Empowering the Modern Data Ecosystem: Unlocking Value Through the Dremio Semantic Layer

This session explores how Dremio’s AI-powered semantic layer—backed by the Dremio Enterprise Catalog and Apache Polaris—bridges complex, heterogeneous data sources with business-friendly views that everyone can understand. You’ll learn the “layer cake” pattern Dremio recommends: a physical/raw zone, a staging layer for standardization and cleansing, a business layer where shared business rules live, and an application layer that BI tools and AI agents actually query. We’ll walk through how to design these namespaces and folders so you can plug-and-play data sources (e.g., moving from Oracle to Iceberg), reuse building blocks across use cases, and ensure every team—and every AI agent—gets the same answer to the same question.

From there, we dive into the practical value of this semantic layer for governance, performance, and AI. You’ll see how fine-grained access control, row/column security, and clear separation of responsibilities (data engineers, SMEs, and application consumers) create a governed self-service environment. We’ll also cover how Dremio’s data lineage and autonomous reflections work hand-in-hand with the semantic layer to optimize queries transparently, and how semantic search, business terms, wikis, and MCP-based integrations give agentic AI the business context it needs to generate high-quality answers. If you’re building a modern, AI-ready data ecosystem, this talk shows how to structure your semantic layer so humans and agents alike can discover, trust, and accelerate value from your data.