Google has introduced Gemini AI, a workplace artificial intelligence (AI) agent designed to act like a colleague rather than just a tool. Announced by Thomas Kurian, CEO, Google Cloud, at the Gemini at Work 2026 event, the agent integrates across Gmail, Docs, Drive, Sheets, and also third party platforms such as Slack and Microsoft 365.
Gemini is built as a universal agent capable of writing and running code, creating media, and managing workflows. Employees can assign tasks to it, and it delivers results independently. It can also generate ‘sub agents’ — temporary assistants with their own identities — to handle projects that may run for several days.
The agent can operate as a personal assistant or team member, taking on roles such as project manager or analyst. It builds memory and skills over time, meaning users don’t need to re brief it or track which device holds the task. All activity is logged to the agent’s identity, not the employee’s, a design that has already attracted regulatory attention in Europe.
Gemini runs in the Cloud and continues working even after human users log off, routing jobs across Google’s own models and Anthropic’s Claude. This persistent capability positions it as more than a productivity tool—it is effectively a digital coworker.
The launch places Google in direct competition with Meta’s Muse, introduced last year, and OpenAI’s Dots, released recently for ChatGPT Pro and Business accounts.
Gemini signals a shift toward hybrid workforces where humans and AI agents share the same digital environment. It raises pressing questions: how to measure performance when tasks are split, who carries responsibility for errors, and how career paths evolve when part of the workforce is non human.
Gemini marks the start of a workplace where AI agents are treated as colleagues, not just software.
