They love it… but I don’t have any idea how to do it… There is no source where I can find tutorial steps. So it will be really, really hard for me. I ask claude do an investigation , there is the result. if you have something extra to add, it will be great .
------- Claude findings-----
The docs never use the word “multiplayer”; that’s the video’s name for it. The closest docs page is Multi-user mode, and it rests on several features that already existed.
You were right, is not one unique feature: it’s built from several features
Feature
What it does, simply
Docs
Shared sessions
One AI conversation that many people can open, read and add to
Update: I’m starting to build this at my company, and I have a question about how the AI model account (the “brain”) works when several people use one Gateway. I want to make sure I understand it before I set it up.
My current setup (personal, works great)
OpenClaw Gateway runs on my Mac mini.
The model is my own ChatGPT subscription, connected with openclaw models auth login --provider openai.
I connect from my Mac mini and my MacBook. Only me.
What I want to do at work
One Gateway on a Google Cloud VM.
A small group of developers (3–5) connect to it, each signed in as themselves, so the multiplayer features work (owners, mentions, presence).
Everyone chats with the same agents.
My understanding (please correct me if I’m wrong)
If a teammate connects to my Gateway and asks the agent something, the agent answers using the Gateway’s default model account. In my case that would be my ChatGPT subscription.
Technically that works, but OpenAI’s Account Sharing Policy says one account = one person. So letting teammates use my subscription is not allowed, even though OpenClaw doesn’t block it.
The right way is that each teammate opens Settings → Profile → Connected accounts and connects their own ChatGPT account. Their new chats then use their own subscription.
My questions
Are points 1–3 correct?
The docs say selecting a personal account “does not guarantee that every turn bills that account”. When would a teammate’s message still use the Gateway’s default account (mine)? Is there a way to make sure it never falls back to my subscription?
In a shared session, when person A started the chat and person B sends a message, whose account answers B’s message: A’s, B’s, or the Gateway default?
For a small company team, what do you recommend:
(a) each person connects their own ChatGPT subscription, or
(b) one company OpenAI/Anthropic API key as the Gateway default?
What do I understanding wrong? Thanks! Once it’s working, I’ll share my step-by-step setup notes here so others can follow them.
My understanding was mostly right, but the key nuance is that selecting a personal account is a preference, not a billing guarantee. A teammate’s new chat prefers their connected account, but an existing shared session keeps its selected account, and same-provider failover can still use another eligible profile. For a company deployment, I’ll use a company-owned OpenAI API key as the Gateway/agent default and let developers connect personal accounts only for their own new chats.
It is quite a lot to learn and since my company is not a developer company, I don’t know how I’d use it yet. We’re the kind of company that if we ever have to do a “meeting” something is going not quite right.
Apologies for the naming conventions being a bit off. We only recently settled on the term “Multiplayer” for this feature, so expect to see some rectification of our docs in short order to reflect.