At its Gemini at Work 2026 event, Google unveiled a universal agent and API for Gemini AI, designed to coordinate enterprise AI use. The agent can start from a simple prompt and work autonomously across workplace applications, systems and devices to complete complex tasks. It can act as a personal assistant or team member and manage workflows over extended periods.
The agent is model-agnostic: it can select an appropriate, cost-effective AI model for each task, including competitors’ models such as Claude. Google said it plans to add more models. The launch signals the company’s ambition to shape how businesses use AI across their organizations, beyond providing AI models.