Show HN: AI search for every photo and every frame of video on macOS (github.com)

A new AI search tool for macOS has been announced, which allows users to conduct deep searches across every photo and video frame stored locally. Notably, this tool operates without needing cloud services; all processing occurs directly on the user's machine, ensuring privacy and avoiding account setups. Users can describe memories in natural language, and the software employs local vision models to locate corresponding images or video moments. The technology also includes Optical Character Recognition (OCR) for visible text and dialogue searches via the Whisper neural network, making it possible to find specific spoken lines in videos. This development is significant for the AI/ML community as it demonstrates an advanced application of local models, particularly in multimedia search capabilities. Unlike conventional search tools that rely heavily on metadata or file names, this application uses semantic understanding to accurately retrieve content, enhancing user experience and efficiency. Its infrastructure allows for dynamic content management with features like automatic folder watching and deduplication, and it supports multiple operating modes for different types of queries—images, video scenes, text, and dialogue. By processing data locally, it addresses privacy concerns while enabling powerful AI-driven searches, setting a precedent for future applications in digital media management.