Google Cloud introduced Gemini Agent on October 8 as a universal work agent available through a conversational interface and APIs. Google says the agent can combine tools, reusable skills and memory to carry out multi-step work across enterprise applications.
One agent layer across models and systems
The company says Gemini Agent can route tasks among Google models and selected third-party models, including Anthropic’s Claude, while drawing on enterprise context and permissions. The announcement also describes governance and observability controls intended for administrators.
Reuters independently confirmed the product announcement and its competitive context. Google’s claims about productivity gains and task performance are vendor-reported and have not been independently reproduced here.