Google Cloud has introduced Gemini agent, a universal AI assistant designed to function as an autonomous employee within corporate environments. These agents are assigned their own email addresses, calendar access, Google Drive storage, and entries in corporate directories, allowing them to operate continuously in the cloud.
The system integrates chat, task execution, and coding within a single interface. Users define goals, and the agent manages the planning and execution process. The platform automatically selects between Gemini and Anthropic Claude models based on performance and cost requirements, while maintaining operations even when the user is offline.
For security and accountability, each agent receives a cryptographically verified identifier. All actions are logged in an audit trail, and agents operate within isolated environments. Companies can set spending limits, which automatically pause agent activity if exceeded. Access is restricted to data explicitly shared with the agent.
Google is currently offering preview versions of Gemini agents tailored for the legal and financial sectors. While the company has not disclosed pricing or a general release date, it noted that previous internal testing revealed potential social risks, such as agents sharing inappropriate content with senior management.
