OpenAI Launches Workspace Agents for ChatGPT to Automate Multi-Step Team Workflows
GPT
Enterprise AI
AI Agent
OpenAI launched workspace agents in research preview for ChatGPT Business, Enterprise, and Edu plans. These shared AI systems handle complex, multi-step tasks and long-running workflows across an organization's software stack. Unlike standard assistants, these agents operate independently on schedules or in the background while users are away.
This update marks a transition from conversational AI to agentic AI (autonomous systems that plan and execute multi-step goals) within the enterprise. It builds on the parallel subagent architecture and follows the secure sandbox execution recently introduced for developers. Organizations can now standardize automated workflows across teams rather than relying on individual custom GPTs.
You can build agents that pull context from docs and chats to take approved actions like updating Linear issues. Admins manage these systems through a central console with audit logs and approval gates. The feature is currently available for ChatGPT Business, Enterprise, Edu, and Teachers subscribers.
Frequently asked questions
- What are OpenAI workspace agents?
- Workspace agents are shared AI systems in ChatGPT designed to automate entire multi-step workflows across a team. Unlike standard chatbots that respond to individual prompts, these agents can pull context from various enterprise tools, take actions like updating tickets or sending messages, and continue running in the background or on a set schedule without constant human direction.
- Which ChatGPT plans include workspace agents?
- Workspace agents are currently available in a research preview for users on ChatGPT Business, Enterprise, Edu, and Teachers plans. This rollout allows organizations to test the autonomous capabilities of these agents within their specific environments. Access is managed through the organization's workspace, where admins can control who has the ability to build and deploy them.
- How do workspace agents interact with other apps like Slack or Google Drive?
- These agents use integrations to gather information and execute tasks across different platforms. For example, an agent can monitor a Slack thread to understand a request, pull relevant data from a document in Google Drive, and then update a task in Linear. They operate across these tools to complete end-to-end processes that previously required manual coordination.
- What security and admin controls are available for workspace agents?
- OpenAI provides centralized management for admins to oversee agent activity. This includes role-based access control to determine who can build agents and which tools they can access. Admins can also review detailed audit logs of agent actions and implement approval gates, which require a human to verify and sign off on sensitive actions before the agent proceeds.
- Can workspace agents run while I am offline?
- Yes, workspace agents are designed to handle long-running workflows and recurring tasks independently. They can be set to run on specific schedules or continue working in the background after a user initiates a process. This allows them to handle time-consuming work like summarizing support requests, reviewing leads, or generating reports without requiring the user to stay active.
