- Published on
Google Wants Gemini to Be Your Next Coworker—Complete With Its Own Email Address
- Authors
- Name
- DeepHorus
Google Cloud introduced the Gemini Agent on Thursday, a universal AI for work that the company says can accept a goal, operate autonomously, and return with a finished result. CEO Thomas Kurian unveiled the product at the company's Gemini at Work event, declaring that "work now starts in the prompt window." The launch positions Google's cloud division directly against a growing field of agentic workplace tools from rivals.
The agent operates inside Gmail, Docs, Sheets, Slides, Drive, Chat and Calendar, and also reaches into Microsoft 365 and Slack. According to Google, it keeps running in the cloud after a user closes their laptop, meaning a single task can stretch across hours or even days without supervision. That persistence is the core of the pitch: continuous work rather than session-bound assistance.
The headline feature is what Google calls coworker agents. A manager describes a role such as events coordinator, and Gemini creates an agent with its own Workspace account, complete with an email address, calendar, Drive storage and a listing in the company directory. The approach echoes earlier experiments from other vendors, including Grok Bot, OpenAI's GPT dots, and Hermes's bots feature.
Integration is broad. The agent connects to Salesforce, ServiceNow, Jira, Snowflake and BigQuery, plus any server speaking Model Context Protocol, an open standard that functions like a universal plug between AI and software. It is also not limited to Google's own models: Gemini currently orchestrates across Gemini and Anthropic's Claude, routing each job to whichever model Google judges best for quality and cost.
Letting software act independently raises obvious safety questions, and Google's answer is to treat each agent like an employee. Every agent receives a cryptographically attested identity, and every action lands in an audit trail attributed to the agent rather than a person. Agents run inside an Agent Sandbox, a walled-off environment, while their traffic passes through Agent Gateway, which Google describes as an "AI network firewall." Containment has failed before: in July, researchers at Accomplish AI reported that the local mode of Claude Cowork could escape its Linux virtual machine on Macs and reach host files.
Cost is the other concern, since agents consume enormous volumes of tokens, the text chunks that AI models read and write and that customers are billed for. Google says nearly 500 customers each processed more than one trillion tokens over the past year. To address runaway spending, it added real-time caps: when a project hits its limit, the agent pauses until someone clicks to resume.
Google published no pricing and no general availability date for the Gemini Agent. Financial services and legal versions are in preview, while government, healthcare and retail are listed as coming soon, suggesting a staged rollout that leaves the broadest enterprise deployment still ahead.
