Home
Local AI inference and retrieval-augmented generation over your own files.
Point LocalLens at any folder — notes, a repo, a research dump — and it builds a private brain you can chat with. Every answer comes back with citations, and not a byte leaves your machine.
The whole thing is eight TypeScript files in src/, written with
Bun and the QVAC SDK.
Small enough to read in an afternoon, complete enough to ship.
Where to start
What it does
The product, the user flow, and what local AI buys you.
How it works
The retrieval and completion pipeline, drawn end to end.
Run it locally
Install, run the CLI, run the browser app, verify the build.
Build it from scratch
Module by module, from domain types out to the UI.
Why these files, in this shape
The responsibility split, the request trace, and what each file should not own.
Extend it
Add image and PDF parsing with QVAC OCR, voice questions with QVAC transcription, and richer diagnostics.
What you'll need
- Bun
>=1.3.0 - Node
>=22.17.0 - A machine that can run a 1.7B-parameter quantized model. There's a 600M fallback for slimmer hardware.
No OpenAI key. No Pinecone account. No GPU required. The whole pipeline runs locally through QVAC.
What this site is not
This is the manual for a sample app, not a full reference for QVAC or Bun. The goal is to show you how the pieces fit together. When you need the depth of either upstream, the links point you there.