Google announces 'Gemini agent' as ‘universal agent for work’

Key Points:

• Google Cloud announced the Gemini agent at Gemini at Work 2026, a universal AI assistant designed for enterprise use across the Gemini Enterprise app, Workspace, and third-party services, capable of handling tasks like answering questions, managing knowledge work, creating media, and coding.
• The Gemini agent can autonomously manage complex tasks such as scheduling meetings by identifying participants from chat history and coordinating calendars, acting both as a personal assistant and a team member fulfilling roles like project manager or analyst.
• It operates using multiple AI models, including Gemini and Claude families, selecting the best fit for each task to optimize quality and reduce costs, and employs sub-agents to manage multi-step workflows that may span hours or days.
• The agent maintains four types of memory—session, semantic, procedural, and episodic—to retain context, knowledge, skills, and past activities, enabling continuous and personalized task execution across cloud environments.
• Accessible via @Gemini mentions in Gmail, Docs, Sheets, Slides, and Chat, and integrable through APIs into various platforms like Slack and Microsoft 365, the Gemini agent is currently in private preview with plans for broad availability to Workspace Business and Enterprise customers soon.