It would be good to see the ability to collaborate on Assistant threads with other users. As a first step, I think this would make most sense on Team and/or Family plans. It would follow OpenAI's fairly recent introduction of group chats for ChatGPT (see here).
A few things to note:
- Collaboration could be on a case-by-case basis: per thread; per tag; or all threads.
- Tags could be shared and synced between users, so that new threads in those tags automatically appear for all users in the shared plan.
- Non-team/family users could share threads on a case-by-case basis with other users too.
- All of these settings should be customisable for the plan admin(s), where relevant.
I haven't put too much thought into the UX of this, but queries obviously wouldn't be submit-able concurrently. The benefit of shared context when users are working on projects together would far outweigh any UX difficulties though.
Two potential difficulties:
- What would credit/token usage apportionment be like? On a Team/Family plan, it is all centralised anyway. But if there would be inter-single-user sharing, this would be more complicated.
- What would model availability look like? If one user is on Ultimate but another on Professional (even within Team/Family plans), would the Ultimate user's models be available for the Professional user to use? Or would both have to be on Ultimate?
