❓ Problem

AI has quietly become the most-used tool in every company and the least visible one. Your designer has 40 Claude threads. Your marketer has 60 in ChatGPT. Your ops person has a folder of agent runs nobody else has ever seen. All of it sits inside personal accounts, behind personal logins, on personal billing. The company's best thinking is trapped in a browser tab that only one person can open.

The scale of this is not a rounding error. OpenAI now has 7 million enterprise workplace seats deployed and Anthropic serves 300,000+ business customers, yet a huge share of actual workplace AI runs outside any of it. LayerX's 2026 usage report found roughly 1 in 7 AI interactions inside corporate identities are powered by personal licences, and only 30% of organisations have full visibility into how their people use AI at all.

Here is the part that makes it a product rather than a policy memo. Even the sanctioned team plans do not solve it. ChatGPT Team and Claude Team give you a shared billing account and shared custom GPTs, but the chats themselves stay private by default. You cannot read your colleague's thread. You cannot fork it. You cannot jump into a running agent and redirect it. The "team" plan is really a procurement plan.

And the demand signal is now public. Oliur, the designer behind UltraLinx, tweeted asking for exactly this for his three-person team and finished with the line every founder should be scanning for: if it does not exist, he may have to build it himself. Meanwhile Y Combinator built its own internal version, called QM, and open-sourced it on 31 July under MIT. It is excellent, and it is also a Node, Fastify and Postgres deployment that expects you to init against Fly.io or AWS. A three-person design studio does not want to become a DevOps shop. They want a URL.


✅ Solution

A hosted, multiplayer AI workspace organised by projects, where every teammate's chats and agents are visible, readable and joinable by default.

The wedge is deliberately small: make a three-person team's AI work legible to each other. The platform underneath, shared context plus agent orchestration plus connector governance, expands into the whole AI operations layer for a company.


📊 Key Numbers

Market size