The machine took the bookkeeping. People kept the judgment.
Accounting has already been through what most professions are about to face. In the late 1970s the spreadsheet arrived, and within a few years the work that had filled accountants' days, posting entries and recalculating columns by hand, was done by software. The profession did not disappear. It moved up.
From recording to judging
Before the spreadsheet, a large part of an accountant's value was accuracy in recording: the right figure in the right column, totals that added up. Once the machine did the recording, that value fell to almost nothing. What remained was everything the machine could not do: deciding how a transaction should be treated, spotting what looked wrong, explaining the numbers to a manager, advising on what to do next.
The people who adapted stopped being the ones who produced the figures and became the ones who validated them. The work went from gatekeeping, nothing passes without me, to validating, I check what the system produced and answer for it.
The same shift, in other jobs
AI is doing to drafting, summarising, scheduling, routing and first-level answering what the spreadsheet did to posting. The pattern is the same. A planner stops building the schedule and starts approving and adjusting the one the system proposes. A customer service agent stops answering routine questions and handles the cases the system escalates. A lawyer stops searching for precedents and judges which ones apply.
The value moves to the person who knows when the output is wrong, why, and what to do about it.
The warning in the precedent
There is a catch. Junior accountants learned judgment by doing the posting: years of entries taught them what normal looks like, so they could recognise the abnormal. Remove all the junior work and you remove the place where judgment was learned.
So the shift has to be designed, not assumed. Decide which jobs in your company are moving from producing to validating. Give those people the authority to reject what the system produces, and train them for it. And keep a deliberate path for juniors to learn what good looks like, even when the machine does the first draft.