Google Cloud launches Gemini Agent: a single general-purpose work agent integrating Workspace, multi-model routing, and enterprise-grade governance.

Beating AI News Flash: Google Cloud CEO Thomas Kurian unveiled Gemini Agent at the Gemini at Work 2026 conference, positioning it as a single universal work agent — users simply provide goals rather than instructions, and it can complete knowledge work, content creation, image and media generation, as well as code writing and execution. The agent features a unified entry point, ubiquitous access (web, iOS, Android, Windows, Mac, and channels including Workspace, Microsoft 365, and Slack), persistent execution (running in the cloud, tasks continue after closing the laptop, with memory and personalization graphs unified across devices), multi-agent orchestration (can dynamically create temporary sub-agents with independent identities, and can also serve as a long-term colleague agent with a dedicated email and storage), deep contextual understanding, and model selection flexibility — the underlying model can be freely switched across Google's Gemini series and Anthropic's Claude, among others, to match the best model to the task and reduce costs. Currently, nearly 80% of Google Cloud customers use its AI products, and nearly 90% of the Fortune 100 use Gemini Enterprise.

The agent is directly embedded in Gmail, Drive, Docs, Slides, Sheets, Chat, and Calendar, carrying the same memory, skills, and controls; a new data analysis skill enables both technical and business users to obtain actionable insights in minutes using natural language; it also provides industry-specific tools and knowledge for financial services and legal teams. On the security and governance front, Google offers mechanisms including identity and policy management, authorization and permission controls, secure sandboxes, and network gateways; for cost control, it achieves this through multi-model orchestration, intelligent routing, and real-time spending caps. Early testing customers include sports brand On, which uses dynamic model selection to accelerate time to market; Shopify mixes frontier models for millions of merchants, and PayPal routes 10 million multi-model requests per week.