NL
Natural Language
RAG
Retrieval
ChromaDB
Vector Store
Auto
Chart Gen
Business intelligence agent that lets you query your data in plain English using RAG over a ChromaDB vector store. Ask questions like 'which products drove Q3 growth?' and get SQL-backed answers with chart generation — no BI tool required.
Skip the BI tool. Ask your data questions in English. Get SQL-backed answers with charts.
Tech Stack
PythonLangChainChromaDBStreamlitPandasDuckDBPlotlyOpenAI
Features
NL to SQL
Natural language query → RAG retrieval of schema context → SQL generation → execution.
ChromaDB Context
Business glossary, schema metadata, and query examples stored in ChromaDB for retrieval.
Auto Chart Gen
Results automatically visualized with Plotly. Picks chart type based on result shape.
LangChain Agent
ReAct agent handles multi-step queries: join, filter, aggregate — all from one natural language request.
How It Works
01
Query
User question → LangChain agent + ChromaDB schema retrieval
→
02
Generate
LLM generates SQL using retrieved schema context
→
03
Execute
DuckDB executes SQL → structured result set
→
04
Visualize
Plotly auto-generates chart → Streamlit renders answer