OpenAI Workspace Agents

OpenAI has started rolling out workspace agents in ChatGPT, letting teams build shared agents that run multi-step work across company tools in the background. (openai.com) OpenAI said the product launched April 22, 2026 in research preview for ChatGPT Business, Enterprise, Edu, and Teachers plans. The company said the agents can run in ChatGPT or Slack and are powered by Codex. (openai.com) The new agents are built for repeatable office tasks rather than one-off chats. OpenAI said teams can connect apps such as Google Drive, Google Calendar, Slack, and SharePoint, then schedule recurring runs or trigger work inside connected Slack channels. (help.openai.com) OpenAI’s pitch is that a team can build one agent once and share it across the workspace, instead of each employee making a separate custom bot. The company said agents can gather context from connected systems, follow team processes, ask for approval when needed, and keep working when the user is offline because they run in the cloud. (openai.com) That pushes ChatGPT further from a chat window toward an internal automation layer for companies already using Slack, document stores, calendars, and ticketing systems. OpenAI’s business page says the agents can update tickets, edit documents, send messages, and generate reports without step-by-step prompting. (openai.com) The product also extends OpenAI’s broader “agent” push from 2025, when the company introduced ChatGPT agent for individual users. In that July 17, 2025 launch, OpenAI said ChatGPT could browse sites, run code, analyze files, and complete tasks on a virtual computer with user approval for consequential actions. (openai.com) Workspace agents add admin controls that matter in corporate rollouts. OpenAI said Enterprise workspaces have the feature off by default at launch, admins can enable it for eligible workspaces, and Enterprise customers using Enterprise Key Management are excluded at launch. (help.openai.com) OpenAI also said admins can control who builds and uses agents, which apps and actions are allowed, and when human approval is required before an agent sends a message or updates a record. The company says audit logs and monitoring are available so organizations can review what an agent did. (openai.com) For teams already using custom GPTs, OpenAI said those GPTs will remain available during testing and that conversion tools are coming later. The company’s release notes say workspace agents are rolling out over the next few weeks, which means availability may vary by workspace even after the April 22 announcement. (openai.com; help.openai.com)