Google Cloud has introduced the Gemini agent, a versatile AI assistant designed for workplace tasks. Individual digital assistants based on this technology will function like company employees, complete with their own email addresses, calendars, storage on Drive, and entries in the corporate directory.

Today, we are announcing our new Gemini agent for your business—a universal cloud agent running 24/7 to get your work done. It arrives knowing your tools, data, and work history; it works across knowledge work, content, and code; and it comes with built-in governance to help your data stay secure and cost stay in control → https://t.co/NRlFcjz5zu

— Google Cloud (@googlecloud) October 8, 2026

How It Works

The Gemini agent integrates chat, task execution, and coding within a single interface. Users can set objectives, and the agent will plan the tasks and deliver the results. Operating in the cloud, it can continue working even after the user closes their laptop.

Interface of the Gemini agent. Source: Google Cloud.

The system will select models based on the quality and cost for each task, currently utilizing Gemini and Claude models from Anthropic.

A colleague agent can be created based on a role description, such as "event coordinator." Employees will be able to add this agent to chats, assign tasks, and receive edits on documents, with its activities recorded under its own name.

According to Google representatives, the digital assistant will only see what has been shared with it.

Control Measures

Each agent will be assigned a cryptographically verified identifier, with all actions logged in an audit trail attributed to the agent rather than an individual.

They will operate in a secure environment, and companies can set spending limits; if these limits are exceeded, the agent will pause until manually restarted.

Industry-specific versions of Gemini for the finance and legal sectors are currently available in preview mode. Google has not disclosed the timeline for a full launch or pricing details.

A previous five-month study by the company indicated that the AI agent in team settings made social errors, such as contacting a vice president about routine bugs and publishing a colleague's personal draft.

In October, Google also introduced the Nano Banana 2.1 image generation model and launched an AI platform for creating games from text descriptions.

Follow ForkLog on social media:

Telegram (main channel) Facebook X