How AI is reshaping financial planning and budgeting processes

In the ever-automating world of Enterprise Performance Management (EPM), it is no surprise that AI – overall the next step in automation – is gradually gaining a foothold. It is, however, important to recognize that certain tasks are far more complex, making them less suitable for full automation.

Enterprise Performance Management covers four major interconnected domains: planning, budgeting, forecasting, and reporting. As a strategic asset, EPM delivers criticial insights into business processes that enables organizations to steer effectively toward their goals..

EPM also connects strategic objectives with operational execution, ensuring that all resources are aligned to implement and adapt the organization’s vision. This alignment requires seamless integration of practical, bottom-up processes with top-down defined KPIs.

Artificial intelligence (AI) refers to the field of computer science focused on creating systems or machines that can perform tasks typically requiring human intelligence. These tasks include learning from experience, understanding natural language, recognizing patterns, and making or suggesting decisions.

AI encompasses a range of technologies and approaches, such as generative AI, the use of AI to create new content, machine learning (where algorithms improve their performance over time), and neural networks, designed to mimic the human brain’s processing. The ultimate goal of AI is to enable machines to perform complex functions autonomously and adaptively.

EPM should ideally be a continuous process, given the ever-changing business environment that confronts organizations with a challenging strategic landscape. Sophisticated automated support systems are essential. Clever use of EPM tools entails a variety of benefits, particularly in terms of efficiency and transparency.

This brings us to the pivotal role of AI in the pursuit of high-quality automation. But is AI's transformative impact on the EPM landscape already a reality? How far can it realistically advance areas like planning and budgeting? And what emerging machine learning applications are beginning to reshape the finance domain?

Although not all EPM solutions have fully integrated artificial intelligence in their engines, many technology players are already experimenting with AI at various levels. Depending on a company’s specific needs, the presence or absence of integrated AI features is becoming an influential factor in EPM software selection.

"I wouldn’t consider AI the top criterion for choosing EPM software at this stage, but it’s definitely a factor worth considering," says Maarten Lauwaert. "As companies place more emphasis on AI, vendors will increasingly invest in innovations, which will lead to a steady rollout of new functionalities."

Exciting times lie ahead in the EPM landscape, that’s for sure.

"It’s crucial to point out that there are major differences between AI tools, functionalities and capabilities," explains Maarten Lauwaert. "We’re looking at a broad spectrum where a realistic approach is essential. Understand what tools can achieve based on your specific requirements, but also be aware of their current limitations."

One practical application of AI which is gaining traction is Microsoft Copilot, which positions AI as an assistant to users—a "sidekick" that helps you find your way in EPM tools. Copilot can retrieve information from vast datasets, generate workflows, conduct basic analyses, and prepare reports, providing practical support for daily tasks.

"Although these technologies may not be revolutionary, they definitely boost productivity," Lauwaert emphasizes. When used correctly, tools like Copilot can become virtually synonymous with operational efficiency.

Machine learning models are increasingly making their way into EPM software. However, simpler tasks—like predicting customer payments or churn rates—consistently outperform more advanced (predictive) models in terms of accuracy. Models that produce integrally automated planning or far-reaching cash flow forecasting with highly detailed ‘what if’ scenarios, do not exist at this moment and may never fully materialize.

Maarten Lauwaert, Practice Leader of Data & Analytics at TriFinance, provides insight: “Generally speaking, the closer the focus is to operational tasks within the EPM flow, the more accurate and effective AI-driven models tend to be. This makes it far more feasible to apply machine learning to specific sub-areas of EPM, creating strong automated foundations to build upon.”

Lauwaert emphasizes that a degree of human interaction will likely always be necessary, even as AI enhances efficiency in certain areas. This balance between automation and expertise underscores the current potential and limitations of machine learning in the EPM landscape.

EPM should never be a silo, it thrives as part of an interconnected ecosystem, where automation and artificial intelligence can impact broader organizational processes and reverberate within EPM. Think of the many (master) data flows from all corners of the data household, all of which are essential to arrive at the appropriate strategic roadmaps.

AI’s ability to identify patterns—particularly anomalies—makes it a powerful tool for enhancing data quality. One of the known strengths of AI is its ability to identify anomalies and (deviant) patterns. That makes it a powerful tool for enhancing data quality. By continuously scanning for outliers in databases and triggering automated alerts when irregularities are detected, AI tools act as a watchdog for data integrity. This, in turn, ensures that EPM processes are built on accurate, reliable data, ultimately improving strategic decision-making and organizational outcomes.

“Artificial intelligence will eventually become even more interesting when the focus on transparency and substantiation intensifies”, says Maarten Lauwaert. “Then the tools won’t just deliver numbers or analyses. They will also explain how they have arrived at those results.”This shift could unlock advancements toward more complex tasks, such as scenario planning, and perhaps even the step from predictive to prescriptive analytics. Therefore, AI-driven software could offer actionable recommendations.

Crucial to this, though, is the robustness of the technology. “When working with highly complex functionalities and calculations, all those models and underlying mechanisms must be implemented with absolute precision," Lauwaert emphasizes.

A second consideration is the adaptation required from users. As these innovations introduce new parameters, functions, and ways to handle the data, organizations must invest in building user skills. . “Fortunately, a digital assistant like Copilot can lower the threshold”, says Maarten Lauwaert, who nevertheless acknowledges that for many people it will be necessary to adopt new skills.

Read more on AI and EPM in our blog post.

Picture by Freepik