Google has introduced a new 'universal' AI agent for its Gemini Enterprise platform, designed to autonomously handle workplace tasks across various applications.

Google announced a new AI agent for its Gemini Enterprise platform during the 'Gemini at Work 2026' event on Thursday. The tool is designed to act as a 'universal agent' that can autonomously plan and execute complex workplace tasks by integrating with Google Workspace, Microsoft 365, Slack, and other enterprise systems. Google Cloud CEO Thomas Kurian stated that the agent can be assigned objectives rather than simple instructions, allowing it to manage workflows and create specialized sub-agents.

Capabilities and Integration

The new Gemini agent operates in the cloud and maintains context across mobile, desktop, and web interfaces. It can perform functions such as writing code, creating media, and managing project tasks. To complete assignments, the agent can generate temporary, job-specific sub-agents. Users can monitor the agent's progress, including its thinking process and task delegation, through a 'tasks inbox' interface. The agent is designed to function as a co-worker, even possessing its own dedicated Workspace account and email address.

Model Flexibility and Enterprise Focus

By default, the system selects the most appropriate AI model for a given task, though users maintain the ability to manually select models, including third-party options like Anthropic’s Claude. Google plans to expand this model picker to include open-source and other private models in the future. The agent is currently available to enterprise customers in private preview, with Google prioritizing business adoption before a planned rollout to consumers.