Google's Gemini Agent Can Route Work to Claude and Act as an AI Teammate
Google has introduced a Gemini agent that can plan tasks, use connected business tools, and return completed results. It can route jobs between Gemini and Anthropic's Claude models, with future support planned for more private and open models. The agent runs persistently in the cloud and includes enterprise controls such as permissions, audit trails, sandboxes, and spending limits.
Google’s new Gemini agent is positioned as a coordinator rather than a single chatbot. It can take a goal, devise a plan, operate linked business applications, and deliver a finished output. Its routing layer may send work to Gemini or supported Claude models, with more private and open models expected later.
The agent can keep running in the cloud after an employee signs off, preserving context across devices. It may spawn temporary sub-agents for parallel or sequential tasks. Separate coworker agents can hold persistent identities, including company email, calendar, storage, and directory presence, while enterprise controls cover permissions, audits, sandboxes, and spending.
This shift could affect knowledge workers and teams by letting them delegate multi-step tasks, potentially changing workflows, oversight, and job roles. Managers may need new practices for reviewing agent output, while IT and security teams could focus on permissions, audit trails, and spending controls. Smaller firms may gain access to coordinated AI work previously requiring custom integration, though reliance on cloud agents and model routing may raise concerns about transparency, accountability, and skill development.