Reor is a desktop note-taking application designed for individuals who manage large, interconnected bodies of information. It leverages local AI models to automatically link related notes, answer questions based on a user's corpus, and provide semantic search capabilities, all while prioritizing local data storage and privacy.

How It Works

Reor utilizes a RAG (Retrieval-Augmented Generation) architecture, treating both the LLM and the human user as "generators." Notes are chunked and embedded into a local LanceDB vector database, enabling automatic connection of related notes via vector similarity. In Q&A mode, the LLM retrieves relevant context from this corpus to answer user queries. The editor mode offers a sidebar displaying semantically related notes, augmenting the user's current thoughts. This approach aims to provide AI-powered knowledge management without relying on cloud services.

Quick Start & Requirements

• Install: Download from reorproject.org or releases. Mac, Linux, and Windows are supported.
• Local Models: Integrates with Ollama for downloading and running local LLMs. Can also connect to OpenAI-compatible APIs.
• Note Import: Manually populate a chosen directory with markdown files. Frontmatter parsing may be inconsistent.

Highlighted Details

• Local-first AI for personal knowledge management.
• Obsidian-like markdown editor.
• Automatic note linking via vector similarity.
• RAG-based Q&A on user's notes.

Maintenance & Community

• Active development with a focus on shipping features quickly.
• Discord server available for feedback and community interaction.

Licensing & Compatibility

• AGPL-3.0 license. This is a strong copyleft license, requiring derivative works to also be open-sourced under AGPL-3.0.

Limitations & Caveats

• Importing notes from other applications requires manual file copying, and frontmatter parsing may be unreliable. Integrations with other apps are planned but not yet implemented.