About this site

I write about how large regulated organizations adopt AI, and why so much of it stalls.

My working view is that the hard part isn't the technology. AI touches product, operations, risk, data, security and procurement at the same time, which means the work that determines whether it succeeds falls between teams where nobody is accountable and nobody is measured. Exit readiness is my recurring example. European financial institutions are legally required to prove they can leave a critical technology dependency, and almost none of them can do that for a model provider. Not because it's technically impossible, but because it belongs to no one.

I'm a solutions architect working with financial services organizations in the Benelux, mostly insurers and banks, after nine years in cloud. Dutch and Egyptian, based in The Hague.

Disclosure: I work for a cloud provider. That gives me an obvious commercial interest in how organizations think about provider dependency, and you should read anything I write about lock-in with that in mind. Everything here is written on my own time, from public sources, and represents my views rather than my employer's.

If you're working on the same problems — particularly the evaluation question, which I think is the least solved piece — I'd like to hear from you.