Google Cloud launches Gemini agent

Google is positioning Gemini agent as the control plane for enterprise work, but its real test will be reliable execution—not another polished chat interface.

Cross-application workflows could eliminate much of the manual handoff between email, documents, code, and business systems.

Model routing across Gemini and Claude gives enterprises flexibility while making the agent layer more important than the underlying model.

Persistent coworker agents with identities, permissions, calendars, and storage push AI toward organizational infrastructure.

MCP support and integrations with tools such as Jira, Git, Salesforce, Snowflake, and BigQuery make the platform relevant to developers.

Google has not yet disclosed pricing, general availability, or independent reliability benchmarks, leaving deployment economics unclear.