Google develops its own desktop Agent to compete with Cowork

Google is steadily expanding Gemini for Business and may be getting ready to introduce a much broader update before Google I/O. One of the clearest signs is a new Agent tab that has appeared in Gemini Enterprise, placed directly next to the standard chat interface. The addition matches earlier changes already spotted in Gemini and suggests that Google is aligning the two experiences around a broader product shift.

Inside this new Agent area, there are two entry points: New Task and Inbox. When a task is started, the interface opens a chat view with an additional panel on the right. That panel contains Goal, Agents, Connected apps, Files, and a toggle called “Require human review.” The structure is notable because it is no longer just a chatbot layout. It is starting to look like a task execution workspace.

The design strongly resembles systems like Claude Cowork, where the model is given a goal, access to tools, connected services, and files, then asked to carry out a broader workflow. In that sense, this looks like Google’s attempt to move Gemini beyond simple prompting and toward a more agent-driven product built for multi-step work.

The “Require human review” option is especially important. Even if some of these agents are meant to run in the cloud, that toggle suggests Google is preparing for actions that may need oversight before execution. That could mean browser-based actions, but it also points to a wider ambition around desktop-level task handling rather than a purely web-based assistant.

At the same time, Google continues refining other parts of the Gemini experience, especially Projects and Skills. Those areas have also been evolving, and it increasingly looks like all these changes are part of a single larger rollout. Together, they suggest Google is shaping Gemini into a more complete work platform centered on agents, persistent workflows, and connected tools.

That is why the possibility of a desktop app feels increasingly realistic (yet remains a speculation at this point). Google is already known to be building a desktop app for AI Studio, so the bigger question is whether these efforts stay separate or eventually merge into a single product. Either way, the direction is becoming clearer: Google seems to be preparing a stronger response to OpenAI and Anthropic's desktop and agent-focused products.

There is clearly more on the way. What remains unclear is how much of it Google is ready to show at once!