Feature Request: Shared Project State Across Multiple Chats
I frequently work on larger projects using several ChatGPT conversations in parallel.
For example, one chat may focus on backend development, another on licensing, another on a Windows client, another on testing, and another on business logic.
The current Projects feature helps organize chats and provides some shared context, but what is still missing is a persistent and structured shared project state that all chats inside the same project can read and update.
Today, important decisions, implementation progress, test results, dependencies, bugs, versions and architectural changes are often trapped inside individual conversations.
This creates unnecessary manual work. Users must repeatedly copy summaries between chats, explain the current project status again, or rely on memory that is not precise enough to function as a technical project state.
A useful solution would be a shared Project State available to every chat inside a Project.
It could contain:
- current project status and milestones
- completed and open tasks
- architectural decisions
- frozen specifications or contracts
- component versions
- test results
- known bugs
- dependencies between modules
- next steps
- important changes made in other chats
Each conversation could read this shared state before answering and, with user approval or automatically according to configurable rules, update it after meaningful work is completed.
Example:
Chat A finishes License Manager v0.4 and updates the shared project state.
Chat B, which is working on a Windows client, immediately knows that the server contract is now finalized and can continue implementation without the user manually transferring information between conversations.
This would be especially valuable for software development, engineering, research, business projects and any long-running work where several specialist conversations are used in parallel.
The key distinction is:
Memory remembers the user.
Project State remembers the project.
Ideally, users should also be able to inspect, edit, lock or revert the shared project state, so they remain in control of what becomes the project’s source of truth.
This would significantly reduce repetitive prompting and make ChatGPT Projects much closer to a real persistent working environment.