DigUp brings on-device semantic search to Mac files
DigUp is a free, open-source menu bar application for Apple Silicon Macs that searches selected files by their contents. Its GitHub project says it uses Google DeepMind’s EmbeddingGemma 2 to index images, PDFs, documents, audio, and video, with code search available as a separate option. A query can locate a relevant passage, image, or moment in a recording or video.
The project describes the search as on-device: indexed files stay on the Mac, with no account or telemetry. DigUp supports macOS 14 or later and downloads an 865 MB model on first launch. Users choose which folders to index; the README also details exclusions and format-specific limits. For example, long documents are indexed in stages, and video search samples frames and audio. The release demonstrates a practical local application of multimodal embeddings, while its scope and performance depend on the selected files and indexing behavior documented by the project.