Google Cloud on Thursday unveiled the Gemini agent, a single universal AI agent for enterprise work designed to answer questions, handle knowledge work, create images and media, and write and run code from one interface. Google Cloud CEO Thomas Kurian introduced the agent at the company's Gemini at Work 2026 event, describing it as an agent that holds an organization's business context, plans the work, uses skills and tools, and connects to customers' business systems.
The agent runs persistently in the cloud with shared memory and a personalization graph, so jobs that take hours or days continue after a user closes their laptop. It can spin up temporary sub-agents for multi-step work and create persistent 'coworker' agents with their own email addresses, calendars and storage. It draws on a tools registry spanning Salesforce, ServiceNow, Jira, Snowflake, Databricks and any Model Context Protocol server, and works inside Google Workspace, Microsoft 365 and Slack.
A notable feature is model flexibility: each job is routed to the best-fit model, drawing on Google's own Gemini models and Anthropic's Claude models today, with OpenAI and open-weights models to follow. Governance features include cryptographically attested agent identities, OAuth-propagated permissions, action-level audit trails, an Agent Sandbox and an 'Agent Gateway' described as an AI network firewall, alongside real-time per-project spend caps.
Google says the agent is in private preview with general availability expected by the end of October starting in North America, plus industry specializations for financial services and legal now in preview and government, healthcare and retail to follow. Kurian said nearly 90% of the Fortune 100 now use Gemini Enterprise, signaling how aggressively Google is pushing the agent era into its cloud business.
