Google Cloud unveils Gemini Agent for workplace tasks
In short
Google Cloud announced Gemini Agent, a workplace agent that it says can answer questions, create content and write code from goals set by users. IT Home reported that it is in private preview, with broader access planned for eligible Workspace customers.
Read the full story 1 min read
Google Cloud announced Gemini Agent on October 8, presenting it as a general-purpose agent for workplace tasks, CLS and IT Home reported. Google says users can set a goal rather than supply detailed operating instructions, and the agent can answer questions, perform knowledge work, create content and write programs. IT Home said the announcement took place at Gemini at Work 2026 and that the product supports Gemini Enterprise, Workspace and third-party services. [ 1 , 5 ]
The agent currently supports Google's Gemini models and Anthropic's Claude models, according to CLS and IT Home. Both reported plans to support additional proprietary and open-source models. IT Home said Google would provide an API allowing developers to integrate Gemini Agent into third-party applications. It reported that the product is in private preview and is expected to become more widely available soon to eligible Workspace customers on selected Business and Enterprise plans. [ 1 , 5 ]
36Kr, citing Jiemian, reported that the agent can use internal company tools and data to complete user-defined goals. It listed direct integration with Gmail, Drive, Docs, Sheets, Slides, Chat and Calendar, along with the ability to create colleague agents with separate identities and permissions. The report also described new skills for data analysis, data engineering and machine learning, plus specialised versions for financial services and the legal industry. [ 3 ]
IT Home said the agent can act as a personal assistant or a team member, including in project-management and financial-analysis roles. 36Kr reported that enterprise controls include permission management, auditing, sandboxing and real-time spending limits, and that the system can dynamically select among different models. [ 3 , 5 ]
Why it matters
Google says the agent can work across its workplace services and third-party applications. The coverage describes both personal-assistant uses and roles within teams.
Key facts
- Google Cloud announced Gemini Agent for workplace tasks. [ 1 , 3 , 5 ]
- Google says the agent can answer questions, handle knowledge work, create content and write code. [ 1 , 3 , 5 ]
- The agent supports Google's Gemini models and Anthropic's Claude models. [ 1 , 3 , 5 ]
- IT Home reported that the agent is in private preview. [ 5 ]
- 36Kr reported enterprise controls including permissions, auditing, sandboxing and real-time spending limits. [ 3 ]
Confirmed by several sources
Still unclear
- The date of broader availability. IT Home says wider access is expected soon but gives no date.
- Enterprise controls and industry-specific versions. Only 36Kr describes the controls and versions for financial services and law.