--- title: "AI clip suggestions: find the shorts inside your long videos" canonical: "https://blitzreels.com/changelog/long-videos-short-stories" --- # AI clip suggestions: find the shorts inside your long videos Milestone: 2025-10-29 Article published: 2026-09-26 Status: Product update BlitzReels reads the transcript of a long recording and suggests short clips, each with a title and exact time range, so you review real moments first. ![A frame from a 3:13:57 podcast episode next to three vertical shorts suggested from it.](https://blitzreels.com/images/changelog/long-videos-short-stories.jpg) ## Finding the clip takes longer than cutting it Cutting a 45-second clip is quick. Finding which 45 seconds of a 90-minute podcast deserve it is the slow part: you scrub back and forth and lose your place more than once. On October 29, 2025, we added AI short suggestions to the media library. You open a transcribed video, click Generate, and BlitzReels proposes candidate clips based on what was actually said, each with its own start and end time. ## What a suggestion contains The AI reads the full transcript and aims for ten distinct segments of roughly 20 to 60 seconds, letting one run longer when the thought needs it. Each suggestion has: - a title written to open the clip as on-screen text, - a hook and a short summary of what happens in that stretch, - a call to action when one fits, - exact start and end times inside the recording. In the media library you see the title and a preview of the segment that you can play right there. Two rules shape the batch: no two suggestions reuse the same stretch of transcript, and when you ask for more, the ones you saved are passed along so the new ideas don’t repeat them. The title and hook also tell you why the AI picked a stretch, which makes it quicker to say yes or no without watching every candidate in full. ## How it worked in October 2025 1. Upload or import a long video and let it finish transcribing. 2. Open it in the media library and click Generate in the Short-Form Suggestions panel. 3. Play the previews. Save the ones worth keeping and dismiss the rest. 4. Click Create Project on a suggestion to open that segment as its own project, ready to edit. 5. Click Generate New when you want another batch. Your saved suggestions stay where they are. The editorial call stays with you. A suggestion tells you where to look; you decide whether the idea is clear on its own, needs a few more seconds of context, or isn’t worth posting. ## What it grew into That panel was the first piece of the clipping flow. Today you click “Clip a long video”, paste a link or upload a recording, set a target between 1 and 30 clips or leave the count on Auto, and BlitzReels picks the moments, cuts them, adds captions and frames them vertically in one run. Later updates added [auto reframe](https://blitzreels.com/changelog/reframing-that-moves-with-the-video) for landscape footage and framing for [podcasts with two to four speakers](https://blitzreels.com/changelog/podcast-clips-that-show-every-speaker). The clips come out ready for [YouTube Shorts](https://blitzreels.com/use-cases/youtube-shorts), TikTok or Reels, and each one opens as a normal project in the editor if you want to change the cut. ## Which videos work best The October version worked from the transcript alone, so it suited spoken recordings best: [podcasts and interviews](https://blitzreels.com/use-cases/clips), webinars, lessons and livestreams. A video where very little is said gave it much less to work with. - Clipping is built for recordings longer than five minutes. Upload something shorter and BlitzReels offers to open it as a regular edit project instead. - You can end up with fewer clips than your target when moments overlap or don’t pass quality checks. - Generating clips uses AI credits from your plan. - The public API and the CLI can create clips from a workspace video or a YouTube link too. ## Questions ### How does BlitzReels choose which moments to suggest? It reads the transcript of your video and looks for complete ideas that can stand on their own as short clips. At launch it aimed for ten suggestions of roughly 20 to 60 seconds, each with a title and exact time range. ### Do I have to use the suggested clips as they are? No. You can save, dismiss or open any suggestion as a project, then change the cut, the captions and the framing in the editor. ### How long should my video be for AI clipping? Clipping is designed for recordings longer than five minutes, such as podcasts, interviews, webinars or lessons. When you upload a shorter file, BlitzReels suggests opening it as a regular edit project instead. ### How many clips can I get from one video? You can set a target between 1 and 30 clips or leave it on Auto, which bases the count on the video’s length. You may get fewer when moments overlap or don’t pass quality checks. ## Find the next clip in a long video Upload a recording or paste a link on the clip page and review the moments BlitzReels pulls out. [Try AI clipping](https://blitzreels.com/clipping) We wrote this story later, looking back at our product history. The date at the top marks the work described here; features and access may have changed since then.