You're Designing an AI-Enabled Finance Function Around People You Won't Be Able to Hire
You're Designing an AI-Enabled Finance Function Around People You Won't Be Able to Hire
Every AI roadmap ends the same way: automation absorbs the transactional work, humans move up into judgement. There is a supply assumption buried in that sentence, and it isn't holding.
Sit through enough AI-in-finance discussions and you notice they converge on an identical destination. The processing work — coding, matching, chasing, reconciling — gets absorbed by automation. The people who used to do it move up into analysis, business partnering, and the judgement calls that software can't make. Leaner function, higher-value work, better jobs.
I think that end state is broadly right, and I've argued for versions of it on this blog. But it contains an assumption that almost never gets stated: that the people are available to move up into, and that when you need more of them, you can hire them.
Look at the pipeline and that assumption is the weakest part of the whole model.
The supply assumption, in numbers
Research by Victoria University for Jobs and Skills Australia, cited by CPA Australia, projects the number of accounting roles in Australia rising from 201,600 in May 2024 to 234,000 by May 2034 — an extra 32,400 accountants over the decade from growth alone. That is before replacing anyone who leaves, and the leaving is not trivial: work by Oxford Economics Australia for the Future Skills Organisation estimates a 21.1 per cent attrition rate in accounting between 2024 and 2030 through retirements, career changes and moving overseas. CPA Australia's own read is that Australia needs an average net increase of roughly 4,000 accountants and auditors every year over the next decade.
Against that, the supply side. CPA Australia's preliminary analysis of 2024 Department of Education data counted 7,020 accounting graduates — 1,528 domestic and 5,492 international. But CPA Australia estimates that around 70 per cent of international students leave Australia after completing their qualification, and on that basis puts the number actually available to enter the Australian accounting workforce in 2025 at only 3,000 to 3,500. Both figures are CPA Australia's estimates, made in a submission arguing for skilled migration, and should be read as such — but the direction is not seriously contested.
|
+32,400
Additional accounting roles projected over the decade to May 2034 — growth alone, on top of an estimated 21.1% attrition from the profession between 2024 and 2030.
|
7,122 → 340
Enrolments in the Accounting Professional Year Program, 2018 to 2024. The decline made the program financially unsustainable and it has now ceased.
|
These are structural figures, not a news event. There is no announcement here and nothing changed this week. That's rather the point: this constraint has been assembling quietly for years while the conversation in finance has been about tooling.
The pathway that closed this year
One concrete marker is worth pulling out, because it moved from "declining" to "gone" recently enough that many finance leaders haven't registered it.
In February 2025, CPA Australia, Chartered Accountants Australia and New Zealand, and the Institute of Public Accountants jointly announced they would cease the Accounting Professional Year Program. Enrolments had fallen from 7,122 in 2018 to 340 in 2024, which made it financially unsustainable. New enrolments stopped in March 2025, and the program concluded its activity by 1 May 2026.
The Accounting PY Program had run since 2008 and more than 48,500 participants had graduated from it. It was a recognised pathway to permanent residency for graduates on 485 visas — which is to say, it was one of the mechanisms by which internationally-educated accountants converted a degree in Australia into a career in Australia. That mechanism has now stopped.
Reasonable people disagree about migration settings, and this post isn't an argument about them. The relevant fact for a finance leader is narrower: a channel that supplied entrants to the profession has closed, and no replacement pathway of comparable scale has been announced.
Why this binds hardest in funded sectors
National shortages are experienced unevenly, and the unevenness is not random.
An NFP, an NDIS provider or an aged care operator competes for the same finance talent as everyone else, but with salary bands that are effectively set by a funding model rather than by the labour market. When a commercial employer responds to scarcity by paying more, that is a decision. When a funded organisation responds to scarcity, it is a submission — to a funder, on a cycle, with an outcome that arrives long after the vacancy did.
So in a tight market these organisations don't lose the competition occasionally. They lose it structurally, and the loss compounds: the roles stay open longer, the remaining team absorbs the work, the work that gets dropped is the analytical work rather than the compliance work, and the function becomes less attractive to the next candidate precisely because it has become more transactional.
That is the loop worth naming at board level. Understaffing doesn't just reduce capacity. It degrades the job, which reduces future capacity.
What actually follows from this
The wrong conclusion is "so we need AI even more urgently." That's true but useless — everyone already believes it, and it doesn't tell you what to automate first.
The useful conclusion is that it changes the sequencing.
The default automation logic is to go where the biggest measurable saving sits: the highest-volume, highest-cost process, usually with a headcount number attached to the business case. That logic is sound where the labour is genuinely replaceable — and for routine data entry and transaction processing it often is, which is why savings-led automation there remains the right call. It stops being sound at the point where the role you release is one you could not re-hire. Cross that line and you've converted a reversible position into an irreversible one, and you'll discover the cost the first time volumes rise or somebody resigns.
The alternative is to automate first where it retains the people you already have — the work that is tedious, repetitive, low-judgement and quietly corrosive to morale. Manual reconciliations. Rekeying between systems. Chasing the same missing information every month. Nobody's business case looks impressive on that basis, because the benefit is a retained senior finance officer rather than a line in a savings schedule. But a retained senior finance officer is worth more than the saving, because the replacement doesn't exist at any reasonable price.
Put more bluntly: in the model everyone is designing towards, the scarce resource is not the AI. It's the judgement. Design your automation programme to protect the scarce thing.
Three questions for the FY27 plan
1. If your most experienced finance person resigned next month, how long would the role stay open — and what's the honest evidence for that estimate? Not the number you'd like. The number based on the last three recruitment rounds you actually ran. That figure is the single most important input into how aggressively you should automate for savings versus for retention.
2. Which of your current automation candidates would make someone's job better, and which would make a position redundant? Both are legitimate. But you should know which is which before the business case is written, not after, because the second category carries a re-hiring risk that belongs in the paper.
3. Where does the judgement actually live right now? In most small and mid-sized finance functions it lives in one or two people's heads, undocumented. Automation programmes routinely make that worse — the process gets encoded, the reasoning doesn't. If the judgement is the scarce resource, capturing how it's exercised is worth more than automating the steps around it.
The uncomfortable version
There's a reading of all this that finance leaders in funded sectors tend to arrive at privately and rarely say out loud: that the AI conversation is not really about efficiency for them. It's about whether the function remains viable at the staffing level they can realistically sustain.
I don't think that's defeatist. I think it's a more honest framing than the one in most vendor material, and it produces better decisions — because it forces the question of what the function must be able to do with fewer people, rather than what it could do faster with the same people.
The organisations that handle the next few years well won't be the ones with the most sophisticated tooling. They'll be the ones that worked out early which capabilities they cannot afford to lose, and built the automation programme around protecting those, rather than around whatever produced the best-looking business case.
Is your automation plan built around savings or around retention?
In a market where senior finance capability is genuinely hard to replace, those two plans look very different. PFL provides senior-level outsourced finance, management reporting, and AI automation for Australian NFP, NDIS, and SME organisations.
Talk to PFL →Accounting bodies left with no option but to cease Accounting Professional Year Program for international students — CPA Australia, CA ANZ and IPA joint media statement, 21 February 2025
Joint media statement — Institute of Public Accountants
Employment projections — Jobs and Skills Australia
Skilled migration mooted to address accountant shortage — Financial Newswire
Comments
Post a Comment