Text-to-SQL Agent
A multi-model GenAI system that converts natural language questions into SQL, executes them against Snowflake, and returns instant charts and insights.
Problem
Business users previously waited 24–48 hours for basic reporting and insights, creating friction between data teams and decision-makers and slowing execution.
Solution
Designed a multi-model GenAI system that allows users to query proprietary data using natural language, automatically generating SQL, executing it securely in Snowflake, and returning tables or visualizations in real time.
How it works
User submits a natural language question via UI (Chat-style interface).
Router / message broker evaluates intent.
Prompt-evaluator agent validates request.
SQL-generation agent converts prompt → executable SQL.
Query executes against Snowflake.
Visualization agent renders tables or charts.
Outcome
The system eliminated reporting lag entirely, reducing wait time from 24–48 hours to instant insights and enabling 0 hours wait time for business users. By providing self-serve analytics with no SQL required, the system reduced load on data teams while enabling instant iteration and faster decision-making.