Built a local GraphRAG desktop tool on top of LightRAG + Ollama

[Semm6]

Hey everyone,

Me and two friends (all devs) were experimenting with GraphRAG setups locally and kept running into the same problem: most pipelines are a pile of Python scripts and take quite a bit of setup to get working properly.

So we started hacking on a small desktop wrapper around LightRAG that runs locally. The goal was mostly to make it easier to experiment without having to wire everything together manually.

Right now it basically:

runs locally through Ollama (no data leaves your machine)

lets you build and query a knowledge graph from your documents

is packaged as a desktop app instead of a Python pipeline

One idea we’re exploring is having pre-built knowledge graphs for specific domains (dev docs, research papers, legal frameworks, etc.) so people don’t always have to index everything themselves.

Curious about a couple things:

Would something like this actually be useful to you?

If you could download a ready-made graph, what data would you want in it? (React docs, a specific repo, tax codes, etc.)

If anyone wants to try it or give feedback, it's here:
https://retriqs.com/

For local models we’ve had decent results using qwen3:30b-a3b-instruct-2507-q4_K_M in Ollama for the extraction step. After the graph is built you can switch to other models for querying.

Happy to answer questions or share more details if anyone’s interested.