OpenAI

OpenAI Launches Workspace Agents for Collaborative Workflow Automation in ChatGPT


Executive Summary

OpenAI has introduced workspace agents, an evolution of GPTs, for ChatGPT Business, Enterprise, Edu, and Teachers plans. These shared, cloud-based agents are designed to handle complex, long-running business workflows by integrating with various tools, following team processes, and operating continuously. The feature allows teams to automate tasks like reporting, lead qualification, and IT support, turning organizational knowledge into reusable, automated processes with enterprise-grade administrative controls.

Key Takeaways

* Product: Workspace agents, powered by Codex, are shareable AI agents for automating team-based workflows.

* Functionality: Agents run autonomously in the cloud to handle multi-step tasks, gather context from different systems, write/run code, ask for approval, and operate across tools like Slack.

* Creation: Users can create agents by describing a workflow in natural language, using templates, or providing files. The system helps define steps and connect tools.

* Use Cases: Examples include agents for software review requests, product feedback routing, weekly metrics reporting, lead outreach, and third-party risk management.

* Target Audience: Teams using ChatGPT Business, Enterprise, Edu, and Teachers plans.

* Enterprise Controls: Admins can manage access, control which tools can be connected, and monitor agent activity via a Compliance API. Built-in safeguards protect against prompt injection.

* Availability & Pricing: Available now in a research preview. The feature is free until May 6, 2026, after which credit-based pricing will apply.

Strategic Importance

This launch positions OpenAI more deeply into the enterprise software ecosystem, shifting from a focus on individual productivity to automating complex, collaborative business processes. It directly competes with other workflow automation platforms by offering a more integrated and conversational approach to building autonomous agents.

Original article