Every few years a new technology promises to transform the finance function. Artificial intelligence is the current candidate, and the pitch is familiar: faster forecasts, fewer spreadsheets, sharper decisions. A recent IMA article by FP&A consultant Ayman Abdellatef is more useful than most on the subject, because it is honest about the plumbing beneath the promise.
His starting point is financial planning and analysis (FP&A) as it actually runs in most organisations: spreadsheet-based planning, static budgets, and variance analysis done after the fact. Budgets are assembled in silos, each team preparing its own segment before everything is stitched into a master plan. Because sales and operations cycles move faster than budget revisions, the numbers finance reports tend to lag what the business is actually doing - and cautious manual adjustments widen the gap.
That lag is the problem AI is meant to solve, and on paper it solves it well.
Abdellatef points to capabilities that are real rather than aspirational. AI can aggregate and clean data from scattered sources, removing much of the manual preparation that eats the first week of every reporting cycle. It can re-forecast close to real time as new data lands, rather than once a quarter. It can run what-if scenarios in minutes instead of days. And because it can connect to the tools other departments already use, it can begin to break the silos that keep finance working from stale numbers.
The evidence he cites is encouraging. According to Gartner research quoted in the piece, 64% of finance teams that have adopted AI say its impact has met or exceeded expectations, pointing to better accuracy, faster insight and greater efficiency. His headline case is JD.com, the Chinese e-commerce group, which reached an 85% automation rate in procurement, cut inventory turnover to 30 days, and - through AI-driven cost measures - reported a 92% rise in net income in the second quarter of 2024.
Impressive numbers. They also come from one of the largest, most data-rich companies on earth. That is the part worth pausing on.
AI does not create a good finance function; it scales whatever function you already have. Feed it clean, connected, well-governed data and it will produce faster, sharper analysis. Feed it fragmented spreadsheets and inconsistent definitions, and it will produce fast, confident, wrong answers - which is arguably worse than slow ones.
The author is candid about the risks, and they are not minor. Poor data quality erodes trust in every output. Models trained on historical data can carry forward the biases embedded in it, skewing decisions on sensitive matters such as lending or how resources are allocated. And because these systems extrapolate from the past, they tend to struggle precisely when it matters most - during the unprecedented events, from pandemics to geopolitical shocks, where judgment beats pattern-matching.
So the human role does not shrink so much as change. Abdellatef's framing is that finance professionals move from number crunchers to strategic partners: validating what the model produces, supplying the context it lacks, and owning the decisions it informs. The tooling grows more powerful; the accountability stays human.
For a founder or a lean SME finance team, the honest takeaway is not ‘buy an AI platform.’ It is that the return on any of this is set by the quality of what sits beneath it. A messy chart of accounts, three disconnected systems and a definition of revenue that shifts by department will not be rescued by a model - they will simply be automated, faster.
So sequence matters. Get the numbers clean and consistent first. Then pick one genuine pain point - a forecast that is always late, a cash position nobody trusts - and pilot against that, as the article sensibly recommends, rather than trying to transform everything at once. This is unglamorous work, and it is exactly the groundwork that lets a smaller business borrow the leverage that larger ones get from sheer scale. It is also where a good fractional CFO earns their keep: fixing the plumbing and asking the sharper questions, so that whatever you layer on top has something solid to stand on.
Abdellatef closes by arguing that the future of FP&A is not just automated but empowered. Fair enough. But empowerment assumes there is something worth empowering. The businesses that get the most from AI in finance will be the ones that did the dull, foundational work first.
Source: How AI is Redefining the Future of Financial Planning and Analysis - Ayman Abdellatef, IMA (Institute of Management Accountants), 2025.
