HelixDB/helix-foundry

Connect your company's data, let an agent draft an ontology over it, and load it into HelixDB as one graph you can explore and ask questions of. Hosted at foundry.helix-db.com, or run it on your own machine.

What it solves

Helix Foundry is a local-first data workspace that allows companies to connect disparate data sources—such as databases, Stripe, PostHog, and APIs—into a single, connected ontology. It eliminates the need for external hosting or accounts, allowing users to explore and query their company data privately on their own machine.

How it works

The system uses HelixDB and DuckDB to ingest and sync data from various connectors (SQL databases, SaaS tools, and files). It automatically suggests a connected ontology that matches records across different sources. Users can interact with their data using a natural language interface powered by local AI (via Ollama) or external LLMs (Claude, OpenAI), which generates SQL to be executed in an isolated DuckDB executor for security.

Who it’s for

It is designed for data analysts and company operators who need to synthesize information across multiple business tools without compromising data privacy or relying on a cloud-based data warehouse.

Highlights

  • Local-first architecture: Runs entirely on the user's computer with local AI by default.
  • Unified Ontology: Automatically connects records from different sources into a single graph of objects and relationships.
  • Diverse Connectors: Built-in support for PostgreSQL, MySQL, Stripe, WorkOS, PostHog, and various file formats (CSV, JSON, Parquet).
  • Secure Execution: Generated SQL runs in a sandboxed DuckDB executor with no internet access.
  • Versioned Data: Uses immutable Parquet snapshots to maintain history and handle schema drift.

Related

  • Project
    julien040/anyquery

    Anyquery is an open‑source SQLite‑based engine that lets you query files, databases, and SaaS apps (e.g., Notion, Apple Notes) with SQL. It uses a plugin system, can run as a MySQL‑compatible server, and provides a Model Context Protocol (MCP) so LLMs like ChatGPT can fetch data from your sources. Installable via Homebrew, APT, Scoop, etc., it’s licensed under AGPL‑v3 (core) with MIT‑licensed RPC for plugins.

  • Project
    DataArcTech/DataArc-SynData-Toolkit

    DataArc SynData Toolkit is an open‑source Python framework that lets you generate synthetic training data (text or image‑text) for LLMs via a single YAML config, then fine‑tune models (SFT/GRPO) and evaluate them with DeepEval—all accessible through a CLI or a FastAPI + React GUI.

  • Project
    ModernRelay/omnigraph

    Omnigraph is an open‑source graph database built on the Lance columnar format that runs on any S3‑compatible object store. It offers typed schemas, multimodal retrieval (graph + vector + full‑text), Git‑style branching, and Cedar‑based security policies, all managed via Terraform‑like `cluster` files. Designed for long‑horizon AI agents, it lets multiple agents share a versioned world model and retrieve context in a single query.

  • Project
    awslabs/open-data-registry

    A centralized repository that helps users discover and share publicly available datasets hosted on AWS, streamlining data access for cloud-based analysis.