Talk to Your Data - A Semantic Layer Meets Agentic Graph-RAG for Conversational Analytics

Getting an LLM to answer real business questions over a data warehouse isn't a SQL problem — it's a meaning problem: which of your hundreds of tables does "margin on Milano Fashion, excluding returns" actually touch? In this talk we show how we organize the semantic layer — tables, join paths, and named business metrics — into a knowledge graph, and how that graph lets an agent search the schema and follow relationships instead of drowning in the entire schema. The result is conversational analytics that stays accurate and cost-stable as the warehouse grows.
Format: TECHNICAL TALK, RESEARCH TALK
Location: ROOM B
Speakers: Luka Vitosevic
Level: Intermediate
Audience: Registered participants