Reor

Reor is a local AI knowledge management application that stores, links, and searches personal notes using large language models and vector embeddings entirely on the user's device. It functions as a private AI note assistant, keeping all data and processing local for full privacy without relying on external cloud services.

The application integrates with Ollama to manage the lifecycle of local LLMs and embedding models, handling downloads, updates, and execution. Notes are imported from markdown files, preserving existing file structure, and are automatically linked through vector-similarity comparisons that surface semantic connections without manual tagging. Users can search their note corpus by meaning rather than exact keywords, and ask natural language questions that retrieve relevant passages and generate answers using the local LLM.

Reor also provides an OpenAI-compatible API bridge, allowing users to optionally connect to external AI services as an alternative to local models. The embedding pipeline runs entirely on the machine, computing and storing vector representations locally to ensure data never leaves the device.

Features

AI Knowledge Management - Provides a complete private AI knowledge management system that stores, links, and searches personal notes using local LLMs.

Local AI Model Runtimes - Downloads and manages LLMs and embedding models through Ollama, keeping all processing on the user's machine.

Model Lifecycle Managers - Manages the full lifecycle of local LLM and embedding models through Ollama for private AI processing.

Ollama Model Runners - Downloads, updates, and executes LLMs and embedding models through the Ollama runtime for local AI processing.

Ollama-Powered Note Tools - Integrates with Ollama to run local LLMs and embedding models for AI-assisted note features like semantic search and question answering.

Retrieval-Augmented Generation - Retrieves relevant note passages via vector search and feeds them to a local LLM for grounded answer generation.

Local Embedding Pipelines - Computes and stores vector embeddings entirely on the user's machine using locally-run models for full privacy.

Local Model Query APIs - Answers questions by sending a natural-language query to a locally-run large language model and returning the generated response.

Contextual Note Surfacing - Automatically connects notes by comparing their vector embeddings, surfacing related content as you write.

Question Answering - Retrieves relevant passages from the note corpus and feeds them to a local LLM to answer user queries.

Local-First Knowledge Bases - Operates as a local-first knowledge base that stores and processes all notes and AI models entirely on the user's device.

Automatic Note Discovery - Discovers semantic connections between notes as they are created or edited and surfaces those relationships automatically.

Embedding-Based Note Searches - Searches notes by meaning using local embedding models and vector similarity, enabling natural language queries over the corpus.

Note-Taking App AI Assistants - Answers questions and surfaces related content using locally-run AI models, keeping all data on-device as a private note assistant.

Note-Taking App Similarity Searchers - Links notes by comparing their vector embeddings via cosine similarity, surfacing related content without manual tagging.

Semantic Note Retrieval - Finds notes by meaning using vector similarity and local embedding models, without relying on external cloud services.

Note Importers - Reads markdown files from a designated directory, allowing migration of notes from other applications.

Markdown Note Managers - Imports, organizes, and edits markdown notes with automatic semantic linking and full-text search capabilities.

Markdown Note Importers - Imports existing notes from markdown files, preserving file structure and enabling migration from other applications.

reorprojectreorArchived

Features

Star history