OpenAI Space gives people shared, editable Pages and files to work on, with each person bringing their own ChatGPT agent. I like where this is going. Teams can put the brief, supporting material and work in progress somewhere everyone can use.
That’s a practical improvement over passing answers between separate conversations. One person could ask their agent to work through a proposal while another checks the supporting research. They’d have a common document to improve together, with someone still responsible for deciding what’s ready.
Space replaces Library; Projects remain. You can organize Pages inside other Pages and give collaborators permission to view or edit. Sharing a Page doesn’t open your private chats or Memory to other people.
What you put on the Page does become visible. That includes personal context an agent brings into the writing. Uploaded files follow the Page’s permissions. A link to an external file keeps that service’s access rules, while a summary copied onto the Page is readable by its audience.
Training settings also follow the person whose agent is interacting with the content. Your setting doesn’t control a collaborator’s. OpenAI says Business and Enterprise data aren’t used for training by default.
The separate Dots controls govern capabilities such as computer access and app actions. Permission to read a shared Page isn’t permission to do everything described in it.
I’m enthusiastic about people working with teams of agents. A shared place for the actual work makes that easier to imagine and organize. I’d start with a clear assignment: who is working on which part, what evidence belongs alongside it, and who needs to review it before anyone acts. Space gives that collaboration somewhere to happen.