Twenty-five years spent translating complexity in the finance function — into structure, and into decisions.

Building the operating model finance needs for AI, with governance built into its foundation. That's what earns trust.

Finance runs on deterministic processes. AI is probabilistic. The real question is what operating model lets the two coexist: decision rights that specify where the system can act and where it can only recommend, an audit trail a regulator would accept, and controls that catch problems before they compound. Get governance right and adoption follows. Skip it, and the tool either goes unused or gets used in ways nobody in the room can fully explain.

Build that foundation first, and speed and cost savings follow. Skip it, and speed itself becomes the risk. Here is what that looks like in practice, and what the 2026 CFO surveys say about it:

I work on this hands-on — I build the applications myself, which is how I know what the technology can and cannot do. I'm not a technical AI specialist and don't write as one; I build to understand the finance implications. The insight agent I built for QuipuCFO is one example: a Python pipeline that pulls filings from SEC EDGAR, computes a fixed set of KPIs for a company and its competitors by ticker, and feeds the result to a model that drafts an executive-ready competitive analysis, structured the way finance is meant to land with a leadership team. Nearly all the effort went into the deterministic pipeline; the prompt itself was almost an afterthought. That's the pattern I keep returning to: human judgment sets the logic, AI scales it, human judgment validates the result.

That gap — between what AI can do and the operating model a CFO's organisation needs to govern, audit, and afford it — is the work.

I have spent twenty-five years in the finance functions of regulated, capital-intensive institutions across insurance, asset management, and banking.

Whatever the title, the recurring work was the same: turning complex financial and regulatory material into something a senior team could decide on.

Three things describe how I work. I build structure out of complexity before I build anything else. I reason from evidence rather than consensus. And I am drawn to the frontier of a field rather than its settled middle — which is why AI in finance holds my attention now. I also work between functions that do not naturally agree — finance, actuarial, risk, technology — and bring them to a shared decision. An actuarial assumption can be perfectly sound for a technical calculation and, read as management, still be an ambition the organisation has to hit for the business model to hold. Connecting those two readings — the technical and the steering — is the kind of work I mean.

I hold an EMBA from Rotterdam School of Management and an MSc in Financial Economics from the University of Amsterdam. I live and work in Amsterdam.

I'm talking to CFOs, Controllers, and Finance Directors about how they're actually governing AI in the finance function.

Where the authority sits, what breaks when the data underneath isn't ready, how cost gets forecast when pricing is usage-based — 30-minute conversations, no sales pitch. If you're navigating this, I'd value your perspective.