Healthquant, an AI analyst that queries databases and builds interactive dashboards
Conversational data analysis with SQL queries, interactive charts, and shareable dashboards.
| When | Sep 2024 to Feb 2025 |
|---|---|
| Role | Software engineer |
| Status | Built |
| Tools | TypeScript, React, Next.js, Vercel AI SDK, PostgreSQL, Drizzle, SQLite, React Flow, Recharts |
| Link | Source code on GitHub ↗ |
I built Healthquant to let people ask questions about structured data and explore the results through a conversation. The agent queries a connected database, interprets the returned data, and creates charts alongside its answers.
Users can upload a SQLite database or configure a database connection. The agent inspects the schema, generates SQL, and can repair a failed query using the available tables and columns. The SQLite query path restricts execution to SELECT statements.
Charts can be added to a dashboard and arranged on a draggable canvas. The interface supports line, bar, scatter, histogram, heatmap, and other chart types, with saved conversations and shareable views.
The application combines streaming model responses and tool calls with persistent chat history, database connections, and dashboard state.