Jumper is a machine learning-powered search engine designed for video footage, operating entirely on-device without requiring cloud uploads. It integrates directly into popular non-linear editing (NLE) software such as Adobe Premiere Pro, Final Cut Pro, DaVinci Resolve, and Avid Media Composer, or can be used as a standalone desktop application for Mac and Windows.
Core Functionality
- Visual Search: Users can search their footage using keywords to retrieve relevant clips, eliminating the need for manual tagging. Features like "Match Similar" help find visually analogous moments, while "Match Source Monitor" allows searching for clips that match a specific frame.
- Face Recognition: Jumper can automatically detect and index every person in footage, enabling instant searches for specific individuals across hours of material. This feature operates locally, ensuring privacy.
- Speech Search: It provides instant speech-to-text search across 111 languages, allowing editors to quickly locate specific words or phrases spoken in interviews or other audio.
- Script Editor: A new feature allows users to work with full transcripts, non-destructively edit them, remove silences or unwanted lines, and send selected segments directly to the timeline.
Jumper emphasizes speed and privacy, performing all analysis locally on the user's machine, which ensures fast performance and that media never leaves the device. It also supports agentic editing, where an AI assistant can search footage based on plain English descriptions.