The finance team of 2030: fewer keystrokes, more judgement
Most of the routine work in a small finance function will soon run itself. What is left will be harder to automate and worth more: judgement, control and the conversation about what the numbers mean. What we expect, what is already decided, and how to get ready.
Predictions about technology are usually wrong in the detail and right in the direction. This is our view of where a small or mid-sized finance function is heading over the next four years. Some of it is opinion, and we say so. A surprising amount of it has already been decided.
The work most exposed to AI is office work
In January 2026 the government published its first assessment of how AI could affect the UK labour market. Drawing on IMF estimates, it found that around 70% of UK workers are in occupations with tasks AI could perform or enhance, against around 60% in the US and other advanced economies. The reason is the shape of the economy: Britain earns its living from services, and services run on documents, data and decisions.
- Exposed, where AI is more likely to help
- Exposed, where AI is more likely to take over tasks
- Little exposure
Source: DSIT, Assessment of AI capabilities and the impact on the UK labour market (January 2026), citing IMF estimates.
Exposure measures how much of a job AI could in theory do. Whether AI is more likely to help or to take over depends on how much judgement, responsibility and human contact the job involves. Rows may not add to 100 because of rounding; the estimates are indicative.
Show the data
| United Kingdom: Exposed, where AI is more likely to help | 35% |
|---|---|
| United Kingdom: Exposed, where AI is more likely to take over tasks | 32% |
| United Kingdom: Little exposure | 33% |
| United States: Exposed, where AI is more likely to help | 30% |
| United States: Exposed, where AI is more likely to take over tasks | 30% |
| United States: Little exposure | 41% |
| Advanced economies: Exposed, where AI is more likely to help | 26% |
| Advanced economies: Exposed, where AI is more likely to take over tasks | 32% |
| Advanced economies: Little exposure | 42% |
Finance roles sit squarely in the exposed group. But exposure is not replacement. Around half of exposed workers are in what the IMF calls high-complementarity roles, where the cost of a mistake, the need for judgement and the human relationship make AI more likely to make people more productive than to take their place. Good finance work lives there.
What is already decided
Several of the biggest changes are not forecasts. Making Tax Digital for Income Tax began in April 2026 and reaches sole traders and landlords with income over £20,000 by April 2028. Most benefits in kind move into payroll in 2027 and 2028. VAT invoices become e-invoices from April 2029, the same year Direct Debit becomes the normal way to pay VAT and PAYE. HMRC’s plan is for at least 90% of its interactions to be digital by 2029-30, and its AI assistant, Ask HMRC, already handled around 6.3 million customer interactions in 2025-26.
Decided, not predicted
- April 2026Making Tax Digital for Income Tax begins, for qualifying income over £50,000.
- April 2027Cars, vans, fuel and medical benefits taxed through payroll; the MTD threshold falls to £30,000.
- April 2028Most other benefits payrolled; Making Tax Digital reaches income over £20,000.
- April 2029E-invoicing for VAT invoices; Direct Debit for VAT and PAYE.
Together they mean the data a finance team handles will arrive structured, more often and with less room for error. That removes much of the keying, chasing and checking that fills a month today.
Where the time will go
Where the time goes
- Entering and coding transactions
- Chasing receipts and approvals
- Reconciling at month end
- Building reports by hand
- Answering the same questions twice
- Reviewing the exceptions systems flag
- Testing controls and access
- Explaining results and trends
- Forecasting and planning scenarios
- Advising on decisions as they arise
The speed-up is already measurable. Studies summarised in the government’s assessment found AI tools cut the time taken on writing tasks by 59% and on consulting work by 25%. Not every task is like that, and controlled trials flatter real offices. But a month-end close that takes ten working days today will not take ten in 2030.
- Writing tasks59%
- Software development56%
- IT support44%
- Legal work34%
- Consulting25%
Source: DSIT, Assessment of AI capabilities and the impact on the UK labour market (January 2026).
Results vary by study and setting: signs of what is possible, not averages for every business.
Show the data
| Writing tasks | 59% |
|---|---|
| Software development | 56% |
| IT support | 44% |
| Legal work | 34% |
| Consulting | 25% |
What changes for people
Junior roles will change most. The tasks that used to teach people the basics, posting invoices and reconciling accounts, are the ones being automated. That is a real problem for training, and early signs are visible: the same assessment notes that between 2022 and 2025, UK job adverts fell by 38% in the occupations most exposed to AI and by 21% in the least exposed, while cautioning that AI has not been shown to be the cause.
Our view is that judgement has to be trained on purpose rather than picked up by repetition: juniors reviewing the exceptions a system flags beside an experienced colleague, explaining a variance before they have ever posted a journal, and learning controls as a skill in their own right.
Four jobs every finance function will need
- A reviewer, who looks at what the systems produced, questions what looks wrong and signs it off.
- A controller, who owns the rules: who can approve what, what the software may do alone, and whether it does.
- A data keeper, who keeps the chart of accounts and customer and supplier records clean, because every automated step depends on them.
- A business partner, who can sit with an owner or a manager and turn the numbers into a decision.
In a small business these may be one or two people, or partly your accountant. What matters is that each job belongs to someone.
How to prepare, starting now
- Get your data clean and in one place: AI is only as good as the records beneath it.
- Write down your controls: who approves what, and what the system may do on its own.
- Choose software with open connections, so new tools can plug in without a rebuild.
- Invest in people who can explain numbers to colleagues who are not in finance.
- Review the plan once a year, because the tools and the rules will keep moving.
This is a perspective. The dates and figures come from the sources listed; the view of how finance work will change is ours, and we will revisit it each year.
Sources
More on
This insight is general information, not advice for your circumstances. Talk to us before acting on it.