Both ChatGPT and Claude now let a team share a project: the same files, the same instructions, the same AI. If your team already pays for one of them, this is the first place to look for shared AI context. But the two work differently in one way that matters a lot for teams - whether you can see each other's conversations.
How ChatGPT shared projects work
- Sharing: you can invite people one by one, invite a group, or share a link. Each person gets chat access or edit access. On Business, Enterprise and Edu workspaces you can share a project with the whole workspace by link, and a project can have up to 100 collaborators (OpenAI help).
- What's shared: ChatGPT can draw on everything in the project - its chats, uploaded files and custom instructions - so one person can pick up where another left off.
- Memory: a shared project uses project-only memory. It remembers what members tell it inside the project, never reads anyone's personal memories, and stays that way permanently once shared.
How Claude Projects sharing works
- Who can share: Team and Enterprise plans (Claude help).
- Visibility: a project is Public (everyone in the organization can view and use it) or Private (only invited members).
- Permissions: Can view lets someone read the project and chat in it; Can edit lets them change the instructions, knowledge and members.
- What's shared: the instructions and the knowledge base. Chats are not - Anthropic's help page says your chats in a project are private and inaccessible to other members unless you share a specific chat.
The difference that matters
In ChatGPT, a shared project's conversations feed the shared context. In Claude, everyone shares the same documents and instructions, but each person's conversations stay their own.
Neither is wrong. Claude's approach keeps personal drafts private; ChatGPT's lets work carry over between people. Pick based on how your team works:
- Your team mostly needs the same reference material (a style guide, product docs, a brief): either works well.
- Your team needs to build on each other's AI work (one person researches, another writes, a third reviews): ChatGPT's shared chats, or a shared channel, fit better.
What both are not
A shared project is still a set of separate conversations. Two people asking about the same thing usually do it in two different chats, and the AI answers each one alone. That's fine for reference work. It's less good for a team making decisions together, where you want everyone - and the AI - in the same conversation, seeing the same answers.
Where a shared team channel is different
In a channel-based tool like Teamski, the AI agent sits in the team's channel. Everyone talks in one conversation, the agent sees all of it, and it remembers the project's decisions across channels. Because it lives where the team works, it can also run tasks in the background, post scheduled summaries, and act in tools like GitHub or Linear after someone approves. See what multiplayer AI is.
Which should you use?
- Already on ChatGPT Business or Claude Team and happy with it? Use their shared projects - they're good, and you already pay for them.
- Want the whole team in one live conversation with the AI, with background and scheduled work, on free models? That's what Teamski is for.
Features in both products change often. Check each one's current help pages before you decide.
Try Teamski free with your whole team.
A shared AI teammate that lives in your team's channels. Free to start, unlimited people.