Google launches universal Gemini agent that runs across apps and devices

Google has announced a "universal" Gemini AI agent designed to carry out tasks across applications and devices in the background, according to The Verge. Unveiled at the company's Gemini at Work event on Thursday, the agent lives inside the Gemini Enterprise app, where users can chat with it and hand off work from a single interface.

How the agent works

The agent is not limited to Google's own productivity tools. According to The Verge, it can operate inside Workspace apps such as Gmail, Drive, Docs, Sheets and Calendar, and users can reach it from mobile devices, desktop computers, the web and third-party services including Slack and Microsoft 365. Because the agent runs in the cloud, it keeps the same context across every connected device, so a task started in one place can continue elsewhere without starting over.

Google also says the agent automatically selects whichever AI model best fits the job at hand, rather than locking users into a single model for every request.

Coworker agents with their own email

Two details in the announcement stand out. First, the agent can work with job-specific sub-agents to complete tasks, suggesting an architecture in which a general-purpose orchestrator delegates to specialised helpers. Second, it can operate as a "coworker agent" with a dedicated identity and its own "@agents.company.com" email address, effectively an AI colleague that can be addressed like any other member of the team.

Availability and the consumer picture

For now, access is narrow. The Verge reports that the agent is available only to enterprise customers in private preview. The publication notes, however, that it may signal Google's broader ambition: a single assistant window that can answer questions and perform work on the user's behalf.

Google already has a consumer foothold here. In May, the company launched Gemini Spark, a consumer-facing agent that subscribers can use to perform tasks across Workspace and on the web. Spark, though, runs through a separate interface inside the Gemini app, whereas the new universal agent is meant to consolidate everything in one place.

Why it matters

The announcement marks another step in the shift from chatbots that answer questions to agents that complete multi-step work in the background, across whatever apps and devices are involved. Google's pitch combines three things that competitors have largely offered separately: cross-app reach that extends into rival territory such as Slack and Microsoft 365, persistent cloud-side context that follows the user between devices, and an identity layer that lets an AI take part in ordinary workplace communication.

The coworker email detail is particularly consequential. Giving an agent its own address normalises delegating to AI through the same channels people already use, blurring the line between human and machine colleagues. Automatic model selection, meanwhile, points toward a future where users ask for outcomes rather than choosing between model versions.

The catch is availability: for now this is an enterprise-only preview. But if Google ports the unified design into the consumer Gemini app, folding in what Spark does today, it would have a single default assistant spanning the Android, Chrome and Workspace surfaces it already controls, a distribution advantage few rivals can match.