Overview
- Google announced the Gemini agent at its Gemini at Work event, presenting a single, persistent AI colleague that lives in the Gemini Enterprise app, Gmail, Calendar, Drive, Docs, Sheets, Slides, and Chat.
- The platform is a full agent lifecycle system called the Gemini Enterprise Agent Platform that includes an Agent Development Kit, low-code Agent Studio, an Agent Runtime, and the Gemini Enterprise app for discovery and management.
- Agents can run autonomous, multi-step tasks for extended periods with a current maximum runtime of seven days and maintain four kinds of memory—session, semantic, procedural, and episodic—to preserve context across jobs.
- Google built-in features let agents have persistent identities with their own email, calendar, and Drive space, use sub-agents to coordinate parallel and sequential work, and connect via an API to third-party apps and services.
- The platform offers broad model choice—more than 200 models including Gemini 3.1 and third-party options like Anthropic’s Claude—includes cryptographic agent identities for security, and is in private preview with wider rollouts planned for select Business and Enterprise Workspace plans.