Skip to content

Offline Text Explainer

Select any text, get an explanation — fully offline, on an 8 GB CPU-only laptop.

▸ local LLM · phi-3 mini
runs offline · 8 GB · CPU only
OFFLINE

The problem

Looking something up breaks focus, and cloud AI isn't always available or private. The explainer runs entirely on the machine you're reading on.

What I built

  • Tauri app: Rust backend, TypeScript front end
  • Single words resolve from a local SQLite dictionary (WordNet / Wiktionary)
  • Sentences and paragraphs go to Phi-3 Mini via llama.cpp, run as a Tauri sidecar
  • Planned: topic-specific “wordbases” for saved vocabulary

Outcome

  • Designed for 8 GB RAM, CPU-only hardware
  • Currently in development

Stack

  • Tauri
  • Rust
  • TypeScript
  • SQLite
  • llama.cpp
  • Phi-3