Google Cloud on October 8, 2026, introduced Gemini agent, an enterprise AI system designed to plan and complete complex work across business software, including coding and tasks delegated to specialized subagents.
The company presented Gemini agent at its Gemini at Work 2026 event as a “universal agent” that can handle knowledge work, answer questions, create content, produce media, and write and run code. Rather than requiring users to provide step-by-step instructions, the system is intended to accept an objective, determine the necessary steps, use relevant tools, and return a finished result. It can work within Google Workspace applications such as Gmail, Drive, Docs, Sheets, Calendar, and Chat, while also connecting to Microsoft 365, Slack, Jira, Git, databases, and other enterprise systems.
For larger assignments, Gemini can create temporary subagents with specialized roles and coordinate their work over extended periods. Google said the agent includes administrative controls, security protections, permissions, sandboxing, and spending limits intended for corporate deployment. The launch places Google in an increasingly competitive race with other technology companies developing AI systems that act more like autonomous digital coworkers.
Google is initially emphasizing enterprise use, where it says organizations can apply the agent to connected workflows and internal data while retaining oversight of its actions.