Data Fabric Boosts AI Business Value

While AI technologies are rapidly spreading across enterprises, data quality and context are emerging as the biggest obstacles to realizing business value. It is estimated that by the end of 2025, more than half of companies will adopt AI for at least three business functions. Various AI solutions, such as chatbots and predictive systems, are increasingly being integrated into core operations including finance, supply chain, human resources, and customer operations.

E. Fan Kang, Chief Product Officer at SAP, pointed out that while AI can process vast amounts of data quickly, a lack of understanding of the business context embedded within the data can lead to incorrect judgments and hinder return on investment (ROI). To ensure AI systems make accurate decisions and reflect business priorities, building a 'data fabric' that goes beyond simple data integration is essential.

Many companies are re-evaluating their data architecture, seeking ways to connect information across applications, clouds, and operational systems rather than moving data to a single repository. Preserving semantic information that explains how the business operates is crucial in this process. This shift is driving increased interest in data fabric as the foundation for AI infrastructure.

Traditional data strategies have primarily focused on data integration. For the past two decades, companies have invested heavily in extracting information from operational systems and loading it into centralized warehouses, data lakes, and dashboards. While this approach facilitated running reports, monitoring performance, and gaining business insights, it resulted in the significant loss of the meaning attached to the data – its relationship to policies, processes, and actual decision-making.

For example, a comparison of two companies using AI to manage supply chain disruptions reveals a difference. One company might rely solely on raw signals like inventory levels, lead times, and supplier scores, while another company adds context across business processes, policies, and metadata. Even if both AI systems analyze the data quickly, they are likely to reach different conclusions. Information such as which customers are strategic accounts, what are acceptable trade-offs when stock is low, and the status of an extended supply chain allows one AI system to make strategic decisions, whereas the other cannot due to a lack of appropriate context. CPO Kang explained that while both systems move fast, only the system with context is heading in the right direction.

The 'context premium' refers to the benefits gained when the data foundation preserves context across processes, policies, and data from the design stage. In the past, human experts implicitly managed this missing contextual information, but in the age of AI, the absence of this contextual information poses severe limitations. AI systems need to understand and interpret data. Without this contextual understanding, the rapid outputs provided by AI are unlikely to translate into tangible business value.