Forget SQL! Create your own AI Data Analyst
summary
An AI Data Analyst connects your database to LLMs and returns SQL plus results: implemented with Gemini + Vanna, Kaggle notebooks, a Vector DB and Compute Engine, and it queries Cloud SQL/BigQuery/Snowflake. It indexes 200+ tables (~600k tokens) to retrieve only relevant schemas, generates SQL, executes read-only queries, and limits outputs (30 rows) to avoid hallucinations. Deployment runs internally, costs are tracked, and the system is in internal test mode with thumbs-up/down feedback.
AI-written summary · about
chapters (9)
- 0:00Introduction and motivation
- 10:10Initial attempts with LLMs
- 14:22Google AI Studio exploration
- 17:55BigQuery SQL generation
- 21:24Demo of AI analyst in Kaggle
- 24:52Vanna AI implementation
- 28:02SQL assistant evolution
- 33:41Alternative tools and approaches
- 35:04Q&A session
related talks
56mDeveloping AI Applications with SQL AI: Ground to Cloud to Fabric
50mLeveraging AI as a PostgreSQL DBA
01h 00mWorkshop: Getting Data Ready for AI - With Unified GraphQL &SQL Endpoints
14mMulticloud Analytics Made Simple
40mData Intelligence Platform on AWS
50mBuilding AI ready applications
30mRyoma: Building Open Source AI Agents for Enterprise Data Analysis
30mMulti-tenant Conversational Analytics