Hey Product Hunt! 👋 I'm Harshad, maker of LoomFlow.

I’ve spent years building data systems and workflow automation, and ran into the exact same frustration over and over: you either wait around on single-core Python scripts and brittle glue code, or pay absurd enterprise licensing fees for legacy visual tools.

I wanted something different: One open-source workspace to blend data, automate pipelines, AI native. Run fast, local-first workflows right out of the box. Ship complete data workflows the same day you clone the repo.

So, I built LoomFlow.

It gives you the intuitive, drag-and-drop power of modern canvas tools backed by a high-performance, multi-threaded engine—benchmarking up to 57x faster than standard data pipelines.

What makes it tick:

⚡ Blistering Speed & Local-First: High-performance execution engine with sub-millisecond transforms and smart disk caching (freeze node outputs to Parquet on the fly).
🧠 AI-Native, Not AI-Bolted: Seamlessly drop LLM nodes, multimodal reasoning, and embeddings directly into your DAG without writing custom glue code.
🧩 Extensible by Design: Zero proprietary lock-in. Drop in your own custom nodes, connect new models or tools with simple Python handlers, and export full pipelines to declarative YAML for autonomous agents.
💻 Interactive Visual Canvas: Smooth node-based DAG workflow built to take you from raw data sources to live pipelines in minutes.

We are 100% open-source. Clone it, run it locally, break things, and tell me what you want to see next.

Check out the repo, drop a star if you dig the vision, and let me know in the comments: what’s the most painful part of your current data pipeline setup? 🚀