Google Cloud launches Gemini Agent for work

Google Cloud has introduced Gemini Agent, a universal AI agent for enterprise work that can answer questions, handle knowledge tasks, create media, write code and operate across company systems.

The important shift is that Google is positioning Gemini less as a chat surface and more as a persistent work layer. The agent can run in Google Workspace, Microsoft 365, Slack, desktop and mobile apps, command-line workflows and headless integrations. Google says it can keep memory and context across devices, schedule tasks, respond to events, and spin up temporary sub-agents for multi-step jobs.

For companies, the practical hook is control. Gemini Agent is meant to connect to tools such as Confluence, Jira, Salesforce, ServiceNow, BigQuery, Postgres and MCP servers while using identity, permission, sandboxing, network gateway and policy controls. Google also says the agent can choose the best model for a task, including Gemini models, Claude models from Anthropic and other models in the future, to balance quality and cost.

This is a direct push into the enterprise agent layer: the place where email, files, data, code and approvals meet. If it works as described, the competition moves from individual AI assistants toward governed agents that can actually sit inside company operations.