datagallery-lab/datafoundry

DataFoundry is an open-source AI workbench for data analysis, unifying data sources, knowledge, tools, and agent runtime into a governed workspace for interactive analytics.

What it solves

When using AI to query enterprise databases, organizations face risks regarding business definition accuracy, data mutation, credential leakage, and lack of verifiability. DataFoundry addresses these by moving beyond simple SQL generation to a system that understands business semantics and operates within strict safety boundaries.

How it works

The system follows a structured workflow: it defines a task using natural language, aligns the request with enterprise business definitions (semantics), executes queries through a secure, read-only Data Gateway, materializes results into assets like tables or charts, and maintains a full audit trail for replayability.

Who it’s for

It is designed for enterprise teams needing a secure, controllable, and verifiable way to perform complex, multi-step data analysis using LLMs without risking production data integrity.

Highlights

  • 28 out-of-the-box data sources including PostgreSQL, Snowflake, BigQuery, and MongoDB
  • Enterprise semantics to map business terms like "GMV" to specific database fields
  • Safety-first execution featuring read-only boundaries, field masking, and row limits
  • Fully auditable with persistent, replayable SQL and tool-call logs
  • Multi-interface support via a Web workbench, a Terminal UI (TUI), and a REST API