About

Many organisations have valuable data trapped across legacy systems—making it difficult for employees to find answers quickly or act with confidence. This session introduces an agentic AI approach that sits between users and enterprise data, enabling teams to query information in natural language to get their answers from the different data sources. We’ll cover how a pragmatic data layer (e.g., warehouse/curated sources) can be combined with agentic capabilities that route questions to the right method: text-to-SQL for governed relational data, and semantic retrieval for documents and unstructured knowledge. Attendees will learn how to combine multiple data types—relational, documents, and selected external sources—into one coherent, user-friendly interface for non-technical users

Takeaways

• Practical patterns to break down data silos created by legacy systems
• Where text-to-SQL fits as a safe interface for non-technical users
• How to combine structured + unstructured + external data in one experience
• Design choices for an agentic layer that routes questions to SQL vs semantic search