Google Cloud Launches Gemini Agent, a Universal Work Agent for the Enterprise

Google Cloud has launched Gemini agent, a single universal agent designed to carry out enterprise work across the applications and systems employees already use. The company introduced the product at its Gemini at Work event.

Rather than a chat assistant that answers one prompt at a time, the Gemini agent is built to take on objectives: it plans, delegates to sub-agents, and keeps persistent context as it works. Google positions it as a one-stop agent for knowledge work — handling information research, document creation, and writing and running code inside one window.

One of the more notable design choices is that the agent routes work across models from more than one vendor. Reports on the launch describe an enterprise agent that can draw on both Google's Gemini models and Anthropic's Claude models, with sub-agents able to use Claude where it fits. That is an unusual posture for a major cloud provider, whose agent layer would traditionally favour its own models.

The agent is meant to reach into the productivity and enterprise stack rather than sit apart from it — spanning Workspace apps such as Gmail, Drive and Docs, developer environments, and connected enterprise systems. The pitch is that a worker states a goal and the agent assembles the steps, moving between mail, files, documents, code and business applications.

Google is pairing the rollout with governance features aimed at the concerns that block enterprise agent adoption. Coverage of the launch points to spending caps and security controls, giving IT administrators a way to bound what the agent can do and what it can cost. For companies experimenting with autonomous workflows, cost and control — not raw capability — are often the deciding factors.

The launch lands in a crowded field. Every major cloud and software vendor is racing to put an agent layer on top of its existing products, and the differentiator is increasingly less about whether an agent can act and more about how safely and predictably it can be trusted to act inside a real business. Google's scale is a factor here: the company has described a large base of paid seats and heavy penetration among the Fortune 100, giving it a distribution advantage for pushing agentic features into firms that already pay for its tools.

Announcing the agent at a dedicated Gemini at Work event also signals how Google wants the product framed — not as a consumer novelty but as infrastructure for corporate workflows, competing for the same budgets as the incumbent productivity suites.

What to watch next is how the multi-model routing performs in practice, how strictly the spending and security controls bind, and whether enterprises treat a general-purpose work agent as something they trust with real tasks or as a feature they test in narrow, low-stakes corners first. Those answers, more than the announcement itself, will determine whether the Gemini agent becomes the way companies work or one more assistant in a growing list.