SQL For Everything: Structure, Unstructured, SQLServer, Onelake

SQL has long been the most powerful and accessible language for data analysis, yet its strengths have been largely confined to structured tables. Meanwhile, the majority of enterprise data now lives outside that boundary, spread across text, documents, logs, JSON, and files in cloud object storage. This session explores how SQL can evolve for the AI era, becoming a universal interface for all data, not just rows and columns.

Using real examples, this talk introduces a new class of AI-native SQL functions on the Dremio Lakehouse Platform that allow analysts and engineers to classify text, summarize documents, and generate structured insights directly inside SELECT statements. Attendees will see how these capabilities eliminate the need for separate AI pipelines, scripts, or tooling, enabling unstructured and structured data to be queried, joined, and governed together.

The session also demonstrates how a single SQL engine can unify the Microsoft data ecosystem, spanning Azure Storage, SQL Server, and Microsoft OneLake via open standards like Apache Arrow and the Iceberg REST Catalog. By combining a governed semantic layer, AI-powered SQL, and high-performance query execution, organizations can simplify architecture, reduce cost, and deliver sub-second analytics across their entire data estate.

Finally, the talk connects these ideas to the rise of agentic analytics, where both humans and AI agents rely on SQL as a shared reasoning layer over trusted enterprise data. Attendees will leave with a clear understanding of how “SQL for Everything” becomes a practical foundation for AI-ready analytics, not a distant future concept.

To demonstrate the power of AI functions and how to leverage them on their Microsoft real estate.