Persistent Multi-GPT Collaboration Inside ChatGPT Projects

I’d like to suggest a feature for ChatGPT Projects: the ability to add multiple Custom GPTs as persistent members of the same Project conversation.
The simplest analogy is a group chat.
The Project would provide the shared environment: common files, Project instructions, conversation history, and context. Each Custom GPT would remain a distinct participant, preserving its own instructions, specialization, personality, and knowledge.
Instead of manually calling one GPT at a time, the user could create a persistent group of GPTs inside the Project. They could read each other’s contributions, respond to one another, challenge ideas, refine proposals, and collaborate toward a shared result.
Ideally, the user could also choose how the group is orchestrated: all GPTs may participate, only relevant ones may be called automatically, one GPT may act as moderator, or the user may explicitly invite specific GPTs into a discussion.
The key feature would be persistence: once added, the GPTs remain members of that Project space rather than being invoked only for a single interaction.
This could combine two of ChatGPT’s strongest concepts:
Projects as a shared, persistent workspace
Custom GPTs as specialized agents with distinct roles and knowledge
Together, they could become a native multi-agent collaboration environment.
I am currently reproducing this workflow manually across several Custom GPTs. The main limitation is that I have to act as the human orchestrator, copying information from one GPT to another and deciding who should speak next. With many specialized GPTs, this quickly becomes difficult to manage.
A persistent multi-GPT Project would remove that bottleneck and could be extremely useful for complex professional workflows, software development, research, creative teams, consulting, and many other use cases.
Thank you for considering it.
— Adriano