AI in Excel is the most exciting way to solve the wrong problem.
by Raphael Steinman
For finance: AI in Excel is the most exciting way to solve the wrong problem.
The formulas write themselves. The vlookups vanish. The pivots assemble in seconds. Finance teams are genuinely faster.
But faster at what?
– Faster at exporting from three ERPs that define “revenue recognition” differently.
– Faster at rebuilding intercompany eliminations that require context no system records have.
– Faster at reconciling a cost center hierarchy restructured twice, with the logic living in someone’s head; someone who may not be there next quarter.
AI in Excel is a faster way to hand-refine raw ore.
It is still hand-refining. Still person-dependent. Still disposable. Next month, you start over.
That is the part nobody wants to examine. A finance team running on spreadsheets is not describing a tool preference. It is describing an architecture failure.
The enterprise systems they were given do not produce answers they trust.
So they export, rebuild, and cross-reference until the numbers make sense. Every month. From scratch. AI just makes the “from scratch” part quicker.
This is the distinction the market is missing: multiplying capabilities is not the same as collapsing complexity.
Multiplying capabilities means each person does more. The work still scales with headcount. The reconciliation logic still lives in one person’s mind. Still evaporates when they leave.
Collapsing complexity means the work resolves once and stays resolved.
Revenue means one thing. Entities are harmonized. Process context is structural, not tribal. The tenth report draws from the same foundation as the first; and costs almost nothing.
One approach gives you a faster finance team.
The other gives you a finance function that compounds.
The companies treating AI as a spreadsheet accelerator will outperform last year’s version of themselves.
The companies collapsing the underlying complexity will outperform everyone else.
That gap widens every quarter. It does not close on its own.