In 30 seconds
- Greenfield Thinking separates constraints from history.
- The Greenfield Model is a branch held beside the real enterprise — its distance from today is the point.
- Tested in the open: a social-insurance institution rebuilt, on paper, from law and citizen need.
The old house and the empty lot
Imagine an architect who has spent years improving a 150-year-old house. Every design must respect its walls, plumbing and foundations. Then the architect is given an empty lot next door. Same family, same climate, same budget, same needs — only history is gone.
The purpose hasn’t changed and the real constraints haven’t disappeared. History has simply stopped masquerading as a requirement.
Strip, don’t shrink
Run the same exercise on an enterprise. Remove the org chart, job titles, application names, steering committees, even current investments. Keep the intent: customer needs, products, economic logic, legal duties, safety limits, physical flows and commitments already made. Then ask what form the work itself requires.
Conway’s Law explains why this rarely happens: organizations design systems that mirror their own lines of communication, so every improvement flows through channels the old structure provides. AI changes the cost of escaping that circle. It can read laws, contracts, products, processes and data, trace thousands of links, draft a coherent candidate — and revise the whole when one premise changes.
A public institution, rebuilt from intent
To test the method in the open, we built an archetypal, synthetic social-insurance institution from public sources only: legislation, benefit guidance and public service information. We call it ArkkiKELA. (It is not an official project of Finland’s Kela; Kela did not commission, approve or validate it.)
The question was deliberately different from typical digitization: what would such an institution look like if built today from law and citizen need? Every rule named the facts it reads, the section of law it rests on, and the exact point where the law reserves the decision for a human. The data model was designed late, on purpose — every element had to argue its way in.
Discipline required
The Greenfield Model is not a plan to demolish the house. It is a branch held beside the real enterprise, giving historical choices something to explain themselves against.
AI read, drafted and exposed contradictions; people approved. The speed is real — and it does not make lawyers and engineers optional. It makes them the bottleneck that matters.
“History has simply stopped masquerading as a requirement.”