Google is expanding Gemini into an enterprise agent that it says can take an assignment, break it into separate tasks, and leave those jobs running in the cloud for hours or days.
Google Cloud CEO Thomas Kurian announced the Gemini agent at the company’s Gemini at Work 2026 event, describing it as a “single, universal agent for work.” He said the agent can handle knowledge work, generate images and other media, as well as write and execute code.
In an accompanying statement, Kurian said the agent can handle delegated tasks autonomously without compromising security, administration, and governance required by client companies.
"It plans the work, uses skills and tools, connects to your systems, and brings back something finished - inside the documents, the inbox, and the developer environments you already work in," he wrote.
Sub-agents and coworker agents
Google is also giving Gemini the ability to spin up temporary sub-agents for individual jobs. Each can have its own identity, with the chief agent coordinating their work as tasks are carried out in parallel or one after another, Kurian said.
"You never have to re-brief it or wonder which machine you told it to do something on," Kurian wrote. "Work that takes hours or days keeps running after you close your laptop, and it is still there when you come back."
The agent can also work alongside “coworker agents,” with their own Workspace accounts, email addresses, calendars, and Drive storage. Their access is limited to information shared with them, according to the statement.
Google said Gemini can route tasks between its own models and Anthropic’s Claude models, with support for private and open models planned for later.
The agent can operate across Gmail, Drive, Docs, Slides, Sheets, Chat, and Calendar.
The system can also connect to outside enterprise software and data services, including Salesforce, ServiceNow, BigQuery, Databricks, Snowflake, Git, Jira, and Microsoft 365.