Introducing workspace agents in ChatGPT
OpenAI ● Covered by 3 sources
OpenAI just launched workspace agents in ChatGPT, powered by Codex, that run in the cloud to handle multi-step work automatically. They're built for teams, not solo chats, meaning ChatGPT is inching toward being an actual coworker.
Based on reporting by OpenAI — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
OpenAI is pushing ChatGPT further into the workplace with a new feature it calls workspace agents. Built on Codex, the company's coding-focused model, these agents are designed to take on complex workflows rather than just answer questions or draft text. Instead of a single back-and-forth exchange, the idea is that you hand off a task and the agent works through it on its own, executing multiple steps without needing you to babysit each one.
What makes this notable is where the work happens. These agents run in the cloud, not just inside a browser tab waiting for your next prompt. That distinction matters for teams that need something running in the background while people move on to other things. It's less like chatting with an assistant and more like assigning a task to a junior teammate who reports back once it's done.
OpenAI is also framing this squarely around teams and tool integration. The pitch is that workspace agents can reach across the various apps and systems a company already uses, stitching together actions that would otherwise require a human to jump between tabs and manually connect the dots. Security gets a specific mention too, which suggests OpenAI knows enterprise buyers will ask hard questions before letting an autonomous agent touch internal systems.
This fits a broader pattern from OpenAI over the past year: moving Codex and agent-style tools out of individual developer use and into something IT departments can deploy at scale. Workspace agents look like an attempt to turn ChatGPT from a smart chatbot into infrastructure that companies build processes around, not just a tool employees open when they're stuck.
My take — AI-written commentary, not fact-checked reporting
Every major AI lab is racing to turn chatbots into background workers, and OpenAI folding this into ChatGPT rather than a separate product tells you where the real money is: enterprise contracts, not consumer subscriptions. I'd watch the security claims closely, because 'runs in the cloud across your tools' is exactly the kind of feature that sounds great in a slide deck and terrifying to an actual security team.
Read more about this at: OpenAI