Flagship case study
Even
“Your studio runs itself.”
- Thesis
- “Your studio runs itself”
- Category
- AI operator
- For
- Movement studios
- Owns
- The booking system
- Built with
- Agent teams
- Domicile
- EU
- Built in
- Bern
What it is
Even is an AI operator for movement studios. Not an assistant that drafts a reply for a human to send: an operator that owns the movement studio’s booking system. It handles the schedule, the bookings, the changes and the follow-ups, and it escalates to the owner when judgement is needed.
Ownership is the point. An assistant saves minutes. An operator removes a job from the owner’s week.
How we build it
We build Even with agent teams. A small number of people set direction and review. Teams of AI agents plan, implement, test and document. Every method we teach in briefings and pilots runs here first, on a product that has to work.
That order matters. We would not teach a practice we have not shipped with.
The agent-team process
- 01
Scope
A goal becomes a set of small, reviewable tasks with explicit acceptance criteria. Humans write the goal; agents draft the breakdown; humans approve it.
- 02
Build in parallel
Implementation agents work on separate branches at the same time, each with the codebase conventions and the task in context.
- 03
Review before a human sees it
A review agent checks every change against the spec and the conventions. Only changes that pass reach a person.
- 04
Evaluate the behaviour
The operator’s decisions are tested against a growing set of evals. A change that regresses an eval does not ship.
- 05
Human gate
A person approves every release and every rule the operator is allowed to apply on its own.
1 week for specification. 2 weeks build time end-to-end. AI team built, human decision-maker.