Rather than operating as a separate assistant in each app, Google says the work agent keeps context between tasks and access points.
One agent across apps and devices
Kurian said Gemini is web-based, can run on any device, integrate with third-party applications and operate without a dedicated interface. Because it runs in the cloud, Google says it maintains one set of memories, context and a personalization graph wherever a worker accesses it.
The agent can also create sub-agents for multi-step jobs and operate as a coworker with its own role and identity. Kurian said Gemini can choose among multiple models to improve quality and lower costs.
Google’s official news account said Gemini will work directly inside Gmail, Drive, Docs, Slides, Sheets, Chat and Calendar, carrying the same memory, skills and controls across those Workspace apps. Google describes the agent as able to answer questions, handle knowledge work, create images and other media, and write and run code.
Data skills and enterprise controls
Google also announced data and analytics skills that let technical teams and other business users ask questions in natural language. Specialized tools, connectors and knowledge for financial services and legal teams are rolling out, according to the company.
The company says built-in governance gives IT teams enterprise data protection, privacy controls and central oversight. Those are Google’s product claims, not independently tested findings.
During the Gemini at Work event, AI podcaster Sam Charrington reported seeing a demonstration in which Gemini Agent built a machine-learning model and trained it on Spark. He also reported that administrators can govern interactions between users and agents while applying policies and budgets across an organization.
Google’s announcement did not give a broad availability date or pricing. The industry-specific tools were described as rolling out, leaving the timetable for many organizations unclear.
