The new Gemini agent works from objectives rather than step-by-step instructions, with business customers getting access before consumers.
TechCrunch reports that Google has rolled out a unified agent built into Gemini, giving business customers access before consumers. Google Cloud's Thomas Kurian said the agent takes objectives rather than step-by-step instructions, so it can plan its own work, use custom skills and tools, and connect to a company's internal systems. Users can let Gemini pick the underlying model or choose one themselves, including Anthropic's Claude, and CEO Sundar Pichai said nearly 90 percent of Fortune 100 companies already use Gemini Enterprise at work.
The agent connects to tools such as Google Workspace, Microsoft 365, Slack, Jira and BigQuery, and can work with any Model Context Protocol server. Its progress shows up in a tasks inbox, and it writes an audit trail attributed to the agent rather than a person. It also gets its own Workspace account and email address, so people can tag it or add it to group chats, and it will be reachable from mobile, desktop, the command line, Workspace, Microsoft 365, ServiceNow and Slack. Early testers included On, Shopify and PayPal, and Google paired the launch with new spend caps and other cost controls.
Why it matters: as Google and its rivals hand agents real objectives instead of narrow tasks, the controls built around that power, audit trails, model choice and spend caps, count as much as the capability itself. Enterprises are still weighing how much autonomy to grant AI agents in production systems, and this launch puts more weight behind governance, not less.