AI in Finance
The Future of the Finance Function: What AI Changes and What Stays Human
27 April 2026
Most predictions about the future of finance land in one of two places: breathless optimism about how quickly everything will change, or defensive insistence that the human element cannot be replicated. Both miss the point.
The real picture is more specific. Some parts of the finance function will change substantially over the next five years. Others will not change at all. The work is knowing which is which, and not mixing them up.
I have built and rebuilt finance functions at different stages of maturity. I have watched this technology develop in practice, not in vendor demonstrations. This is what I actually think will happen.
The Five-Year Picture
By 2030, the finance function in a well-run organisation will look materially different from today. Not unrecognisably so. But different in ways that matter.
Transaction processing will be largely automated in organisations that have done the foundation work. AP and AR at the transactional level will run through AI-assisted systems, with humans managing the exception queue rather than the full population. This is already happening where implementations are mature. By 2030 it will be standard in any finance function that has managed its data and process foundations properly.
Reconciliation will be AI-assisted with human judgment applied to genuine exceptions. The reconciliation role as it currently exists, skilled people whose value is their ability to work through high volumes of matching at pace, will look very different. The volume work is automatable. The judgment on real exceptions is not. That ratio shifts dramatically.
Management reporting will move toward near-real-time in organisations with the right data infrastructure. The monthly management accounts exist because of data latency and manual production time. Both constraints are reducible. By 2030 the CFO waiting until the 10th working day for management accounts will be in the minority.
The month-end close, as currently structured, will not survive in its current form. The close is a consequence of how financial data has historically been accumulated and processed. When accumulation is continuous and processing is automated, the specific event of the close loses most of its rationale. This is a structural change. It affects not just how finance teams work but how they are managed and how their time is organised.
The finance team of 2030 will be smaller, more senior, and more commercially focused. This is not primarily a prediction about redundancy, though in organisations that handle this poorly, redundancy will be part of the story. It is a prediction about shape: fewer roles at the transactional processing level, more at the analytical and commercial level. Lower aggregate headcount. Higher average capability.
The direction is clear from the technology trajectory. The agentic AI post covers the specific capabilities that make it legible. The finance director role piece covers what this means for senior finance leadership specifically.
What AI Will Not Change
Three things. They are connected.
Judgment
Whether to invest in a new market is a judgment call. Whether to write off a receivable, not the mechanical application of a policy but the real decision about a specific counterparty in a specific situation, is a judgment call. Whether the management accounts are telling the true story or a flattering one is a judgment call. These require context that AI does not have: knowledge of the specific people involved, the history of relationships, the organisational dynamics behind the numbers, the strategic direction that makes some outcomes more concerning than others.
AI can deliver better information faster to support these decisions. It cannot make them.
Relationships
The CFO who builds board confidence over time, who manages the audit committee through a difficult year, who knows which non-executive to speak to before a challenging agenda item arrives: that is human work. It is built on trust, on reading a room, on knowing what is unsaid and deciding whether to say it. These skills do not benefit from automation. They become more important as the transactional layer of the role is removed and the relational and strategic layer fills more of the CFO’s time.
Accountability
When the numbers are wrong, the CFO is accountable. This does not change because an AI produced the first draft. The accountability structure in finance exists because decisions are made on the basis of financial information and the organisation needs someone to stand behind it. AI is a tool the CFO uses. It does not share the CFO’s accountability. The governance architecture that makes AI-assisted finance work, see the AI governance framework for the detail, is built around maintaining human accountability for AI-assisted outputs, not transferring accountability to the tool.
Judgment, relationships, accountability: these have always been the core of the senior finance role. The reason the finance function has sometimes failed to deliver on its “strategic partner” billing is not that those elements were missing from the role description. It is that the transactional layer occupied too much of the available time. When automation removes that layer, the judgment, relational, and accountability work becomes proportionally more of what the role is.
That is better for everyone.
The Skills That Become More Valuable
Commercial judgment
Commercial judgment is the ability to understand what the numbers mean in the context of the business, the market, and the specific decisions on the table. It is looking at a set of management accounts and knowing whether the picture is healthy or whether the headline number is hiding something. It does not compound quickly. Finance professionals who develop it early are in a better position than those who wait.
Strategic communication
Strategic communication is the ability to translate financial complexity into clear, useful language that boards and executive teams can use. The management accountant who understands the numbers but cannot explain why they matter will find that gap more visible once production of the numbers is automated. The CFO who can sit in front of a board and say “here is what this means and here is what we should do about it” is more valuable as AI handles the upstream work.
Systems thinking
Systems thinking means understanding how the finance function connects to the rest of the organisation: what decisions it influences, what information flows through it and from where. This becomes more important when the function’s role shifts from production to insight. Understanding the system is a precondition for extracting anything useful from it.
Working effectively alongside AI tools
Working effectively alongside AI tools is a new but specific skill. It means knowing what to trust, knowing when to interrogate the output, understanding the limitations of the model being used, and improving inputs and oversight when the outputs are not good enough. This is not a technical skill in the programming sense. It is professional judgment applied in a new context. Finance professionals who develop it now will have an advantage that compounds.
The Skills That Become Less Valuable
I will say this directly. The indirect version is not useful.
Deep expertise in specific systems and processes that AI will automate
This becomes less scarce when the automation is in place. The person whose primary professional value is knowing how the reconciliation works in a specific legacy ERP is in a different position once that reconciliation is handled by an AI layer above the ERP. The expertise does not become worthless. There will still be a need for people who understand the underlying systems, particularly during the transition. But it is less differentiating.
Manual reconciliation proficiency at volume
This is the clearest example. Working accurately and quickly through high volumes of matching has been valuable. It will be less valuable as automation handles the volume. The residual skill of judging genuine exceptions is valuable. It is also a smaller population of work.
Technical knowledge whose primary value is managing complexity that should not exist
Finance functions in many organisations have accumulated complex workarounds, manual adjustments, and layer-on-layer process additions that exist because the underlying systems were inadequate or the processes were never properly designed. Some finance professionals have built considerable expertise navigating that complexity. When the complexity is addressed, either through AI adoption or the process redesign that should accompany it, the expertise in navigating it is no longer needed.
This is uncomfortable. It is also accurate. The AI will not fix a broken finance function post addresses the mistake of deploying AI on top of complexity that should be eliminated rather than automated. The mirror of that point: expertise in managing complexity that should be eliminated is not a permanent asset.
What Finance Leaders Should Be Doing Now
Not waiting. That is the starting point.
The organisations that will have the best finance functions in 2030 are building the foundations now. Not because they are deploying agentic AI today, most are not ready for that, and deploying before the foundations are in place produces the worst outcomes. Because the foundations take time to build and the benefit accrues across every stage of the journey, not just the final one.
Process documentation
Process documentation is the work of understanding and writing down how the finance function actually works today. Not how the process maps say it works. How it actually works, including the manual adjustments and workarounds that exist in practice but not on paper. This is unglamorous work. It is also a precondition for process redesign, and process redesign is a precondition for effective AI deployment. Organisations that skip this step find out later that the AI is automating a process that was not the right process to automate.
Data quality
Data quality is the foundation everything else builds on. The data quality post has the detailed framework. The short version: the accuracy, completeness, and consistency of the financial data the AI uses determines the quality of what the AI can produce. This is not a technology problem. It is a governance problem. It requires policies, accountabilities, and enforcement. That means senior finance leadership treating it as a priority and maintaining that treatment over time.
Governance
Governance covers how AI tools will be used, what they will be accountable for, how outputs will be reviewed, and what escalation looks like when the AI is wrong. This does not need to be complicated. It needs to be explicit and written down before the first tool is deployed, not worked out retrospectively when the first question about audit defensibility arises. The CFO guide to AI strategy covers the governance architecture in practical terms.
Team capability
Team capability is the least discussed and the most important foundation. The finance teams that will get the most from AI over the next five years are not the ones that invest the most in tooling. They are the ones that invest in developing judgment, communication, and commercial skills that AI cannot replace, while building the working-alongside-AI skills the new environment requires. That development does not happen without deliberate investment. It does not happen by announcing a new tool and expecting the team to work it out.
Where This Is Heading
The finance function of the future is better than the one most organisations have today. More accurate. More timely. More commercially useful. Less occupied with manual production work that should not require human effort.
Getting there requires the same discipline that getting anywhere in finance requires. Start with what is broken. Understand the current state before deciding where to go. Fix the foundation. Build on it deliberately. Do not skip chapters.
The organisations that apply that discipline to AI adoption will have finance functions in 2030 that are materially more effective than what they have today. The ones that chase the tools without the foundations will have expensive implementations that do not work well enough to justify the investment, and teams that do not know what the future of their roles looks like.
The AI in Finance Strategy page is the right starting point for thinking about where your finance function sits and what the path to readiness looks like. The direction is clear. The work is specific. The organisations that start it are in a better position than the ones that watch.
Maebh Collins is a Fellow Chartered Accountant (FCA, ICAEW) with Big 4 training and twenty years of operational experience as a founder and senior finance leader. She writes about AI in finance transformation from the inside out.