Cases/01 · AI investments
A Danish public-sector body had committed to a major IT initiative — an investment at meaningful taxpayer scale, with a delivery plan, a timeline, and political momentum behind it. The work was framed inside the organisation as a technology problem: build the system, deliver to spec, ship.
Challenge
The brief looked clean on paper. But the value calculation underneath it was oversimplified — the spec assumed a narrow set of operating conditions and ignored a range of business-domain factors that would shape the system once it met the real world. The architectural design followed the spec faithfully. The real-world problem the system was being built to solve was being swept under the rug rather than addressed.
The implication: the system would build successfully and fail operationally. Not because the engineering was wrong — because the business problem had been mis-framed before engineering began.
What was done
A leading partner on the engagement walked away. The advice given to the client was direct: the architecture as scoped would fail when it met the real domain it was supposed to serve, the value model didn't survive contact with the operating context, and no amount of careful technical execution would compensate for a problem that had been defined incorrectly upstream.
The client proceeded without us.
Outcome
Four years later the system failed at scale, publicly, with material taxpayer cost.
The failure mode was the one named at the outset: the architecture met the world it had to operate in, and the un-modelled domain factors broke it.
The client returned looking for advice; there wasn't a short-term fix. The engagement that followed was about how to recover honestly from where the project actually was, not where the original plan had assumed it would be.
Practitioner
Lars.
Where this fits
This case is the thesis on the page. We find what's worth building — and build it. Or tell you not to. Sometimes don't build it is the answer; sometimes the client builds it anyway; sometimes they come back. The work is the same regardless: telling the truth about what the project actually is, and being trusted enough to be the call when the truth becomes unavoidable.