Google is introducing a Gemini agent for businesses that can take on work across apps and devices from one interface. Announced at the Gemini at Work event on Thursday, the agent is available in the Gemini Enterprise app and is currently in private preview for enterprise customers.
One place to assign work
Users can chat with the Gemini agent and give it tasks through the Gemini Enterprise app. The idea is to make the agent a common starting point for work that might otherwise involve moving among different services and screens.
Gemini can operate inside Google Workspace apps, including Gmail, Drive, Docs, Sheets and Calendar. Users can also reach it from a mobile device, desktop computer or the web, as well as through third-party apps such as Slack and Microsoft 365.
Because the agent runs in the cloud, Google says it will keep the same context across connected devices. That means a user can change where they interact with Gemini without losing the thread of the work, according to the company.
Agents tailored to particular jobs
The system can work with job-specific sub-agents to complete tasks. Google also describes a “coworker agent” with its own identity and an “@agents.company.com” email address, presenting the agent as something that could take part in work through a recognizable account.
Gemini will automatically choose an AI model based on the task, Google says. In practice, this puts task assignment and model selection behind the same agent interface, while specialized sub-agents can help with work tied to particular roles.
The announcement describes a set of capabilities rather than a broadly available product. Private preview means enterprise customers are the group currently able to access this Gemini agent; the source does not give a public launch date.
A broader interface for Google’s AI plans
The enterprise agent points toward a version of Gemini that can answer questions and carry out work from one window. Its reach across Workspace, devices and outside services is central to that approach: users could assign a task without first choosing which app should handle it.
Google has also launched a consumer-facing agent called Gemini Spark in May. Subscribers can use Spark to perform tasks across Workspace and the web, but it operates through a separate interface within the Gemini app. The enterprise agent’s single interface and work-focused connections distinguish the experience described in this announcement.
What businesses will be watching
For organizations, the proposal is to give an AI agent access to work across several familiar tools while carrying context between devices. It could make tasks that span email, documents, calendars or other connected apps easier to hand off through one conversation.
The announcement leaves practical questions open, including how the experience will work for specific teams and when it may become available beyond private preview. For now, Google’s stated direction is clear: Gemini Enterprise is being positioned as a central place to delegate work, with sub-agents and model selection supporting tasks behind the scenes.