To get started with Reor, users can download it from reorproject.org or releases, and install it like a normal app. It interacts directly with Llama.cpp libraries, eliminating the need to download Ollama. However, users will need to manually download their preferred model. The app works within a single directory in the filesystem, which users choose upon first boot. To import notes from another app, users must manually populate the directory with markdown files. The app is licensed under GPL-3.0.
Key takeaways:
- Reor is an AI-powered desktop note-taking app that automatically links related ideas, answers questions on your notes, and provides semantic search.
- The app runs models locally by default, standing on the shoulders of Llama.cpp and Transformers.js to enable both LLMs and embedding models to run locally.
- Reor is described as a 'self-organizing' app because it chunks and embeds every note into an internal vector database, connects related notes automatically via vector similarity, and allows for semantic searching.
- Reor can be downloaded from reorproject.org or releases and supports Mac, Linux, and Windows. It interacts directly with Llama.cpp libraries, but users need to download their model of choice manually.