Why The Wheel exists.
AI gives everyone access to what the world already knows. The Wheel is for what only you and your team know: the decisions, the drafts, the relationships, and the private context that makes the work yours.
I kept coming back to one question.
What would you be able to do if you never forgot anything? Not in the abstract. In the work: every decision, source, conversation, draft, and dead end available the moment it mattered.
I did not want a better chatbot. I wanted a way to work from the full record without handing that record to every tool that might help.
So I built The Wheel around that constraint: knowledge should compound, and it should stay under the control of the person or team that earned it.
Welcome to The Wheel.
The model should not be the place where your record lives.
Satya Nadella calls it the reverse information paradox: you pay for intelligence, then hand over the knowledge that makes it useful. What a company creates through AI use should compound for the company, not for the vendor.
The Wheel is not a model. It is the platform around models: a private record that outlasts any one tool, opened by keys nobody holds. You decide who gets access, to what, and for how long.
Privacy was never about hiding. It was always about choosing. And that choice shouldn’t be sold as an enterprise feature.
Privacy is exercised control.
Privacy is a human right because people need agency. The point is to decide what can be revealed, to whom, under what conditions, and for how long.
Consent needs architecture.
Policies arrive after data has already moved. The system should start with control: encrypted records, access you choose to give, and the ability to take it back without relying on vague trust.
Keep your record outside the model.
Your record is bigger than a chat history. Notes, documents, sources, and decisions need a home that can outlast any one model or destination.
AI should raise the floor.
AI is at its best when it reduces noise, translates unfamiliar domains, and lowers the cost of first attempts, so more people can enter the room. It should not pretend to think for them.
Judgment stays human.
AI generates options. It does not decide what matters. The person still redirects the work, catches false confidence, and owns what ships.
Built inside its own loop.
Every feature started as a concrete question: how can a system search useful meaning without making the server the owner of the underlying record? Not a slogan. A constraint. Failed approaches didn’t disappear into a folder; they became part of the next answer, because the record of building The Wheel lived inside The Wheel.
The point is not that everyone should build alone. It’s that people and teams should carry their own context forward instead of starting over every time the tool, model, or problem changes.
Alie Cohen
Alie Cohen has spent 25 years building where sensitive information, difficult permissions, and human trust collide: New York’s statewide health information exchange, authorization APIs spanning 300 million patient records, and multi-tenant isolation across a venture studio portfolio.
Her work has mostly lived at the boundary between sensitive information, messy workflows, and the people who need systems to be clearer than the organizations around them. That boundary is where AI becomes useful and risky, and it is where she and a small founding team are building The Wheel.
The future should not require us to choose between useful AI and control over the records that make our work meaningful.
AI visits. It never owns.
The beta is invite-only and free.