When AI reaches finance, rewrite the jobs first

AI is changing what finance people do sooner than how many there are. Rewrite each finance role around the new work before you decide anything about headcount.

By John Wyatt · November 3, 2026

When AI reaches the finance team, the first thing it changes is the job description. Leaders who start by asking how many seats it will save are starting in the wrong place, and their finance team will read the question that way too.

The early numbers point the same way. More finance leaders say AI has increased their finance headcount (32%) than say it has reduced it (20%), according to Pigment's Q3 2026 CFO Index (Pigment sells planning software), as reported by FM magazine. Companies that plan for the new work will get more out of the tools than the ones planning a smaller team.

AI hands finance new work faster than it takes work away

Picture what changes when a tool starts drafting the variance commentary or matching transactions for the close. The draft arrives faster, and a person in finance still verifies the numbers before they go to the board, the bank or a customer. My rule is that AI output in finance can be trusted as directional and no further until finance has checked it. Who checks what, and who owns the data the tool reads from, are new responsibilities, and they are missing from most finance job descriptions.

Companies are already moving on this. In the same survey, 36% of finance leaders say they have changed the scope of existing roles, and 28% have created new finance roles. As FM summed up the report, AI is "reshaping roles" more often than it is replacing people.

What has not moved yet is the reporting load. Pigment puts the typical finance team at roughly 32 hours a month on reports for leadership, and the teams furthest along with AI are more likely than early-stage teams to spend over 40. That is a warning for anyone rewriting a role. If the new job lists only what the person adds, the week just gets longer. Each rewritten role has to say what the person stops doing as well.

Point the freed time at work finance never got to

The best use of the time AI frees up is the work that was never happening at all. In finance that might be margin by customer every month instead of once a year, a cash forecast refreshed every week, or a second look at the contracts that renew next quarter. People decide which of those matter, and that call is where their time should go.

Rewrite each finance role before you change the headcount

Take the finance org chart and go role by role. For each person, write down what the tool now drafts for them, what they now verify before it goes out, what they stop doing, and which piece of undone work they take on. If the last column is empty for someone, find them that work first. The person who has spent years on a reconciliation usually knows best where the numbers go wrong.

Then give one person the checking rules for the whole function; I have written about what finance should verify before AI output leaves the building. And train people on their own work, side by side with the old way, because that is where they learn when the tool is wrong. Pigment's numbers show the gap: against the 36% who have rescoped roles, fewer than a quarter have stepped up training. A role rewritten on paper, with no training behind it, changes nothing on Monday morning.

A two-person finance team has no org chart to redesign

If your finance function is a controller and a bookkeeper, there is no structure to rewrite. The question there is simpler: which one piece of analysis would you add if the close took two fewer days? Start with that and skip the exercise.

Ask each finance lead what they check now

This week, ask each person who runs part of your finance function one question: what do you check now that you did not check a year ago? If the answers are vague, the new work is landing on people without anyone deciding who owns it, and that is the place to start.

Shaping the finance function around new work is a big part of what a fractional CFO does. If you are working out what your finance team should look like a year from now, I would like to hear how you are thinking about it.

Source note: Pigment's Q3 2026 CFO Index surveyed 2,000 finance leaders in the U.S., the UK, France and Germany in August and September 2026; the full report is on Pigment's site.


John Wyatt · John Wyatt's page · All posts · RSS

The seat this answers

Fractional CFO and COO

Read the pageStart a conversation
Read next

Is your finance team checking AI more than using it?

Trust AI finance output as directional and verify the numbers before anything goes out. If a report needs a full check every run, fix the data underneath.

The AI skills gap is a judgment gap

Most AI training teaches what the tool can do. The gap mid-market teams feel is knowing when its answer is wrong. Train on real work.