today, openai published a blog post announcing the launch of workspace agents in chatgpt, enabling teams to create collaborative agents that can handle complex tasks and long-term workflows.
this feature addresses key challenges in cross-time-zone and cross-tool collaboration, powered by codex at its core. it can manage a wide range of office tasks, from report preparation and code writing to message responses. the agents run in the cloud, supporting round-the-clock operation and ensuring continuous task execution even when employees are offline.
in terms of technical architecture, workspace agents offer capabilities for document processing, code execution, tool integration, and memory storage. openai states that this goes beyond simple question-and-answer interactions, allowing for multi-step, complex operations. the system supports scheduled automation and slack integration; for example, product teams can build agents that automatically answer employee questions in slack channels, link to relevant documents, and submit service tickets.
internally, openai has already deployed five use cases for workspace agents, including software reviewers, product-feedback routers, and weekly-metrics reporters. notably, the sales team uses an agent to automatically integrate call logs, filter leads, and draft emails, significantly reducing manual effort.
in terms of knowledge management, this feature transforms tacit knowledge scattered across systems and personnel into explicit workflows. for instance, an agent built by the accounting team can complete month-end closing procedures in just a few minutes and generate working papers that meet internal-control requirements. these agents possess memory and iterative optimization capabilities, continuously improving their performance through interactive feedback, thereby serving as an effective vehicle for capturing and codifying team knowledge.
currently, this feature is available in research preview for users on the chatgpt business, enterprise, edu, and teachers plans, with free usage until may 6, 2026, after which it will transition to a credit-based pricing model. openai plans to add features such as automatic triggers and performance-analytics dashboards in the future, further reducing team-collaboration costs.
