archive – Manohar Sai Jasti
summary
Claims that large language models turn questions into near‑instant SQL answers and cut wait times from hours or days to seconds, demonstrated with a Jaffle Shop demo that actually runs queries. Shows a production design using natural language → SQL, a semantic layer via dbt to supply table/metric definitions and cut hallucinations, and RAG to scale by sending only relevant metadata. Presents a four‑agent loop and cites a 15% change in returning users as impact.
AI-written summary · about
chapters (8)
- 0:00Conversational data analytics intro
- 2:58LLMs need semantic layers
- 5:30dbt for semantic layers
- 7:39Structured data for LLMs
- 10:07Scaling LLMs with RAG
- 11:57How conversational analytics works
- 12:40Benefits for teams
- 13:04From questions to insights
related talks
30mBoosting Sales with Intelligent BI Agents
30mThe Future of Data is Words
13mRAG Without the Hassle: Building AI-Powered Applications
47mHow to query data using natural language - intro to AI features in Oracle 23ai
56mDeveloping AI Applications with SQL AI: Ground to Cloud to Fabric
33mOpening Keynote: LLMs, between fantasy and reality - the poin...
45mManaging Data & Analytics Projects Effectively
24mBuilding Your Intelligent Data Agent