News in Short

• Google Cloud has introduced the Gemini agent as a universal agent for work.
• The agent can use an organisation’s business context to handle different types of tasks.
• It can plan work, use skills and tools, and connect with business systems.
• Gemini can return completed work inside documents, inboxes and developer environments.
• Google Cloud says the agent includes model selection, cost controls and enterprise governance.

Google Cloud has introduced the Gemini agent, a universal AI agent designed to handle work from a single prompt.

The announcement was made at Gemini at Work 2026. Google Cloud says the agent can use an organisation’s business context across different types of work.

Instead of using separate AI tools for different tasks, the Gemini agent is designed to bring knowledge work, content creation, questions and coding into one prompt-driven experience.

Gemini agent turns prompts into completed work

The biggest change is how Google Cloud wants employees to interact with AI.

Users can start with a prompt instead of opening separate tools for individual tasks. Gemini can then plan the work, select the required skills and use available tools.

It can also connect with an organisation’s business systems. This allows the agent to work with business context rather than treating every prompt as an isolated request.

Google Cloud says the goal is to return something finished rather than simply provide an answer.

The agent can work across existing workflows

Gemini is designed to bring completed work back into the environments employees already use.

That includes documents, inboxes and developer environments. As a result, users do not necessarily need to move between multiple applications to use the agent’s output.

The approach also extends beyond traditional knowledge work. Google Cloud says Gemini can support everything from answering questions and creating content to coding.

This makes the Gemini agent broader than a conventional chatbot. Its focus is on completing tasks across a user’s existing work environment.

Gemini can choose the model for the task

Another part of Google’s approach is model selection. The Gemini agent can choose the best model for a particular job.

That could help organisations avoid using the same model for every task. More importantly, Google Cloud says the system includes built-in cost controls.

For enterprise customers, controlling AI usage and spending can be as important as the quality of the output. The model-selection approach is therefore designed to balance the work required with the resources used.

Enterprise security remains a key focus

Google Cloud is also positioning the Gemini agent around enterprise requirements.

The company says the agent includes security, administration and governance capabilities required by enterprise customers.

That is particularly important as AI agents move from answering questions to taking actions across business systems.

An agent that can access organisational context and tools needs stronger controls than a standalone chatbot. Google Cloud is therefore making governance part of the core offering.

Google is pushing AI agents deeper into work

The Gemini agent reflects a broader shift in how enterprise AI is being positioned.

Instead of asking AI to generate a response and then completing the remaining steps manually, users can increasingly ask an agent to handle the workflow itself.

Google Cloud’s pitch is built around that idea. A single prompt can provide the starting point, while Gemini plans the task, uses the necessary tools and returns a finished result.

The company has not detailed every capability or business system supported by the Gemini agent in the announcement. However, its positioning is clear: Google wants Gemini to become a universal AI interface for enterprise work.