Gemini Enterprise agent announced to handle work tasks in one interface

On 8 October 2026, at the Gemini at Work event, Google announced an AI agent that will run in the Gemini Enterprise app and handle tasks across different work applications. The company said it would initially focus on businesses. According to The Verge, the tool is currently available in private preview for enterprise customers.

The agent works toward goals, not just instructions

The new system can do more than answer questions: it can plan work toward a given goal, use custom skills and tools, and connect to internal systems. According to Google Cloud CEO Thomas Kurian, users can define an outcome they want the agent to achieve instead of giving it step-by-step commands.

The agent automatically selects the right model for a task. Users can also choose a model themselves; third-party options initially include Claude models. Google says it plans to add more open-source and other proprietary models to the selector in the future.

Users can attach files, folders or other content related to a specific workflow to their requests. In addition to Google Workspace and Microsoft 365, the Gemini agent can connect to services such as Slack, Jira, Confluence, Git, BigQuery, Databricks, Postgres and Snowflake. It also supports working with internal or external Model Context Protocol (MCP) servers.

How will task tracking and auditing work?

The agent will have its own Workspace account, email address and working context. According to Google, the system will be able to use company information such as teams, approval processes and calendars. Users will be able to assign tasks to the agent by tagging it, emailing it, sharing content or adding it to a group chat.

The task inbox will be used to track what the agent is doing. It will show task progress, delegation to sub-agents and the use of custom skills. Actions will be recorded in an audit trail under the agent’s identity, rather than on behalf of an employee. Google also describes cost-control options for businesses, including multi-model use, intelligent routing and real-time spending limits. The sources do not include product pricing details.

The agent will be accessible from mobile devices, Windows and Mac computers, the command line, and services such as Google Workspace, Microsoft 365, ServiceNow and Slack. According to TechCrunch, On, Shopify and PayPal are among the early testers. No timeline has been announced for broader consumer availability.

Impact on architecture and visualization workflows

For architecture firms and visualization teams, the announcement is not directly about modeling or producing renders; it is about coordinating tasks across different applications through a single agent. Helping teams track work spread across files and communication channels is one potential use case. However, the announcement does not mention dedicated integrations for Revit, Archicad or other 3D production software.

Google’s developer documentation says Gemini Enterprise agents can connect to Workspace data and offers both no-code and code-based options for building custom agents. This provides a framework for organizations to configure agents around their own data and processes, but it does not mean the newly announced agent is compatible with specific architecture software. Before adopting it, teams should verify the scope of integrations, access permissions, audit requirements and licensing terms. Private preview availability is also a limiting factor.