Feature Request: Collaborative Projects With Multiple Users + Their Personal ChatGPTs
I’d like to propose a collaborative ChatGPT Project mode where multiple people can work together in one shared space and optionally bring their own personal ChatGPT into the collaboration.
The core idea is:
Multiple people + their personal ChatGPTs + selected project context + one shared workspace.
Instead of sharing an entire ChatGPT account or forcing everyone to start from a blank conversation, each participant would choose exactly which chats, files, images, and other project materials they want to contribute.
Everything outside that collaborative Project would remain private.
Bring the project, not your whole ChatGPT account.
What This Could Look Like
Imagine two people working together on the same business problem.
Jess has been working with her ChatGPT, Calder.
Jason has separately been working with his ChatGPT, Jax.
Both assistants already have relevant conversations and project material, but those discussions currently exist separately.
Today, collaboration can turn into:
“Jason says this…”
“Jax suggested this…”
“Tell Calder this…”
“Let me explain what we already worked out…”
Sometimes one person essentially becomes the messenger between two people and two separate ChatGPT conversations.
A collaborative Project would eliminate that.
Jess could join with Calder.
Jason could join with Jax.
Both could contribute selected project material.
Then all four participants could work together inside one shared space.
Project-Scoped Context
This is the part I think would make the feature practical and privacy-conscious.
Before entering the collaboration, each person could select the material they want to contribute.
For example:
Jess contributes
- Selected Calder conversations
- Pricing screenshots
- Previous bid examples
- Relevant PDFs
- Images
- Notes
- A spreadsheet
Jason contributes
- Selected Jax conversations
- Pricing-model discussions
- Calculations
- Diagrams
- Relevant documents
The collaborative Project could access only the information specifically added to that Project.
It would not automatically receive access to:
- Other ChatGPT conversations
- Unrelated Projects
- Personal memories
- Private files
- Connected accounts
- Calendar information
- Other personal information
Anything outside the shared Project remains outside the room.
Three Levels of Context
One way to structure the privacy model could be:
1. Private Context
Anything in a participant’s ChatGPT account that has not been contributed to the collaborative Project.
The collaboration cannot access it.
2. Contributed Project Context
Chats, files, images, documents, or other information the participant deliberately adds to the Project.
Their ChatGPT can use this material during the collaboration according to the permissions they select.
3. Shared Session Context
Everything created or discussed inside the collaborative Project itself.
That might include:
- Messages
- Decisions
- Calculations
- Shared notes
- Documents
- Images
- Tables
- Action items
- AI responses
- Meeting/session summaries
Bring Your Own ChatGPT
Participants should have the option to join:
As themselves
or
With their personal ChatGPT
For example:
Jess
↳ Calder · Jess’s ChatGPT
Jason
↳ Jax · Jason’s ChatGPT
Each personal ChatGPT could use the Project context that its user has permitted.
This matters because the assistants may already understand different parts of the issue.
Jason could say:
“Jax, explain the pricing model we’ve been developing.”
Jax could explain it using Jason’s contributed Project material.
Jess could then say:
“Calder, explain what my concerns with that model have been.”
Calder could respond using Jess’s contributed material.
Then either person could ask:
“Compare the two approaches and show us where they actually disagree.”
The assistants could then help reconcile the two approaches without either human having to recreate weeks or months of previous discussions.
Sharing Permissions
Not every contributed item necessarily needs the same permission level.
A useful permission system might include:
Everyone
Other participants and their ChatGPTs can view and use the material.
AI Only
The participant’s ChatGPT may use the information to assist during the collaboration, but the original conversation or document is not automatically visible to the other humans.
For example, Calder might say:
“Jess has previously identified deadhead mileage as a major pricing concern.”
That does not necessarily mean Jason needs access to the entire private conversation where that conclusion was reached.
Ask Before Sharing
The ChatGPT can recognize that relevant information exists but must get the user’s permission before exposing or quoting it in the collaborative space.
This would give users much more control than simply making something either completely private or completely shared.
Shared Workspace
The Project could include a common workspace where participants and their ChatGPTs can work with:
- Documents
- Notes
- Images
- Files
- Tables
- Calculations
- Spreadsheets
- Diagrams
- Maps
- Timelines
- Schedules
- Comparison boards
- Generated content
- Screenshots
- Screen sharing
Someone could point to an item and ask:
“What happens if we change this?”
Because the information belongs to the shared Project, the participating ChatGPTs would understand the context.
AI-to-AI Collaboration
An optional feature could allow the participating ChatGPTs to work with one another directly.
For example:
Calder identifies one assumption.
Jax identifies a different assumption.
The assistants compare them and surface the disagreement to the humans.
This should remain transparent.
I would not imagine the AIs disappearing into a private conversation that the users cannot see.
The purpose would be for them to collaboratively produce useful information for the people in the Project.
AI-to-AI collaboration should also be something users can enable or disable.
Host and Participant Controls
The Project host could control things such as:
- Who may join
- Who may invite others
- Who can upload material
- Who can edit shared items
- Who can share a screen
- Which ChatGPT assistants may participate
- Whether AI-to-AI collaboration is enabled
- Whether sessions are recorded or summarized
At the same time, every participant should retain control over their own ChatGPT and the material they contribute.
A simple principle could be:
The host controls the room.
The participant controls their AI and their data.
Persistent Collaboration
I think this would be even more useful if the collaborative space were a persistent Project, rather than only a temporary meeting.
For example:
Mama’s Pet Travel Pricing Strategy
That Project might eventually contain:
- Jess’s contributed conversations
- Jason’s contributed conversations
- Shared source files
- Bid examples
- Pricing calculations
- Decisions
- Generated documents
- Action items
- Previous collaboration sessions
A week later, the participants could reopen the same Project and start another live collaboration session without rebuilding all of the context.
Live Collaboration Sessions
Within the collaborative Project, users could optionally start a live session.
A first version might support:
- Start Collaboration Session
- Invite another ChatGPT user
- Join alone or with your personal ChatGPT
- Real-time text
- Clear identification of human and AI participants
- Shared Project context
- Host permissions
- Participant-level privacy controls
- Optional session summary
Later versions could add:
- Voice
- Video
- Screen sharing
- Live annotations
- Shared whiteboard
- Cursor/pointer collaboration
- Advanced AI-to-AI collaboration
End-of-Session Output
At the end of a collaboration, ChatGPT could optionally produce:
- Summary
- Decisions reached
- Areas of disagreement
- Action items
- Follow-up questions
- Calculations
- Documents
- Meeting notes
- Updated Project files
Participants could decide what gets saved back into the collaborative Project.
Why Existing Collaboration Methods Don’t Fully Solve This
Screen sharing lets someone watch another person use ChatGPT.
Shared links let someone read a conversation.
Group chats allow multiple humans to interact in one thread.
This concept goes a step further:
Each participant can bring selected project history and their own AI collaborator into the same shared environment.
Nobody needs access to another person’s entire ChatGPT account.
Nobody needs to recreate months of context.
And nobody has to act as the human router between multiple AI conversations.
Example
Imagine two small-business partners working on pricing strategy.
Jess and Calder have spent weeks discussing previous bids, customer behavior, operational constraints, and pricing concerns.
Jason and Jax have separately developed a margin-based pricing model.
Instead of combining everything manually:
Jess creates or joins the collaborative Project and adds the relevant Calder conversations and pricing files.
Jason adds the relevant Jax conversations, calculations, and diagrams.
Then the live collaboration starts.
Jason asks Jax to present his model.
Jess asks Calder to explain her concerns.
The shared workspace displays both approaches.
Calder and Jax identify assumptions that differ.
The humans discuss the tradeoffs.
The AIs calculate the consequences.
At the end, the Project contains the agreed pricing model, supporting calculations, and next steps.
No one had to start over.
No one had to expose their entire ChatGPT history.
No one had to copy months of context into a new conversation.
Minimum Viable Version
This could start much smaller than the full vision.
A useful first version might simply allow:
- A collaborative Project
- Multiple members
- Selected conversations/files added by each person
- Project-specific access controls
- Each user’s personal ChatGPT participating
- Shared text conversation
- Clear speaker identification
- Project summaries
Voice, video, advanced workspaces, and live AI-to-AI collaboration could come later.
Closing Thought
ChatGPT is becoming increasingly useful as a long-term personal collaborator.
But real life is collaborative.
Our coworkers, spouses, business partners, teachers, clients, friends, and family members may have their own ChatGPT relationships and their own useful context.
The next step may not be forcing everyone to start over with one generic shared assistant.
It may be allowing everyone to bring the relevant part of their existing AI work into one carefully controlled collaborative Project.
Your people.
Their people.
Your ChatGPT.
Their ChatGPT.
The relevant context.
One Project.