Find any moment
in your footage

FrameSeek indexes your raw video folders using on-device AI. It transcribes audio, describes what's visible in each frame, and makes all of it searchable. You type what you're looking for in plain English, like "close-up of someone laughing" or "overhead drone shot of a city", and FrameSeek finds the matching 5-second clips across your indexed footage, ranked by relevance.

You need an Apple Silicon Mac (M1 or later) running macOS 15 Sequoia or later. On macOS 27 and later, FrameSeek uses Apple Intelligence for local processing and no model downloads needed. On older macOS versions, about 2 GB of AI models download on first use. After setup, no internet connection is needed.

Very little. When idle, FrameSeek uses ~50 MB of RAM. During a search it briefly rises to ~300 MB while the embedding model is loaded, then drops back to ~50 MB after a few minutes of inactivity. When indexing locally on macOS 27+, Apple Intelligence runs on the Neural Engine with minimal RAM impact. On older macOS, the on-device AI models use 4–5 GB during indexing, but that memory is fully released when done. If you use Gemini API mode, indexing stays lightweight since the heavy work happens in the cloud. FrameSeek actively unloads models you're not using so your RAM stays free for Final Cut, Resolve, and everything else.

No. All processing runs locally on your Mac. Your footage never leaves your machine. On macOS 27+, no downloads are needed at all (Apple Intelligence is built in). On older macOS, FrameSeek downloads AI models (~2 GB) once on first launch. After that, it works fully offline.

It depends on your footage and hardware. FrameSeek shows a time estimate before indexing starts. On macOS 27+, local processing is about 2.5x faster thanks to Apple Intelligence, roughly 1 hour of footage in about 20 minutes. On older macOS, expect about 50 minutes per hour of footage. If you have a large library and want the fastest processing, you can switch to Gemini API mode in Settings, which uses Google's cloud API and is faster still. You can switch between local and Gemini at any time. Learn more about Local vs Gemini.

Most standard video formats: .mp4, .mov, .m4v, .avi, .mxf, .mkv, .mts, .mpg, and more — including H.264, HEVC, and ProRes. Anything macOS can decode natively through AVFoundation. Camera-proprietary RAW formats like R3D and BRAW are not supported.

No. FrameSeek is read-only. It indexes your footage in place without creating proxies, copies, or re-encoded files. Your originals stay exactly where they are.

You can export search results as .fcpxml or .edl files. Each matched scene becomes a clip on the timeline that references your original video file, so there's no media duplication. Open the export in Final Cut Pro, DaVinci Resolve, Premiere Pro, or Avid and your clips are ready to edit.

Yes. You can download and use FrameSeek without a license key. The trial lets you index up to 25 videos and search across them. A license key is only required to index beyond that.

One payment of $59. No subscription. You get a perpetual license, 1 year of free updates, unlimited video folders and scenes, and you can activate it on up to 2 Macs.

The app keeps working. Your license is perpetual and never expires. You just won't receive new feature updates after the first year. If a new version comes out that you want, you can upgrade at a discounted price.

Yes. Contact support@frameseek.app within 14 days of purchase.

FrameSeek runs everything locally by default. No cloud, no API keys. But if you have a large library and don't want to wait for local processing, you can switch to Gemini API mode in Settings. This sends extracted frames to Google's API for visual descriptions. The estimated cost is shown before you start, and you can switch between local and Gemini at any time. Transcription and search always stay local. Read more about Local vs Gemini.