Viewpoint: Building a Trusted Semantic Foundation for Generative BI
September 2, 2026
Generative BI promises to make analytics easier to access by allowing users to ask questions in natural language, receive explanations, and explore business problems without writing SQL or navigating complex tools.
However, large language models do not inherently understand an organization’s business definitions. They can interpret language and support exploratory analysis, but they are probabilistic. In traditional BI, inconsistent answers already undermine trust. In generative BI, the problem can spread faster because an AI system can generate and distribute answers at much greater speed.
To examine how organizations can address these challenges, TDWI spoke with Dave Mariani, co-founder and CTO of AtScale. Mariani argues that successful generative BI requires a semantic foundation that makes probabilistic AI behave consistently when it interacts with governed enterprise metrics.
Download this special Viewpoint today to learn about the need for business ownership, evaluation, explainability, and security as organizations move toward agentic analytics.