OpenAI announced the introduction of workspace agents in ChatGPT, a new capability that enables teams to create shared AI agents designed to handle complex, multi-step workflows across organizations. The feature marks a shift from individual AI assistance toward collaborative, agent-driven work systems operating within enterprise environments.
Workspace agents build on earlier GPT-based tools by enabling persistent, cloud-based execution of tasks that continue even when users are offline. Powered by Codex, these agents can perform a wide range of functions, including generating reports, writing and executing code, managing communications, and coordinating workflows across teams and systems.
The new capability is designed to address a key limitation in existing AI tools, the lack of shared context across teams. Workspace agents can gather data from connected systems, follow predefined processes, request approvals when necessary, and manage handoffs between teams, enabling more seamless execution of complex organizational tasks.
Organizations can create agents by describing workflows within ChatGPT, which then guides users through building, testing, and deploying the agent. Once created, agents can be shared across teams and integrated into platforms such as Slack, allowing them to operate directly within existing workflows and communication channels.
OpenAI highlighted several use cases already in deployment, including agents that automate software request reviews, route product feedback, generate weekly business reports, qualify sales leads, and manage third-party risk assessments. These agents are designed to transform repetitive, time-intensive tasks into automated processes while maintaining human oversight.
The platform also introduces a system for turning institutional knowledge into reusable workflows. Agents can retain memory, improve over time through usage and feedback, and be shared across organizations, enabling teams to standardize processes and scale best practices more effectively.
Enterprise controls and governance are central to the rollout. Administrators can define permissions for data access, tool usage, and actions, while requiring approval for sensitive tasks. Monitoring tools provide visibility into agent usage, performance, and configuration, and a compliance API enables oversight of agent activity across the organization.
Workspace agents are currently available in research preview for ChatGPT Business, Enterprise, Edu, and Teachers plans, with free usage available through early May 2026 before transitioning to a credit-based pricing model.
The launch reflects a broader evolution in enterprise AI, moving from tools that assist individuals to systems that actively coordinate and execute work across teams. By embedding agents directly into workflows and communication channels, OpenAI is positioning workspace agents as a foundational layer for the next generation of organizational productivity.
