Google is turning Gemini into an agent that gets things done. On October 8, it launched Gemini Agent for the workplace. It can answer questions, handle knowledge work, and write and run code. Set a goal and wait for the results: it can act as a project manager or financial analyst.

It’s available through a wide range of platforms: the web, mobile, Windows, Mac, Workspace, and Microsoft 365. It can also be embedded in third-party apps and run as a headless agent without an interface. It’s more like a colleague than a chatbot, with its own enterprise identity and the ability to work toward a goal for several days.

This is different from earlier assistants that simply followed your instructions. The Google Cloud CEO says it can choose the most suitable model, manage costs, and provide the security and governance capabilities businesses need.

The backdrop is money: hundreds of billions of dollars in AI infrastructure need to be monetized through the enterprise market, and Microsoft, OpenAI, Anthropic, and Amazon are all competing for the same high-margin contracts.