Beyond the Prompt: Scaling AI Reliability with Semantic Models

Abstract

Natural language interfaces are transforming how people access data, but their reliability depends far more on modeling clarity than on prompt engineering. Just as retrieval-augmented generation (RAG) grounds AI for unstructured data, semantic models provide the essential grounding layer for structured data—bridging the gap between physical schemas and business meaning. Without this layer, AI is forced to infer relationships from table names and structures, often leading to inconsistent queries, incorrect joins, and answers that cannot be trusted.

This session explains why the semantic layer is emerging as the control plane for both AI and traditional analytics. You will see how AI interprets data with and without semantic models, and how ambiguity in relationships, metrics, and grain definitions leads to unpredictable behavior. We will break down the core components of high-quality semantic models—including governed business terminology, explicit relationships, clear aggregation rules, and embedded validation policies—that serve as guardrails for reliable and repeatable AI-generated queries.

Finally, the session addresses the realities of implementation at scale. As data ecosystems grow in complexity, semantic models must evolve alongside changing business logic and usage patterns. We will introduce practical approaches for reverse engineering semantics from existing data assets—leveraging metadata and observed query behavior to surface implicit business logic and strengthen the foundation for trustworthy AI-driven analytics.

Bios

Image

Nick Lee is a Lead Product Manager specializing in Data & AI initiatives, driving unstructured data strategy for global consumer entertainment experiences. Previously, he helped Fortune 500 companies solve complex AI problems at leading enterprise data and AI platforms. Nick was recently recognized as a Georgia Tech Alumni Association "40 Under 40" honoree.

Image

Nandakishor Koka is a Principal Technical Architect for Data & AI, specializing in data engineering, machine learning, and semantic modeling at an enterprise scale. He designs metadata-driven platforms, knowledge-graph-powered semantic layers, and AI systems that bridge raw data to business meaning for large-scale consumer ecosystems. With a Master’s in Machine Learning from Georgia Tech, he has led innovations in computer vision, complex analytics, and automated semantic lifecycle management.