DigUp is a free, open-source Mac app that searches user-selected files by what they contain. Its documentation says natural-language queries can find photos, screenshots, PDF pages, passages in documents, moments in audio or video, and—when enabled—matching code.
How the search works
The app runs Google DeepMind’s EmbeddingGemma 2 locally through llama.cpp and Apple’s Metal framework. Google describes EmbeddingGemma 2 as a 740-million-parameter model that maps text, code, images, video and audio into one shared embedding space. DigUp uses those representations to compare a query with indexed file contents, while also boosting exact-word matches for details such as names, invoice numbers and error codes.
The project’s documentation draws a narrower distinction from Spotlight than the original post’s “much better” verdict: Spotlight searches names, text and metadata, while DigUp is designed to also match what a file shows or says and jump to the relevant page, moment or line. That comparison comes from the developer rather than an independent benchmark.
Local processing and availability
The repository says indexed data stays on the Mac; network access is limited to downloading the model and checking for updates, with the update check optional. DigUp requires an Apple Silicon Mac running macOS 14 or later, and its 865MB model is downloaded once. The app is released under the MIT license.
Comments