AI Will Not Replace Your Finance Team. But It Will Expose the Ones Who Are Not Ready.

The Prediction That Keeps Not Coming True

In 2016, a widely cited Oxford study suggested that accountants and auditors faced a 95% probability of automation within twenty years. Variations of this prediction have circulated ever since, usually accompanied by a chart and a sense of urgency.

It is 2026. Accounting teams are still employed. Finance functions are still hiring. CFOs are still trying to close the books by Working Day 7. The automation wave arrived — but it arrived differently than predicted.

The tasks that were automated were real. Reconciliation matching, invoice processing, expense categorisation, VAT return preparation, bank feed ingestion — all of these are now handled by software that did not exist in a usable form ten years ago. Tools like Candis, Payhawk, and modern ERP platforms with embedded AI have removed thousands of hours of manual data entry from finance teams across Europe.

But the finance professionals who were supposed to be displaced are still there. In many cases, they are busier than before — because the automation surfaced more decisions that require human judgement, not fewer. ———————————————————————-- AI has automated the data. It has not automated the decisions. And in finance, the decisions are the job. ———————————————————————--

What AI Actually Does Well in Finance

It is worth being precise about what modern AI tools are genuinely capable of in a finance context. The capabilities are significant and should not be undersold.

Transaction processing and categorisation

Large language models trained on financial data can categorise transactions with accuracy rates that match or exceed experienced bookkeepers on routine items. Expense claims, supplier invoices, and payroll entries that once required a human to assign GL codes are now handled automatically in well-configured systems.

Anomaly detection and variance flagging

AI-powered tools can scan thousands of transactions and surface anomalies — duplicate payments, unusual vendor patterns, outlier expenses — in seconds. The same work would take a senior accountant several hours and is more prone to fatigue-related error.

Forecasting and scenario modelling

Machine learning models trained on historical financial data can generate rolling forecasts, sensitivity ranges, and scenario outputs faster than any FP&A analyst working in Excel. Tools now exist that update a 13-week cash flow model in real time as bank feeds and AR data change.

Regulatory and compliance drafting

AI assistants can produce first drafts of VAT return workings, statutory accounts disclosures, and tax provision calculations. The first draft is faster and often structurally correct. It is not auditable without human review, but it saves hours.

Report generation and narrative commentary

Board packs, investor reports, and management account commentaries that once took a finance manager a day to produce can now be drafted by an AI system trained on the company's financial data and prior reporting style. The output requires editing. But it is a starting point rather than a blank page. ———————————————————————-- The common thread: AI compresses the time it takes to move from data to output. What it cannot do is determine whether the output is right, whether the business context changes the interpretation, or whether the number is telling the leadership team something they do not want to hear. ———————————————————————--

What AI Cannot Do — and Will Not, Any Time Soon

The capabilities listed above are real. So are the limits. Here is where AI consistently fails in a finance context, and why those failures matter.

Contextual judgement on unusual transactions

An AI system categorising transactions does so based on patterns. When a transaction is unusual — a one-off payment to a new counterparty, an intercompany loan with non-standard terms, a provision for a contingent liability — the pattern-matching fails. A senior accountant applies judgement based on understanding the business, the relationship, and the accounting standards. That judgement is not replicable from a data pattern.

Audit relationships and professional accountability

Auditors work with people. They ask questions, probe responses, assess credibility, and exercise professional scepticism. An AI can prepare an audit support pack. It cannot sit in the room and defend the numbers. The professional accountability that a qualified finance team member carries — the ACCA or CIMA designation, the personal liability, the professional judgement — is not something a language model can substitute for.

Stakeholder communication

Finance is not just numbers. It is the translation of numbers into decisions. A CFO explaining a covenant breach to a board, a finance director walking an investor through a reforecast, an accountant advising a founder on the tax implications of a restructure — these conversations require relationship, trust, and the ability to read a room. They require a person.

Navigating ambiguity in accounting standards

Revenue recognition under IFRS 15, lease accounting under IFRS 16, fair value measurement under IFRS 13 — these standards require interpretation. The accounting treatment for a complex transaction is often not obvious, and the right answer depends on facts and circumstances that require professional judgement to assess. AI systems produce plausible outputs. Plausible is not the same as correct.

The first time something happens

AI systems are trained on historical data. When something has never happened before — a new transaction structure, a change in the business model, an acquisition of a type the company has never made — there is no historical pattern to generalise from. The finance team members who handle the unprecedented situation are the ones who understand principles, not just patterns.

Why Senior Finance Professionals Are More Valuable, Not Less

There is a version of the AI-in-finance story that ends with junior accountants displaced and senior professionals unchanged. That version is also wrong.

What is actually happening is a compression of the pyramid. The work that once justified a team of ten accountants — data entry, reconciliation, basic reporting — can now be handled by a smaller team with better tools. The work that required two senior professionals — interpretation, judgement, stakeholder management — now requires those same two professionals, who are freed from the routine work and expected to add more value in the time that was previously consumed by it.

The net effect is not that finance teams get smaller. It is that the expectations placed on every finance professional increase. The team member who was previously valued for their ability to produce a reconciliation quickly is now expected to interpret it. The FP&A analyst who was valued for building the model is now expected to explain what it means and what the business should do about it. ———————————————————————-- The finance professionals who thrive in an AI-enabled environment are not the ones who resist the tools. They are the ones who use the tools to do more of the work that only they can do. ———————————————————————--

The Risk Is Not Replacement. It Is Irrelevance.

The genuine risk for finance professionals is not that AI will take their job. It is that the finance professionals who do not adapt will find their role narrowed to the tasks that AI has not yet been configured to handle — which is a diminishing set.

The finance professional who adds value in the current environment is one who:

• Understands the AI tools well enough to configure, validate, and override them

• Applies professional judgement to the outputs the tools produce

• Translates financial data into commercial decisions and stakeholder communication

• Manages the audit and compliance relationships that require personal accountability

• Handles the situations the tools were not trained for

This is not a lower bar. It is a higher one. The finance professional of 2026 who is doing their job well is doing work that is more intellectually demanding than their equivalent in 2016 — because the routine work has been automated away and what remains is harder.

What This Means for Hiring and Team Structure

Finance leaders who have absorbed the above are restructuring their teams accordingly. The patterns we see across Serana's client base in DACH and Benelux are consistent:

Fewer junior roles, higher expectations

Businesses that once had three or four junior accountants handling transaction processing are now running leaner, with the tooling handling the volume and one or two people managing exceptions and quality. The expectation on those people is higher.

The FP&A function is growing

As routine accounting is automated, the bandwidth for commercial analysis increases. Finance teams are investing in the FP&A capability that was previously crowded out by month-end close work. The demand for finance professionals who can build models, run scenarios, and advise on commercial decisions is higher than at any point in the last decade.

Embedded finance as a bridge

Businesses that cannot justify a full-time senior finance hire but need the judgement and stakeholder skills described above are increasingly using embedded finance models. A named, senior accountant embedded inside the business — working in the company's systems, attending its meetings, accountable to its leadership — provides the human layer that the AI tools cannot. Serana's embedded finance model exists precisely for this reason.

The value of the qualified professional increases

ACCA and CIMA qualifications are not becoming less relevant. They are becoming more relevant, because they certify exactly the capabilities — professional judgement, technical standards knowledge, ethical accountability — that AI cannot replicate. The qualified accountant who also understands the AI toolset is not being replaced. They are becoming the most valuable person in the room. ———————————————————————-- The businesses that will have a finance function problem in three years are not the ones deploying AI tools. They are the ones deploying AI tools without ensuring that the senior human layer is in place to govern, interpret, and act on what the tools produce. ———————————————————————--

The Answer Is Not Either/Or

AI or people. Automation or human judgement. Efficiency or expertise. These are false choices, and finance leaders who frame the question this way will make the wrong structural decisions.

The finance function that performs best in the current environment is one that has deployed the available automation intelligently, and ensured that the human capacity freed up by that automation is directed toward the work that requires professional judgement, relationship, and accountability.

That is not a technology story. It is a talent and operating model story. The technology is available. The question is whether your finance function is structured to get the most from it — and whether the people in it are equipped to do the work that remains.

+———————————————————————--+ | Serana deploys embedded finance teams and senior finance | | professionals into European scaleups. If your finance function is | | navigating the transition to an AI-enabled operating model and you | | want a senior, qualified professional inside your business — not | | just reviewing outputs from the outside — speak to us. | | | | Book a discovery call — seranapartners.com │ | | info@seranapartners.com | +———————————————————————--+

Serana Partners B.V. | Keizersgracht 391A, Amsterdam | www.seranapartners.com