It is the last week of the month and the numbers still move. Not the closing figures, those are not due yet, but the ones you are steering on this week. Four systems, four answers, and somewhere in there the figure you will have to defend on Thursday. Almost every finance leader I speak to knows that week, and almost nobody talks about it out loud.

Everyone is talking about AI in finance. Almost nobody talks about the question underneath it: are your numbers right enough to let a machine loose on them?

That question is not rhetorical. Gartner expects organisations to abandon sixty percent of the AI projects that run without AI-ready data through 2026, which is to say before this year is out, and that deadline is now four months away. In the same breath, eighty-seven percent of CFOs call AI important or extremely important for the finance function in the year ahead. The ambition is universal. The foundation is not. And where AI in finance falls apart, it is almost never the model. It is the data underneath.

So we went looking for a signal that cannot be spun. Not what companies say about their AI strategy, but something more honest: who they are hiring right now. A finance department getting ready for the AI era is already hiring the people who build it. Whoever is not hiring those people is not building the foundation, however hard the board hammers on AI.

On 11 June and 10 July 2026 we opened the live finance careers page of ten of the largest e-commerce companies in the Netherlands and sorted every open finance role into three groups. Building the future: finance systems, data inside finance, automation, reconciliation at scale. Running the present: classic control, credit, procurement, administration. Standing still: barely any finance hiring at all.

Two of the ten came out in the first group: bol and Booking. Eight did not.

At bol, the controllers now carry data and reporting in their titles, and there is a data analyst open specifically for finance, separate from central BI. At Booking, between the classic controllers and tax roles, sits something you see almost nowhere else: a product manager for AI in finance, and a specialist who builds the finance systems. In both companies, the data layer under finance is treated as a discipline of its own rather than a byproduct of central IT.

The other eight hire to run the present. There is no verdict in that. We looked at public job posts on a single day, and a company can have a strong foundation and simply not be hiring. But direction is exactly what a snapshot shows, and the direction is consistent.

The frontrunner builds quietly. The lead does not appear in a press release. It appears in a job title, months or years before it becomes visible in the numbers. By the time you can see it in someone’s reporting, the hiring that made it possible happened long ago.

The biggest untapped lever sits with the companies that already know how to build. The organisations with the strongest engineering cultures, the ones that write their own logistics software and put robots in their warehouses, are often the ones hiring finance most classically. Capex controller. Fixed asset accountant. Junior associate. All that engineering strength sits in the operation and not yet in the numbers. That is where the opportunity sits, because the muscle already exists. Of everyone in this study, they have the shortest distance to a modern finance function. It only has to be pointed inward.

The third one we did not see coming. Nobody hires the person who owns the match between the payment provider and the ledger. Across more than fifty open finance roles, not one. In most teams that work is nobody’s job and everybody’s evening. Automation was barely better: one of ten. These are the two things most often named when finance leaders talk about what AI should fix, and they are the two things nobody is staffing.

Read the matrix vertically and the pattern is unmistakable. The three rows where the future gets built stay almost empty, while the bottom rows fill up with the machine of today.

Six roles build the finance of tomorrow. A financial systems owner who owns the data pipeline inside finance and turns separate sources into one truth. Someone who owns the match across payment providers and entities and turns the match rate into a hard, daily number. A data engineer inside finance who builds the pipelines that feed the ledger rather than central BI. A controller who codes, speaking SQL and Python and accounting, as a bridge between two worlds. An FP&A profile that forecasts on reconciled input instead of on last month. And an automation lead who takes the manual work out of the close so the team can work on the question behind the number.

Six roles across ten companies is sixty chances to build one. Three were taken, always by bol or Booking. What the other fifty-seven would have bought is not a dashboard. It is the Thursday where the number is already right before anyone asks, and the auditor’s email that no longer decides how your week goes.

That is the finding, and it is also the opportunity. An empty quadrant is not a shortfall. It is an open goal. The department that hires those people now builds a lead that is difficult to catch later, and it does so at a moment when almost nobody is bidding for the same people.

You can run this exercise on your own finance function in ten minutes, without a report and without us. Open your own finance job posts and write down the titles exactly as they stand. Not what you would like to hire. What you are hiring now. Plot them: do they fall in the rows that build the future, or in the rows that run the present? Then look at which row stays empty. That empty row is your next hire.

The sequence is the reverse of what the hype suggests. Truth first, cleverness after. As long as four systems give four numbers, every AI output stays an opinion with a dashboard around it. Not because the model is weak, but because the source is.

You do not buy the finance of tomorrow. You hire it.