Security & Defence

In this domain there is rarely a shortage of data. It sits scattered across registrations, reports, systems belonging to chain partners and files that were each set up for a purpose of their own. So the question is almost never which new source to add, but what becomes visible when existing sources are placed side by side.

That is precisely the analysis we carried out for the Dutch police: which combinations of existing datasets yield new, relevant information, and which mainly yield spurious connections. That second half is the hardest and, here, the most important, because a connection that lights up by chance costs investigative time that nobody has to spare.

Why it weighs more heavily here

A wrong figure in a sales report costs a wrong decision. A wrong connection in an investigative environment affects people. Explainability is therefore not a quality requirement afterwards but a design requirement beforehand: for every outcome it must be possible to reconstruct which sources contributed to it and which processing was applied.

We believe technology should be reliable, transparent and explainable, certainly where decisions affect people. In this sector that is not a principle we add, but a requirement the organisation already sets itself.

The Dutch police and DITSS were clients; about that work we publish only what they permit us to. At the Ministry of Defence and the Royal Netherlands Marechaussee it concerned the deployment of individual specialists. About the nature of that work we make no statements.

We are not a defence supplier and do not present ourselves as one. What we bring is data and AI work: opening up and linking sources, building models that stay explainable, and keeping the facility running afterwards. If you are working on an assignment in this domain and lack those specialisms, that is the conversation we are glad to have.