Google had already tested the waters in May with Gemini Spark, a consumer-facing agent designed to carry out tasks across Workspace and the web. But that product lived in a separate Gemini interface; the company’s latest move is a more ambitious attempt to put one agent at the centre of work.¹

On Thursday, at its Gemini at Work event, Google launched a unified Gemini agent for enterprise customers in private preview. Sundar Pichai cast it as “a single, universal agent for work” with access to a company’s business context—an assistant meant to answer questions, handle knowledge work and create across connected tools.Sundar Pichai (sundarpichai)²

The proposition is broader than a chatbot. The cloud-based agent can operate across Gmail, Drive, Docs, Sheets and Calendar, while also connecting to services including Slack and Microsoft 365. Google says it retains context across devices and can use specialised sub-agents, effectively turning a request from a phone, desktop or web browser into work carried out in the background.¹

Google Cloud chief Thomas Kurian described the dividing line in stark terms: customers can give Gemini “objectives, not just instructions.” The agent is designed to plan the work, use tools and skills, and connect with internal systems such as Jira, Confluence, Git, BigQuery and Snowflake.³

That autonomy is also where the stakes rise. Gemini will receive its own Workspace account, email address and enterprise context, allowing colleagues to tag it, email it or add it to group chats. Its actions are logged in an audit trail under the agent’s identity rather than an employee’s.³

For now, Google is keeping the rollout business-first. Pichai said the approach gives the company room to tackle the “harder problems around security, scale, and performance” before taking the technology to consumers.³