RAG + Apache Iceberg™: Structured Retrieval Made Easy
summary
DonutBot maps free‑form donut orders to real menu items by combining semantic vector search over product descriptions with exact filters pulled from Apache Iceberg tables stored on S3 and exposed via a catalog linked database in Snowflake so no data movement is required. Cortex Analyst converts natural language to SQL against a semantic view; Cortex Code auto‑edits that view to add a time‑of‑day dimension so the agent can answer “most popular in the morning.” TruLens and an LM judge flag logical inconsistencies, and iterative fixes reduce hallucinations and produce more grounded, data‑backed recommendations.
AI-written summary · about
chapters (5)
- 0:00Introduction and setup
- 4:02Demo overview and data architecture
- 12:09Building the AI agent
- 22:23Evaluating and improving the agent
- 25:39Q&A session
related talks
30mAI-Driven Operational Data Intelligence
16mTransforming Analytics With AI: A Comprehensive Guide to an Advanced Tech Stack
40mData Intelligence Platform on AWS
14mMulticloud Analytics Made Simple
24mBuilding Your Intelligent Data Agent
15marchive – Manohar Sai Jasti
15mFrom Air Quality to Aircraft and Automobiles, Data Is Everywhere
52mSerg Masís | AI Track | Compass AI & Data Summit 2025
