According to the latest 28th PwC Global CEO Survey, nearly half of all CEOs expect generative AI to boost profitability as early as next year. Yet only 31 percent report systematically integrating AI into their workforce and skills strategies. One reason: while data exists, it’s not AI-ready.
Companies are already investing heavily in data lakes and reporting systems, but operational questions still take days or even weeks to answer. “Which customers are affected by the delivery delay?” a simple question that triggers complex ETL processes.
The paradigm shift: moving from Build & Query to Ask & Understand makes data conversational and enables natural interaction with business information in real time.
This transformation marks the shift from technical data processing to a semantic data culture, where information is no longer just moved but understood within its business context.
Rethinking operational performance: Understanding data instead of loading it statically into reports
Familiar scenario? A production manager asks on Monday: “Which orders are affected by the delivery delay from Asia?” The answer arrives later as a report.
In recent years, many companies have started developing numerous data products, often within a Data Mesh architecture, where business units take ownership of their own data products and make them available across the organization. What’s changing now is not just how these products are developed, but especially how they are used.
The Build & Query mindset was valid for a long time: collect data, model it, and deliver it via predefined reports. Today, however, it slows down operational decision-making. As market demands accelerate, companies still rely on IT specialists to run new analyses.
End users increasingly want to understand data deeply, not just consume it through prebuilt reports. Operational questions arise spontaneously, and users expect context-rich, well-founded answers. Usage is shifting from predefined evaluations to interactive queries that flexibly respond to individual questions.
The alternative: Ask & Understand doesn’t mean sacrificing data quality but gaining direct access. Semantic models and AI make information conversational, without compromising governance.
In the future, data products will be modeled to be interpretable by generative language models. What matters most is the semantic curation of content: clear descriptions of context, relationships, and meaning.
As a result, data products evolve from static reporting sources to dynamic, dialogue-based interfaces that actively contribute to value creation.
Technological Implementation: Semantic Data Architectures in Practice
The theoretical concepts of “Ask & Understand” can already be realized today using modern data platforms. One example is Microsoft Fabric, which brings together the essential components for semantic data architectures. Companies don’t need more reports, they need clear answers to operational questions: direct, contextual, and at the moment of decision-making.
The core challenge for many organizations: operational data is scattered across systems and cannot be efficiently utilized. Modern platforms enable cross-system analysis of control metrics: flexible, interactive, and AI-supported. Users receive answers to specific questions without relying on centralized reporting processes.
The following sections illustrate how “Ask & Understand” can be implemented using Microsoft Fabric.
Authors
