Short-staffed on a Tuesday morning: what AI changes in the working day of an accounting firm
How AI helps accounting firms with the staff shortage: what a Tuesday looks like, which tasks speed up and what changes for the people doing the work.
Four in ten Dutch accounting firms have open vacancies. Asked about the consequences of the staff shortage, 36 percent say they no longer take on new clients, according to the Exact SME Barometer 2026 for accountancy, a survey of 195 firms. At the same time, nearly three quarters of those firms already use AI in some form. AI does not solve the accounting staff shortage by replacing people. It helps by taking over the bounded, repetitive steps in the process, such as coding transactions, matching payments and chasing missing documents, so that the people you do have spend more of their day on review, judgement and client contact. That only works when AI has a fixed place in the process, with a person who approves every output.
How big is the staff shortage in Dutch accountancy?
Big, and it will not disappear soon. The Dutch Employee Insurance Agency (UWV) classified the labour market for both accountants and bookkeepers as “very tight” (zeer krap) at the end of 2024. It expects only a slow easing towards 2030, according to its outlook on labour market tightness per occupation. That covers around 122,000 working accountants and 146,700 bookkeepers.
The Exact survey shows what that means on the work floor. Asked about the consequences of the shortage, 57 percent of firms name pressure and stress at work, and 36 percent say they no longer take on new clients. In the same survey, 52 percent name waiting for missing client information as a cause of peak workload. That last figure says a lot: part of the pressure is not in the accounting itself, but in the work around it.
What problems does AI actually solve at an accounting firm?
Three kinds of problems, and they are worth keeping apart because each needs a different approach.
Volume work. Coding bank transactions, matching payments to invoices and preparing standard reports come back every week for every client. This is where AI saves the most time, because the steps are similar and the input is digital.
Scattered information. Client documents arrive by email, portal and app, and someone has to keep track of what is still missing. AI can compare what came in with what was expected and prepare a reminder. It does not make clients faster, but it takes the bookkeeping of the chase off someone’s desk.
First drafts. A first version of a management letter, a summary of changes compared with last year or an answer to a routine client question. AI writes the draft; the professional decides what is correct and what goes out.
What AI does not solve is the judgement itself. The professional guide of the Dutch accountancy body and the IT auditors’ association is clear on this. In “AI Applied” (Leidraad 2: AI Toegepast) by NBA and NOREA, June 2026, the professional keeps full and undivided responsibility, and AI may never be a black box for an accountant.
What does a Tuesday morning look like when AI is part of the process?
Here is an illustration, not a case: a composite morning at a firm of around twenty people, built on the processes the NBA and NOREA guide describes. An opinion piece on Accountant.nl by the founder of an accountancy software company recently sketched a similar morning. We describe our own version, because we think the interesting part is not the software but what the people do differently.
08:30: the bookkeeper starts with exceptions, not with a pile
Overnight, bank transactions for her clients were coded automatically. The system proposes a general-ledger account for each line, as the guide describes for smart coding: proposals that the user authorises, with full automation only once there is enough confidence. She does not go through four hundred lines. She looks at the thirty the system was unsure about, plus a sample of the rest. Her expertise goes where it is needed.
09:15: matching runs, and she checks the remainder
Payments are matched to open invoices. Most matches follow fixed rules in the accounting software; AI helps with the remainder, such as a payment with a different reference or a partial payment. The guide is honest about the limits: for very large volumes, AI is relatively expensive and inaccurate compared with simple rules. So the rules go first, and AI clears what is left.
10:00: the assistant accountant prepares the monthly figures
For a client with monthly reporting, a first draft of the commentary is ready: which costs rose, which revenue fell, compared with last month and last year. He checks the numbers against the source figures, removes one sentence that is technically correct but irrelevant, and adds the context he knows from the last client meeting. The draft saved him writing time; the judgement is his.
11:00: the missing documents are tracked, not chased by memory
A list shows which clients still owe which documents for the year-end, compared with what they supplied last year. Reminders are ready to send, but a colleague reads them first and decides who gets a phone call instead of an email. The client who always forgets the mortgage statement gets a friendly call.
11:30: time for the conversation that was always postponed
This is the point of the whole morning. The time that came free goes to a client who has questions about expanding, or to a junior colleague who needs half an hour of explanation. Researchers Choi and Xie of Stanford and MIT surveyed 277 accountants and analysed data from 79 small and mid-sized firms. Firms using AI closed their monthly books 7.5 days faster and spent 8.5 percent less time on routine back-office work. Their time shifted towards client and advisory work. It is early, observational evidence from the United States, but it points in the same direction.
What does AI change for the people on the work floor?
The work shifts from doing to checking and deciding. That sounds like a promotion, and for many people it is, but it asks for different skills.
Reviewing becomes a core skill. In the same study, 62 percent of accountants said they were concerned about errors from AI. That concern is healthy. A firm where people check AI output well is safer than a firm where they trust it blindly. Experienced accountants benefited more from AI than juniors, partly because they recognise a wrong suggestion faster.
Junior work changes. In a survey by the British accountancy institute ICAEW among managing partners of mid-tier firms (May 2026), 68 percent expected AI to reduce demand for some early-career roles. At the same time, 83 percent said this would not directly mean fewer roles overall. The tasks juniors used to learn on, such as coding and matching, are exactly the tasks AI takes over. Firms will need to think about how new colleagues still learn the craft, for example by reviewing AI output together with an experienced colleague.
Ownership becomes explicit. Every AI step in the process needs someone who is responsible for it: who checks the results, who adjusts it when the software changes, who decides when a suggestion may be accepted automatically. Without an owner, an AI step slowly becomes the black box the NBA warns against.
How do you start without making the pressure worse?
By choosing one process that hurts now, and redesigning it before adding tools. Write down how it runs today, where people wait and where they redo work. Then decide which step AI takes over, who approves the output, and how you will measure whether it helps, for example hours per client per month including review time. Measure net: a task that takes half the time but needs a full check afterwards has not gained much.
Small firms rarely have someone who can do this alongside client work, especially during the busy season. That is where outside expertise helps: someone who knows the software you use and can set up one process properly with your team. See the AI services and ready-made solutions from our experts, or browse the AI questions other companies have posted for inspiration.
Frequently asked questions
Can AI solve the staff shortage at an accounting firm?
Not on its own. AI can take over repetitive steps such as coding transactions, matching payments and drafting reports, which frees time for the people you have. Whether that eases the shortage depends on how well the process is redesigned and on what the freed time is used for.
Which accounting tasks does AI make more efficient?
The NBA and NOREA guide “AI Applied” (2026) describes coding transactions, matching payments to invoices, monthly reporting, preparing year-end work and drafting the notes to the accounts. In each case AI proposes and a professional approves. For very large volumes, simple rules are often cheaper and more accurate than AI.
Will AI replace junior accountants?
The evidence so far points to changing roles rather than disappearing ones. In an ICAEW survey of UK mid-tier firms (2026), most leaders expected less demand for some early-career tasks but not fewer roles overall. Firms do need a new way for juniors to learn the craft, because the tasks they used to learn on are being automated.
Is it safe to use AI with client data?
It can be, if you use business versions of tools with agreements on data storage and training, decide which data may be used, and keep a person responsible for every output. The NBA guide stresses that AI may never be a black box for an accountant: you must be able to trace every finding back to the source figures. This is not legal advice.
Your next step
Pick the process that costs your team the most time in the busy season and write down how it runs today. For more background, see how firms post an AI question on Baionic. When you are ready, post that process as a question on Baionic: within one working day you receive proposals from one to three matching experts, with their approach, price and planning side by side, and you pay only after you approve the delivery.
Ready to take the next step?
Post your project for free and receive proposals within one business day.