DigUp lets users find Mac files by describing them, not remembering their names

What happened

DigUp, a free, open-source Mac app, was introduced as a tool that allows users to search for files by describing them using natural language. It utilizes Google's EmbeddingGemma 2 model to perform local, multimodal semantic searches across various file types without sending data to the cloud.

Why it matters

For AI developers and technology professionals, DigUp demonstrates the practical application of multimodal semantic search models in improving user experience and data retrieval on personal devices. It highlights the importance of on-device processing for privacy and efficiency.

• DigUp offers a practical solution for users struggling with file organization on Macs
• The use of EmbeddingGemma 2 enables efficient and private file searches
• Combining semantic and exact-word matching improves search accuracy

The bigger picture

DigUp's approach to local, multimodal semantic search could influence future developments in personal data management and AI-driven file retrieval systems, emphasizing privacy and on-device processing.

DigUp is a significant advancement in AI-driven file search, providing users with a powerful tool to locate files through natural language descriptions while maintaining data privacy.

Technical details

• EmbeddingGemma 2: A 740-million-parameter multimodal embedding model that maps different media types into a shared space for semantic matching
• Local processing: All searches are conducted on the user's device without cloud storage
• Multimodal search: Capable of searching across text, images, audio, video, and code
• Semantic and exact-word matching: Combines both approaches to balance fuzzy descriptions with precise queries

FAQ