How visheshjain28/insightsql works
InsightSQL is a Streamlit natural-language analytics application: a user submits a question, the app generates SQL from database schema context, executes it, and presents tabular, visual, and AI-summary results. The UI also exposes schema exploration and an evaluation-report tab. The sampled engine code supports the generation, execution, and summary links shown; the README describes RAGAS evaluation, but the sampled evaluation UI displays fixed metrics and example rows, so no live evaluation pipeline is inferred.
Main entry points: app.py Analytics UI · sql_engine.py SQL Execution
Components 14
Streamlit Interface
- Analytics UI
app.py - Schema Explorer
app.py - Evaluation Report
app.py
Query Engine
- DB Connection
sql_engine.py - SQL Chain
sql_engine.py - Schema Context
sql_engine.py - SQL Execution
sql_engine.py
Analytics Results
- Query Results
sql_engine.py - Data Table
app.py - Interactive Charts
app.py - Business Summary
sql_engine.py
External Systems
- MySQL Database
- Gemini
Other components
- User
Connections 18
- User → Analytics UI: asks question
- Analytics UI → DB Connection: initializes connection
- Analytics UI → SQL Chain: creates chain
- Analytics UI → SQL Execution: submits question
- DB Connection → MySQL Database: connects
- SQL Chain → Schema Context: retrieves schema
- Schema Context → MySQL Database: reads metadata
- SQL Chain → Gemini: requests SQL
- SQL Execution → SQL Chain: invokes
- SQL Execution → MySQL Database: executes SQL
- SQL Execution → Query Results: returns results
- Analytics UI → Schema Explorer: renders tables
- Schema Explorer → MySQL Database: inspects schema
- Analytics UI → Business Summary: requests summary
- Business Summary → Gemini: summarizes results
- Query Results → Data Table: renders rows
- Query Results → Interactive Charts: plots data
- Analytics UI → Evaluation Report: shows report