AI / ML
Clip Finder
An AI system that scans a long-form recording and ranks the most postable 20–60s moments by hook strength, pacing, and topical density — the inverse of human "I'll just clip the funny bits" review.
A clip finder is an AI system that analyses a long-form recording and ranks candidate short clips by likely short-form performance. Inputs typically include the audio waveform (energy / pause detection), the transcript (sentiment, hook detection, topical density), and platform-specific signals (retention curve weights from YouTube, hook patterns from TikTok). Outputs are usually a ranked list of 20–60s segments with a confidence score and a suggested hook. Quality clip finders weight retention curves heavily — the highest-watched moments in the source — rather than just looking for loud / animated segments. The Clipflow Clip Finder ranks by hook strength and supports prompt-filtered ("find moments about pricing") candidate selection.
Related terms
Hook (short-form video)
The opening 1–3 seconds (audio + on-screen text) of a short-form video, which determines whether feed viewers swipe past…
Retention (video)
The percentage of viewers who keep watching at each point in a video. Platform algorithms use the retention curve — espe…
Short-form video
Vertical or square video typically 15–90 seconds long, optimised for mobile feed consumption — TikTok, Instagram Reels, …