We're told that AI will transform finance. That it will automate the tedious work, free up teams to focus on strategy, and unlock business value at scale. But according to a survey of over 3,500 finance leaders across Europe, the implementation is more complex than the hype suggests.
When we look at how AI in finance is actually being used today, it’s evident that there is plenty of opportunity. Despite companies adopting AI-powered tools, finance teams are still drowning in manual work.
So, what's behind this gap between adoption and transformation? In our recent report, A Transformation Moment for Finance: How UK Organisations Can Turn AI Adoption into Lasting Change, we surveyed 3,500 finance professionals to find out. The findings show that most teams have adopted AI tools without redesigning the processes around them.
Here are five hard truths about AI in finance that every finance leader needs to confront.
Let's start with the headline: 98% of UK finance leaders are already using AI. Adoption of AI-powered tools has undoubtedly become widespread across companies of all industries, yet nearly half (49%) of finance leaders report they're struggling with managing AI more than they're struggling with adopting it. This distinction reveals the gap between having a tool and knowing how to best leverage it.
Most organizations deployed AI reactively — bolting it onto existing processes rather than redesigning how work gets done. This becomes clear when we look at how AI is used in finance today. The most common AI applications in finance range from expense categorization to fraud detection, from payment approvals to cash flow forecasting. But these applications can be siloed and create fragmentation — what practitioners now call "AI sprawl."
For AI to deliver real value, it needs to be part of a deliberate redesign of how finance operates. It needs clear boundaries, defined governance, and measurable outcomes. Without these, adoption of AI in finance becomes a cost center disguised as innovation.
Here's a question every CFO should ask their finance team: What's the ROI on our AI investments?
According to the report, 61% of UK finance leaders have invested in AI in areas where the actual return on investment is unknown. This reflects a deeper challenge: many organizations deploy AI without first understanding where the benefits of AI in finance will be greatest. The result is scattered deployment — AI handling tasks that don't move the needle, tools solving problems that don't exist, while real bottlenecks remain unaddressed.
This scattered approach compounds into what practitioners call "AI sprawl." 47% of finance leaders have over-invested in different AI tools and now find themselves managing overlapping, incompatible systems. Together, they've become a maintenance burden that consumes time and budget.
This matters because it undermines confidence. CFOs can't make a case for expanding AI investment when they can't articulate what the last round of investment delivered. Finance teams become frustrated when they're asked to adopt tool after tool with no clear connection to their actual work. And the cycle perpetuates: more tools, less clarity, diminishing returns.
Despite AI adoption, finance teams are more stretched than ever.
The report identified several areas where teams spend enormous amounts of time on repetitive, low-value work.
Ultimately, this adds up to over 16 hours per week — two full work days — spent on checking, chasing, and correcting information. That’s time that could otherwise be spent on the business-critical work that finance leaders were hired to do in the first place. In fact, over half (54%) of finance leaders agree they spend more time policing than shaping strategy.
This is where the real opportunity for how to use AI in finance lies — in systematically removing the friction from routine, repeatable processes and redirecting that reclaimed time toward where finance actually adds value.
Here's where most AI implementations fail: they ask the wrong question at the start.
They ask: "How much of this can AI automate?"
They should ask: "Where do humans need to stay in control? Where can AI accelerate? Where should AI own the work end-to-end?"
The answer varies dramatically by task. For spend policies, half of UK finance leaders (50%) want humans fully or mostly in control, and just 3% would hand the task entirely to AI. Fraud detection is where leaders are most comfortable with automation, with 46% happy for AI to take on most or all of the work. Payment approvals sit at the other end: 56% want humans fully or mostly in control, and only 4% favor full automation.
Setting these boundaries successfully requires explainable AI in finance: systems that show their reasoning so every recommendation can be checked and audited. CFOs and stakeholders need to understand why an approval was flagged, how a categorization was made, what triggered an exception. This accountability is foundational to financial control and regulatory compliance.
This is why the most sophisticated implementations separate tasks into three categories:
This framework matters even more as AI agents in finance take on end-to-end work, from capturing and checking receipts to reconciling and recording spend. Clear boundaries keep finance teams in a supervisory role, with people accountable for the decisions where judgment matters most.
One cost rarely gets quantified: the cost of missed opportunity. In the UK, cumbersome approval workflows and ineffective spend controls cause teams to delay decisions and miss business opportunities. The average cost of a lost opportunity due to slow approvals is £231,220 per opportunity. Across a year, these costs compound significantly.
AI can speed up approvals and spend controls considerably, meaning fewer missed opportunities and less manual admin. But applying it to these processes means handing part of that control to technology, which is a significant shift for a function where every prompt carries potential regulatory risk. Data security in AI for enterprise finance is a real concern, and many finance leaders are understandably cautious. The research also shows that confidence is lagging well behind adoption:
Finance leaders clearly see where this is heading. What most teams need now is a practical way to build confidence, starting with the right training and a clear plan for where AI fits into their workflows.
Taken together, these findings point to the same conclusion: the value of AI in finance depends on how deliberately it's built into the way finance teams work.
Our latest research report sets out a practical, six-step plan for finance teams ready to turn AI adoption into lasting change, from choosing the right first use case to setting the human-AI boundary and measuring real value. It also includes a phased implementation timeline and survey findings from over 3,500 finance leaders across Europe.
Download A Transformation Moment for Finance: How UK Organisations Can Turn AI Adoption into Lasting Change here.