You pick the folders to index, and the DigUp app reads the pictures, PDFs, documents, recordings, and videos inside them. It also reads code projects if you turn on code search. Instead of relying only on file names, you describe what you are looking for in your own words across more than 100 languages. For example, typing “zebra in a video” opens the video clip at the moment the zebra shows up, “where they talk about sleep” jumps to that minute of a podcast, “the clause about pets in the lease” shows the matching PDF page, and “a dog on the beach” finds the photo. Exact words such as invoice numbers, error codes, and names are matched in file names and document text too, making it a local AI file search tool that works alongside Spotlight. Everything runs on your computer with no account and no telemetry.
DigUp at a glance
| Item | Details |
|---|---|
| Made by | ARahim3 |
| License and price | Free and open source (MIT) |
| Systems | Mac (Apple Silicon, macOS 14 or later, 0.6.1); Windows (64-bit Windows 10 or 11) and Linux (64-bit Intel and AMD, Ubuntu 22.04, Debian 12, Mint 21 and newer) in beta (0.7.0-beta.1) |
| Model | Google DeepMind’s EmbeddingGemma 2 (740M parameters, Apache 2.0) |
| What it searches | Pictures, screenshots, PDFs, documents, audio, video, and optional code in the folders you choose |
| Shortcut | ⇧⌘Space on Mac; Ctrl+Shift+Space on Windows and Linux |
| Network use | Goes online only to download the 865 MB model once and check for updates once a day (can be turned off) |
How a search works
DigUp’s readers turn the files in your chosen folders into pictures, text, and sound, and pass them through EmbeddingGemma 2 on llama.cpp to build one SQLite file holding a 768-dimensional vector for every picture, page, passage, video frame, and 30 seconds of sound, plus a keyword index. When you search, your words go through the text part of the same model and are ranked by nearest vectors plus exact words.