Desktop Docs
desktopdocs.com- Published
- Category product
- Tags rust, desktop app, semantic search

Desktop Docs is a Mac application that enables local searching of photos and videos using natural language and CLIP embeddings. Initially built with Electron, the app faced significant performance issues, large file sizes, and stability problems.
After struggling with Electron's limitations, the co-founders decided to rewrite the application in Rust and Tauri. This rebuild drastically reduced the app size from 1GB to 172MB, improved indexing speed (a 38-minute video now indexes in ~3 minutes), and enhanced overall stability.
The new version leverages Rust for image/video processing and file I/O, while still using CLIP for embeddings and Redis for vector storage. The UI was also rebuilt for a cleaner experience. Despite the learning curve with Rust and Tauri, the improved performance for core indexing and search functionality validated the decision to rewrite the application.