Google introduces a Gemini agent across workplace tools

Google’s work agent brings reusable skills, connected tools and memory into Workspace workflows.

Google’s Mountain View campus with bicycles outside its glass-fronted building.
Archive photograph, 13 April 2014: Google’s Mountain View campus. Company context for the Gemini workplace-agent announcement. Resized and converted to WebP; full frame retained. Noah Loverbear · CC BY-SA 3.0

Google Cloud has introduced a Gemini agent for workplace tasks, combining reusable skills, connected business tools and persistent context. The announcement describes an agent that can answer questions, create media and run code, with access inside Workspace applications including Gmail, Docs, Sheets and Calendar.

Google separates the agent from its underlying model choice and describes support for Gemini and Claude models. It also outlines identity, permissions and governance controls for delegated work. VCT’s takeaway: shared memory may reduce repeated setup, but teams still need to decide which company data and actions each agent may access. The announcement alone does not establish reliability in a reader’s own workflows.

Source

[Google Cloud announcement](https://cloud.google.com/blog/products/ai-machine-learning/welcome-to-gemini-at-work-2026).

Google Cloud