Google Cloud turns Gemini into a persistent, multi-agent assistant for work

What is changing?

At Gemini at Work 2026, Google Cloud introduced Gemini as a single universal work agent that carries company context. The agent can answer questions, handle knowledge work, create images and media, and write and run code. When a user delegates an objective, it plans the work, selects skills and tools, connects to enterprise systems, and returns finished results inside work surfaces such as Gmail, Drive, Docs, Slides, Sheets, Chat, Calendar, the command line, Microsoft 365, or Slack.

Because it runs persistently in the cloud, the system keeps the same memory and personalization context across devices and can continue jobs that take hours or days. Gemini can also create temporary sub-agents and coordinate parallel or sequential tasks. Google separates the agent identity from the model layer: the system can currently route work between Gemini and Anthropic's Claude models. Tool registries can connect to MCP servers, while Skills registries hold reusable workflows shared by teams. Industry editions for financial services and legal are in preview, with more sectors planned later.