I would like to suggest an improvement to the Projects feature.
Currently, when working on a long-term matter within a Project (for example a legal dispute, technical project, research project, or business case), conversations inside the same Project are still largely isolated from one another. This makes it difficult to maintain continuity when documents, screenshots, contracts, emails, and other evidence have been uploaded across multiple chats.
It would be extremely useful if ChatGPT could, with the user’s permission:
Access attachments uploaded in other conversations within the same Project.
Search across all conversations belonging to that Project.
Build a chronological timeline of events from multiple chats.
Create a centralized evidence/document index.
Reference previously uploaded files without requiring the user to upload them again.
Identify duplicate documents and group related materials together.
Generate summaries and reports based on all Project materials.
A particularly valuable feature would be a “Project Evidence Library” where uploaded files are automatically cataloged and made available throughout the entire Project.
This would significantly improve workflows involving legal cases, contract disputes, software development projects, research, compliance work, and other long-running matters that require document management across many conversations.
The only thing missing is “This cannot be changed” to an option, “in-project memories only” or similar.
But OpenAI is working on “dreaming” for some users, pilfering through your chats about your pimple breakout problems and collecting them for future injection.
I think there may be a misunderstanding regarding my suggestion.
I’m not talking about accessing other users’ chats or documents within a shared business workspace.
What I mean is a situation where I, as a single user, add multiple conversations from my own history into the same Project. In that case, I would expect those conversations, attachments, screenshots, documents, and emails to become part of a shared project context.
The idea is that the AI/agent operating within that specific Project should be able to:
understand the full context of all conversations assigned to that Project,
search across those conversations,
automatically reference attachments that were already uploaded elsewhere in the Project,
build timelines and document indexes from project materials,
avoid requiring the user to repeatedly upload the same files in different chats.
In my opinion, this is different from cross-user access and mainly concerns improving continuity and document management within a single user’s Project.
As for business and team environments, I completely agree that access control is important. However, most enterprise applications already solve this through user roles, permissions, groups, and workspace-level access management. The existence of those requirements should not prevent richer project-level document awareness where appropriate permissions are in place.
My suggestion is therefore focused primarily on project intelligence and document organization rather than unrestricted access to other users’ data.
They generally are, however, when you have an extended session, a lot of that context is replaced with the focus of that session, if that makes sense.
I wanted to point out the library feature that you found on the upper left of this screenshot.
This list that populates might give you a faster read on certain files that you have already uploaded.
It is account-wide, and you have to download the file to upload it to your current session for your use case.
It’s not perfectly ideal for what you’re doing, but it will likely help you fetch troublesome files faster.
Also, consider stacking files in your project container.
By this, I mean if you have two files that are mostly text, consider combining them with the different files spaced, labeled, and a quick note at the top indexing what you labeled the different parts of the file with.
This is perhaps the best solution I can think of to keep files in project memory so they’re always there, no matter what session you’re working with.
If they are images or scans, consider combining them into a PDF so you can save even more space that way.
Doing those things will solve most of the issues you’re describing, for the most part.
