Question 1

What is NL2SQL?

Accepted Answer

NL2SQL is a platform that translates natural language business questions into semantically correct SQL queries. Unlike generic AI tools, it uses a RAG pipeline with an ontology layer and curated query matching to understand what your data actually means.

Question 2

How is NL2SQL different from ChatGPT or other AI SQL generators?

Accepted Answer

Generic AI tools translate words into SQL syntax. NL2SQL translates business intent — it knows that "revenue" means L_EXTENDEDPRICE × (1 − L_DISCOUNT), that "last quarter" must be dynamically anchored to today, and which tables to join. This domain knowledge comes from a vector knowledge base of schema metadata, business terms, and curated query examples.

Question 3

What databases does NL2SQL support?

Accepted Answer

NL2SQL currently targets Snowflake data warehouses. The demo uses a TPC-H commerce schema (DBT.DWH) with tables like LINEITEM, ORDERS, CUSTOMER, PART, SUPPLIER, NATION, and REGION.

Question 4

What are the confidence levels in NL2SQL results?

Accepted Answer

Every query result includes a confidence signal: Trusted (≥88% match to a curated query), Semi-Trusted (70–87% match, review recommended), or Untrusted (AI-generated or validation failed). This helps users know when to rely on results directly and when to review the SQL.

Blog

Subscribe to our blog

A Semantic Layer Is Necessary - But It Is Not Everything You Need

Multi-Agent RAG solution for Natural Language to SQL pipeline

Getting Started with NL2SQL

Understanding Natural Language to SQL

Snowflake TPC-H Schema Guide