OpenAI has launched “workspace agents” for ChatGPT, a new tier of collaborative tools designed to allow enterprise teams to build and share autonomous AI workflows.
The workspace agents utilize the Codex model to execute code-heavy tasks and integrate directly with team-specific data repositories. Unlike previous standalone GPTs, these agents are designed for multi-user interaction, allowing an entire department to utilize a single, persistent agent for complex project management or software development. The release signals OpenAI’s intent to move beyond individual consumer tools and capture the enterprise market for collaborative, autonomous digital labor.
VentureBeat framed OpenAI’s new workspace agents as the start of “the era of the digital coworker,” arguing that the product moves ChatGPT from one-off chats toward persistent, reusable agents that teams can share and govern across enterprise workflows.
While not responding to the OpenAI news, Okta strikes a more cautious note regarding enterprise deployment of AI agents. In a Wednesday post titled “CISO’s Guide: Essential AI Agent Security Questions” Okta warned that as enterprises roll out AI agents, the core issue is no longer adoption but runtime control. AI agents blur the lines of traditional identity systems because they can act autonomously, call tools and touch enterprise data in ways legacy IAM was not designed to govern, the post stated.
