qmd

You can search personal notes and meeting transcripts locally without a cloud connection by using a local-first hybrid retrieval engine that operates offline. This approach indexes your isolated datasets across multiple collections to deliver search results without any cloud dependencies.

Hybrid search for local collections works by blending BM25 keyword search, vector semantic search, and LLM reranking. This combination retrieves and ranks precise results from your project documentation by matching both exact terms and contextual meaning.

Yes, you can use MCP integration to query your local knowledge base. The system supports both a CLI interface and MCP integration, allowing you to access indexed notes, documents, and transcripts through your connected workflow.

No, you do not need an internet connection to use vector search on your local documents. The local-first search engine uses embedded models to perform vector semantic retrieval and LLM reranking entirely offline.

The best way to organize isolated datasets like notes and docs for local search is to group them into local collections with context metadata. This allows the hybrid retrieval engine to effectively index and search across your personal knowledge base.

You should not use a local hybrid search if your personal knowledge management requires collaborative multi-user access or centralized cloud synchronization. This approach is designed for isolated datasets on local machines rather than shared remote environments.