.gif)
Artificial intelligence has become the unavoidable topic at every finance conference, every FP&A webinar, and in every software vendor brochure. The promises are spectacular: automated forecasts more accurate than those of your best analysts, reports generated in seconds, anomalies detected before your teams even look at the numbers, and a finance function finally freed from repetitive tasks to focus on strategy.
Some of these promises are beginning to materialize. Others remain largely theoretical. And several are outright exaggerated by technology vendors more interested in generating excitement than in honestly informing their potential clients.
For CFOs, FP&A leaders, and finance teams looking to navigate this transformation intelligently, distinguishing reality from hype is a strategic skill. We offer an honest and nuanced look at the current state of AI in FP&A — what truly works today, what is still in development, and what finance teams should do right now to position themselves favorably in this evolution.
What Is AI in FP&A, Really?
Before evaluating the impact of AI on the FP&A function, it is useful to clarify what the term actually covers in this context. AI is a generic term that encompasses several distinct technologies, whose maturity levels and applications vary considerably.
Machine learning is the most mature and widespread technology in current FP&A applications. It relies on algorithms that learn from historical data to identify patterns and produce predictions. This is the technology that underlies the automated forecasting features offered today by platforms like Workday Adaptive Planning and Vena Solutions.
Natural language processing (NLP) allows systems to understand and generate text in human language. In the FP&A context, it manifests in conversational interfaces that allow financial data to be queried in natural language — "Which department deviated most from its budget this quarter?" — and in the automatic generation of variance commentary.
Large language models (LLMs) — the best known of which are GPT and its equivalents — represent the most recent generation of AI technology. Their ability to generate coherent, contextual text opens new possibilities for synthesizing financial data, generating narratives, and assisting in report writing. But their integration into FP&A workflows is still in its early stages for most organizations.
AI-based anomaly detection uses algorithms to automatically identify transactions, trends, or ratios that deviate significantly from historical patterns — an application that is particularly valuable for financial control and internal audit.
What Truly Works Today: Mature Applications
Automated Forecasting and Improved Forecast Accuracy
This is probably the most mature and widespread AI application in FP&A today. Machine learning algorithms can analyze complex time series — historical sales data, seasonality, macroeconomic trends, external variables — and produce statistical forecasts that frequently outperform traditional manual approaches, particularly for organizations managing a large number of product lines, geographic segments, or cost centers.
The advantage is not only accuracy — it is also speed and scalability. A forecasting algorithm can process thousands of lines of data in seconds, where a human team would spend days. For organizations managing complex planning models, this capability represents a real and measurable efficiency gain.
That said, this reality must be nuanced. Forecasting algorithms perform better when historical data is abundant, clean, and consistent. For companies undergoing significant transformation, whose business models are evolving rapidly, or for new product lines without historical data, automated forecasting produces less reliable results. Human judgment remains indispensable for integrating qualitative information — a change in commercial strategy, the arrival of a new competitor, a regulatory change — that algorithms cannot anticipate.
Automatic Anomaly Detection and Financial Control
Machine learning-based anomaly detection tools are today successfully deployed in many organizations to automatically identify significant deviations in financial data — unusual transactions, out-of-norm expenses, ratios that diverge from historical patterns.
The operational impact is concrete: financial control and internal audit teams can focus their attention on real anomalies rather than manually reviewing thousands of transactions to find the few dozen that merit investigation. This is an efficiency gain that improves both the quality of control and team productivity.
Automation of Repetitive Data Collection and Reconciliation Tasks
A significant portion of FP&A teams' time is spent on low-value-added tasks: extracting data from multiple systems, formatting it consistently, reconciling discrepancies between systems, and consolidating spreadsheets. These are precisely the tasks that automation — whether described as AI or simply as RPA (Robotic Process Automation) — can take on effectively.
Organizations that have automated these workflows report significant time savings on the monthly close cycle and on the production of management reports. These gains are not trivial — they allow analysts to devote more time to actual analysis, scenario modeling, and strategic conversations with management.
Automatic Generation of Variance Commentary
NLP-powered tools can today automatically generate explanatory comments on variances between actual results and budget — "Segment A revenues are 8% below budget, primarily due to a slowdown in Western European sales" — drawing on the structured data available in financial systems.
These automated comments do not replace the in-depth analysis of an experienced analyst, but they represent a first layer of explanation that accelerates the reporting process and frees up time for root cause analysis and the formulation of recommendations.
What Remains a Promise: Applications Still in Development
Truly Autonomous Forecasting
The idea of a system that produces complete, reliable forecasts without human intervention — automatically integrating all relevant variables, dynamically adjusting its assumptions, and producing actionable recommendations — is compelling but still largely theoretical for the vast majority of organizations.
Current forecasting systems are powerful assistance tools, not replacements for financial judgment. They produce good statistical forecasts under stable conditions, but they struggle to integrate unstructured context changes and qualitative information that any experienced financial analyst naturally takes into account. Human oversight remains indispensable — and will remain so for several years to come.
Generative AI as an Advanced Financial Modeling Tool
The use of LLMs to build complex financial models, generate sophisticated sensitivity analyses, or produce autonomous strategic recommendations is still largely experimental. The tools available today can assist an experienced modeler, suggest model structures, or generate VBA code — but they cannot replace the expertise of a financial modeling consultant for complex mandates.
The specific risk of LLMs in a financial context is that of "hallucinations" — the generation of plausible but factually incorrect content. In a field where numerical precision and rigorous assumptions are critical, this risk requires systematic human validation of all AI-generated outputs.
Causal Analysis and Autonomous Strategic Recommendations
Current AI systems are good at identifying correlations in data — this variable is associated with this outcome. They are much less effective at identifying causal relationships — this variable causes this outcome — and even less so at formulating strategic recommendations that integrate qualitative, political, and organizational considerations.
The ability of an AI system to say "your gross margin has declined because your product mix has evolved unfavorably, and here are the three actions you should take to correct this" remains limited in real-world contexts. This analytical and prescriptive dimension is precisely the one that creates the most value in the FP&A function — and it is also the one that best resists automation.
What Is Clearly Overstated: Promises to Temper
The Disappearance of the FP&A Analyst Role
The prediction that AI will eliminate financial analyst positions is regularly revived in the trade press. It deserves to be seriously tempered.
What will disappear — and is already disappearing — are the repetitive execution tasks that many analysts themselves find unrewarding: manual data collection, file reconciliation, production of standardized reports. What will not disappear — and will actually grow in importance — is the ability to interpret data in context, formulate nuanced recommendations, communicate financial insights to non-financial stakeholders, and exercise the professional judgment that algorithms cannot replicate.
The transformation of the FP&A function by AI looks more like a reskilling than an elimination. Analysts who embrace AI tools and develop their ability to work with them — rather than resisting or ignoring them — will find themselves in a far stronger professional position than those who do not evolve.
Instant, Frictionless Implementation
AI solution vendors often present their products as plug-and-play solutions that transform financial operations within a few weeks. The reality on the ground is invariably more complex. The quality of existing data is rarely sufficient to feed AI algorithms without significant prior work on cleaning and structuring. Integration with existing systems takes time. Team adoption requires training and support. And results improve progressively as algorithms accumulate data — not overnight.
How Finance Teams Should Prepare: Five Concrete Priorities
Priority 1: Invest in Data Quality Before AI Tools
This is the most fundamental and most frequently overlooked priority. AI algorithms can only produce reliable results from reliable data. An organization whose financial data is fragmented across multiple systems, poorly normalized, incomplete, or unreliable will derive no substantial benefit from even the most sophisticated AI tools.
Before investing in an AI solution, honestly assess the quality of your data: is it centralized, consistent, complete, and historically sufficient to train algorithms? If the answer is no — and it often is — foundational data work is the first step, not the acquisition of AI tools.
Priority 2: Develop Hybrid Skills Within the Team
The highest-performing FP&A teams of tomorrow will be those that combine strong financial analytical competence with ease in using AI tools and the ability to interpret and validate algorithmic outputs. This combination — financial expertise and AI literacy — is rare today and will be valuable tomorrow.
Investing in training your teams on the fundamental concepts of machine learning, on the AI tools available in your technology ecosystem, and on best practices for validating algorithmic outputs is a strategic priority. This training does not need to be technical — it needs to be deep enough that your analysts can use the tools intelligently and understand their limitations.
Priority 3: Start with Proven-Value Use Cases
The temptation is to approach AI as a comprehensive transformation — changing everything at once, adopting the most complete platform, aiming for the most ambitious transformation. This approach often generates heavy, costly, and disappointing projects.
A more effective approach is to identify two or three specific use cases where AI can deliver clear, measurable value in your context — improving the accuracy of revenue forecasts, automating monthly reconciliation, detecting anomalies in expense reports — and start there. Concrete successes build confidence, develop skills, and create momentum for more ambitious applications.
Priority 4: Redefine the FP&A Analyst Role
If AI is taking on an increasing share of execution tasks, the question of what the FP&A analyst should do with their freed-up time becomes strategic. Organizations that do not answer this question explicitly risk seeing their teams fill that time with new execution tasks rather than value-added activities.
Redefining the FP&A analyst role to focus more on data interpretation, formulation of recommendations, communication with operational stakeholders, and strategic advisory to management is an organizational design exercise as important as the choice of technology tools.
Priority 5: Choose Technology and Human Partners Who Understand Your Context
Adopting AI in FP&A is not a technology project — it is an organizational transformation project that touches the processes, skills, roles, and culture of the finance function. Organizations that succeed in this transformation generally rely on partners — consultants, integrators, FP&A experts — who understand both the technology and the financial stakes, and who can support the change holistically.
Modelcom's Position in This Transformation
At Modelcom, we have been observing and supporting the transformation of the FP&A function since 1996. We have seen many technology waves come and go — the adoption of ERPs, the democratization of Excel, the emergence of BI tools, the deployment of dedicated FP&A platforms — and AI is not fundamentally different in its structure: a technology that amplifies human capabilities without replacing them, whose value depends on the quality of the foundations on which it rests.
Our approach to AI is consistent with our general philosophy: pragmatic and decision-oriented. We help our clients identify AI applications that create real value in their specific context — not the most impressive applications on paper, but those that concretely improve the quality and speed of financial decisions.
We work with our partner platforms — Workday Adaptive Planning and Vena Solutions — which integrate mature AI capabilities into proven FP&A environments. And we support our clients through the human transition that is often more decisive than the technology transition: helping finance teams evolve toward more analytical and strategic roles, developing the hybrid skills that will make the difference, and adopting new workflows without losing the rigor and reliability that are at the heart of any credible finance function.
AI Is an Amplifier, Not a Substitute
Artificial intelligence will transform the FP&A function in significant and lasting ways. Some of these transformations are already underway. Others will arrive progressively over the coming years. A few will promise more than they deliver.
What will not change is the fundamental value of an FP&A team: the ability to transform financial data into actionable insights, to communicate those insights with clarity and credibility, and to support decision-makers through the most complex choices their organization faces. AI makes the teams that do this work more effective — it does not do it for them.
Finance teams that approach this transformation with clear-headedness — distinguishing what truly works from what is still a promise, investing in the right foundations, and developing the hybrid skills that will be valuable tomorrow — will be the ones who benefit most from it. And that is precisely the support that Modelcom is positioned to offer.
Want to explore how AI can concretely improve your organization's FP&A processes? Contact the Modelcom team for a no-obligation conversation about your specific context.