android mobile devices, Windows and Mac desktops, the command line interface, Google Workspace, Microsoft 365, ServiceNow, and Slack. Google also announced new spending controls for enterprise AI, including multi-model orchestration, smart routing, and real-time spend caps.