Built · v0.1.0
Live
Semantic search for everything on your computer, running fully offline.
Find documents, PDFs, code, spreadsheets, slides, emails and images by describing them. Live indexes your files with a local embedding model and shows exactly where the match is, in a floating launcher.
- Platforms
- Windows, macOS, Linux
- Version
- 0.1.0
- Built with
- Python, PySide6, fastembed / ONNX, SQLite FTS5, Tesseract
What it does
- Hybrid search: meaning plus keywords, so names and IDs still match
- Shows the page, slide, sheet or line of each match with the passage highlighted
- Re-indexes as files are created, edited, moved or deleted
- Filters like kind:pdf, text from images, and a command-line tool
How it’s built
- bge-small embeddings through ONNX with no PyTorch; vectors and the keyword index in one SQLite file, merged with reciprocal-rank fusion
- Text extractors for PDF, Word, PowerPoint, Excel, OpenDocument, email and notebooks