Google Cloud announces ‘Gemini agent’ as ‘universal agent for work’

At the Gemini at Work 2026 event, Google Cloud unveiled the Gemini agent, a groundbreaking tool designed to assist enterprise customers. This new universal agent will be available within the Gemini Enterprise app, Workspace, and third-party services. With the Gemini agent, users can input objectives instead of detailed instructions, enabling it to answer questions, manage knowledge work, generate media, and write or execute code.

The agent serves not only as a personal assistant but can also act as a team member, taking on roles such as project manager within a group or an analyst for a finance department. Under the hood, the Gemini agent dynamically assigns the most suitable model for each task, utilizing both Gemini and Claude models currently, with additional private and open models on the horizon.

The system employs sub-agents to tackle complex multi-step tasks and communicates with these sub-agents to orchestrate workflows, including parallel and sequential processes that may span hours or even days. The Gemini agent operates in the cloud, maintaining a single set of memories, context, and personalization graph. These memories can be accessed from various sources, including Gmail, Google Docs, Sheets, Slides, and Chat, with the mention of "@Gemini" triggering the agent's response.

The agent is accessible across multiple devices and channels, such as command line, Google Workspace, Slack, and Microsoft 365, among others. An API is provided for integration into third-party applications, allowing for a "headless agent" mode where a dedicated user interface is not required. The Gemini agent will soon be widely available for Workspace customers with select Business and Enterprise plans, currently in private preview.