Four years after generative AI entered the mainstream, the question for finance leaders is whether the foundations underneath it are strong enough to make its outputs trusted, explainable and useful.
With almost every board now asking how quickly they can deploy artificial intelligence (AI), the pressure to flip the switch is intense. In the past year alone, 76% of companies have appointed a Chief AI Officer, up from just 26% in 2025. But against this rush to adopt, one uncomfortable truth is becoming harder to ignore: AI does not fix a broken finance stack; it exposes it.
An algorithm is only ever as effective as the systems, processes and data that underpin it. This is a major source of frustration for finance leaders, many of whom are still tethered to fragmented technology stacks, disconnected data sources, and manual workflows. Built on these weak foundations, AI cannot deliver deeper insights. Instead, it simply digitises existing inefficiencies and surfaces hidden structural issues faster. Many finance teams are far more underprepared for AI than they realise.
The spreadsheet layer AI cannot fix
Spreadsheets have incredibly deep roots in enterprise finance. They’re familiar, flexible and accessible. But they were never built to be the unofficial nervous system of a finance team.
The reality for many enterprise finance teams is a chaotic middle layer of systems, spreadsheets and reporting tools. Every closing cycle becomes a marathon of manual reconciliation, with teams forced to unpick dozens of conflicting versions of performance. Data is copied between systems, formulas evolve over time, and vital assumptions become almost impossible to trace.
Introducing AI to that foundation does not solve the problem. It scales it. Rather than draw insights, it will simply generate the wrong answer faster. Automated recommendations may appear sophisticated at the surface level, but each one requires multiple checks for transparency and integrity. Instead of liberating busy finance professionals to focus on higher-value analysis, it adds another layer of administrative overhead. ‘Garbage in, garbage out’ is a truer principle than ever.
What’s concerning is that this infrastructure deficit is driving a severe talent crisis. Today, an alarming one in three finance professionals quit within their first year, overwhelmed by managing fragmented systems and exhausted by manual data stitching. Highly skilled experts are being hired to drive strategic enterprise growth, yet they are burning out on clerical spreadsheet maintenance. Layering AI over this chaos only turns already stretched teams into full-time fact-checkers, forced to reconcile versions, audit formulas and validate outputs that should have been trustworthy from the start.
In light of this, finance data foundations have become a strategic priority. That means more than cleaner data. It means common definitions, governed business logic and planning, reporting and forecasting processes that all work from the same version of performance. Organisations that invest in this foundation create the conditions for AI to deliver meaningful outcomes, while improving transparency, governance and trust in financial decision-making.
How to evaluate AI readiness
Most organisations mistake a full toolbox for true AI readiness. They enthusiastically check off boxes like enterprise model access, pilot programmes, or employee adoption of generative AI tools like ChatGPT or Claude. Yet, this fixation on shiny front-end tools completely misses the important infrastructure underneath.
Real readiness requires looking past the interface and interrogating the finance model itself. It means asking hard questions. Do financial figures match perfectly across every disparate system? Can teams trust the numbers on their screens, or are planning and forecasting processes still heavily siloed? True maturity requires a single, unshakeable source of truth for business performance.
When those foundational answers are missing, deploying AI becomes a risk. Finance leaders never lack ambition, but trying to build a sophisticated, predictive future on top of unstable foundations that were never engineered to hold the weight can be an executive liability.
Readiness, therefore, isn’t a software procurement challenge. Before budgeting for advanced algorithms, leadership teams must audit the architecture that feeds them. If the middle layer is broken, it must be fixed before it can be automated. Taking a step back to build a unified and governed data core is, ironically, the biggest step forward.
The future of finance undoubtedly involves AI. Its ability to improve forecasting and support strategic decision-making is too significant to ignore. But by creating an unshakeable foundation first, finance leaders can ensure that when the switch is flipped, the machine multiplies clarity rather than compounding existing chaos.
