AI can run your finance function. It cannot run your judgement.

By Zaid Aboobaker, Founder & CEO, CompassPoint Consulting

Finance has become one of the most AI-exposed functions in any business. Reconciliation, invoice processing, expense categorisation, and first draft reporting can now be automated to a standard that would have seemed implausible three years ago. Research by McKinsey has consistently identified finance as among the functions with the highest technical potential for automation. In the UAE, where the government has set a target to deliver half of public services through AI by 2028, the direction of travel is unambiguous.

The temptation that follows is predictable. If AI can do this much of the work, why not let it do more? Why keep a senior, expensive human in the loop at all?

This comes down to the volume layer and the judgement layer.

It helps to split the finance function into two. The first is the volume layer. This is the high frequency, rule-based work that consumes time. Posting transactions. Matching payments. Running reconciliations. Producing the standard monthly pack. AI is transforming this layer, and the firms that have embraced it are operating at a distinct competitive advantage.

The second is the judgement layer. This is the work of deciding what the numbers mean, what to do about them, and what to say to the people who depend on them. Which client relationship to end because the margin does not work. Whether to hold cash or invest in hiring. How to frame a difficult quarter to a board or an investor. This layer is not about processing information. It is about taking responsibility for a decision. Are you comfortable letting AI make decisions? I am not.

AI is remarkable at the volume layer. It is not ready for the judgement layer, at least not yet, based on what I have seen.

AI does not hold the institutional memory of why a particular client was taken on at a low margin as a strategic bet. It does not feel the weight of a payroll commitment to thirty families. It cannot sit across a table from a founder and have the honest conversation about cash discipline that actually changes behaviour. And when a decision proves wrong, it cannot be held accountable for it. Someone has to own the call. That someone has to be human.

Consider a simple case. An AI tool flags that a client’s margin has fallen below threshold and recommends ending the relationship. The numbers are correct. But the tool does not know that the client is a reference account that wins three others, that the contract renews next quarter at a better rate, or that the founder gave a personal commitment to see the year through. The recommendation is numerically sound and commercially wrong. Only a human with the full context can tell the difference. This is the distinction that separates firms getting real value from AI from those generating a false sense of control. The winners are not the ones automating the most. They are the ones clearest about where the volume layer ends and the judgement layer begins.  The danger is that the line gets drawn by accident. A finance team adopts a few tools, automates whatever is easiest to automate, and ends up with a finance function that has more software but no clear philosophy about where humans add unique value. That is not a strategy.

The better approach is deliberate. Map every recurring finance activity. For each one, decide whether AI executes with a human reviewing exceptions, or whether a human decides with AI providing support. The split will differ by business, but the act of deciding it consciously is what matters. That is a leadership decision, not a software decision. It belongs with the CFO, not the IT budget.

This is the gap CompassPoint was built to bridge. We pair senior CFO expertise with the reporting technology and AI-enabled tools that most fractional finance firms in this market have not adopted.

We help our clients to understand the technology is in service of the judgement layer, not a replacement for it. A senior finance partner frames what the numbers mean, decides the strategy, and has the conversations that matter. AI should handle the volume so that human judgement has more room, not less.  It is also, I believe, where serious finance leadership is heading across the GCC.

 

 

About the author: Zaid Aboobaker is the Founder and CEO of CompassPoint Consulting, bringing more than 20 years of experience across the Middle East, India, and Europe. A CIMA Fellow and CGMA, he specialises in growth, transformation, M&A and finance leadership, helping businesses scale sustainably through sharper strategy, stronger systems and operational discipline globally.