Enterprise AI’s next test is business reasoning

Why executives need systems that can link decision making to performance

By: David Marmer, Chief Product Officer, Board

David Marmer is the Chief Product Officer at Board. With a strong background in SaaS and hybrid cloud solutions, David is focused on driving innovation and building world-class products that deliver transformational results. Prior to joining Board, he held executive roles at IBM and Cognos, where he led significant growth in analytics and performance management solutions.

In the world of business, artificial intelligence (AI) has found its clearest commercial success in tasks that increase efficiency and productivity: summarizing reports, answering questions, creating presentation decks and processing information at remarkable speed. However, current AI models cannot replicate the reasoning of a seasoned executive. An experienced chief financial officer (CFO), for example, will explore why things happened, whether the problem is structural or temporary, and what the business should do about it.

Enterprise management is fundamentally built on chains of cause and effect. Drops in revenue compress margins, inventory imbalances tie up working capital, and a single disruption can ripple throughout an entire supply chain, impacting pricing, revenue, and margins. Executives rarely make decisions based on isolated metrics, but on their understanding of a wider system: how one part of the organization affects another, and how those interactions affect the company’s financial and non-financial KPIs. Most AI systems still struggle to do this.

Unlike humans, AI-enabled enterprise technologies have difficulty navigating the reality that large organizations are complex and fragmented, and thus miss critical context scattered across spreadsheets, dashboards, and institutional knowledge. As a result, they may produce answers that sound convincing but do little more than identify correlations.

Finance may be the proving ground for AI reasoning

Understanding this business context is becoming an important milestone for next-generation enterprise AI. One area where these new models are making a genuine, tangible difference is finance.

Finance sits at a point where business decisions and financial outcomes meet. For example, a CFO may ask their team to look into why company EBITDA declined. The finance team will examine what’s behind the number: revenue drivers, cost absorption, inventory balances, cash flow, and operational variances. In their investigation, they build a causal model between business activity and the expected financial outcomes.

Today, AI-enabled decision-making systems like Board are replicating this process. Rather than just being a planning and forecasting tool, next-gen AI can follow structured reasoning workflows and move step by step through planning data, operational drivers, and financial relationships to provide executive leadership with both an answer and an explainable chain of reasoning.

Trust in AI requires a chain of reasoning

Clear, reasoned explanations matter. Executives are unlikely to rely on systems that cannot show how they reached a conclusion, particularly when those conclusions affect forecasts, capital allocation or performance targets.

What’s notable is that many of these capabilities are not emerging from entirely new AI architectures, but rather from the combination of planning models, enterprise semantic definitions, and carefully structured prompts.

Companies can define a set of steps for an AI system to follow when investigating a problem. This could include examining financial statements, identifying the operational factors behind a result and flagging uncertainty when the evidence is incomplete. Planning models provide the structure for the reasoning process.

Some companies are already seeing instances where AI is reasoning reliably and consistently across their organisations. For example, EBSCO, Ammega, and Chiesi are using Board AI to unify planning and operational data across finance and operations. The goal is not automation for its own sake, but to enable AI decision-making that can be explained, tested, and trusted across the business.

Enterprise AI needs context

The market is already saturated with copilots, agents, and conversational interfaces. While these tools are useful, they are becoming increasingly easy to replicate. The more difficult task is enabling AI to reason reliably across the complexity of a large organization. That requires more than access to data.

To succeed, companies must make themselves easier for machines to understand. This will help AI systems reason better and provide an explainable chain of reasoning to executives.

For decades, enterprise software has focused on storing information and automating workflows. The next generation will take on a more ambitious task: understanding why a business is performing as it is, and helping executives make better decisions.