I Built an AI That Cuts Your Podcast Into Shorts. But I Didn’t Want It to Edit Your Content.

An engineer built AI Clip Cutter, an AI tool that turns long-form podcasts into short vertical clips by evaluating candidate moments for hook strength, information density, self-containedness, and emotional tone. The tool explains its choices and keeps trim points editable, ensuring the creator remains in control of the final edit.

The story behind AI Clip Cutter — and why we’re building AI editing around one simple idea: the creator should stay in control. Press enter or click to view image in full size There is an uncomfortable truth about short-form content: Most creators don’t have a content problem. They have a time problem. You can spend an hour recording a podcast. Two hours researching. Three hours having a conversation worth sharing. And then discover that turning that one long video into five genuinely good Shorts is going to take another afternoon. Finding the moments. Cutting them. Reframing them. Writing captions. Making sure the captions don’t start halfway through a sentence. Checking whether the clip actually makes sense without the 30 seconds of conversation before it. Then doing it again. And again. And again. That was the problem that led us to build AI Clip Cutter. AI Clip Cutter But there was another question behind it: What if AI didn’t need to replace the editor? What if it could simply do the boring part incredibly well? The idea was simple Take a long-form video. Find the moments worth sharing. Turn them into short vertical clips. Add captions. Let the creator decide what gets published. Sounds obvious. But once we started building it, we realized that “find the best clips” is not actually a simple problem. A 60-minute podcast can contain dozens of technically valid 30-second sections. But most of them aren’t good Shorts. Some start in the middle of an argument. Some need 45 seconds of context. Some contain interesting information but have no hook. Some are emotional but say nothing. And some sound incredible when you’re sitting inside the full conversation — but completely confusing when they’re watched alone. So we needed the AI to understand something more important than: “What was said?” It needed to understand: “Would someone want to watch this?” We don’t ask AI to pick “interesting” moments This was one of our biggest product decisions. Instead of asking the model to vaguely find “interesting clips,” AI Clip Cutter evaluates candidate moments across four dimensions: Hook strength. Does the clip make you want to keep watching? Information density. Does something meaningful actually happen in those seconds? Self-containedness. Can someone understand the clip without watching the entire podcast? Emotional tone. Does the moment contain surprise, tension, humor, conviction, curiosity, or another strong emotional signal? The system generates candidate windows between roughly 18 and 75 seconds, respecting sentence boundaries and natural pauses. It then ranks them and selects a non-overlapping set of clips. But here’s the part I’m particularly proud of. We don’t just show you: Clip 1–9.2/10 We show you why the AI picked it. For example: “She reframes the whole argument in one sentence here — a self-contained hook that needs no setup.” That’s important. Because AI shouldn’t just make decisions. It should be able to explain them. AI gets the first draft. You get the final word. There is a dangerous direction AI editing products can take. Upload your video. Click a button. The AI decides everything. The crop. The cuts. The captions. And eventually, you’re not really editing anymore. You’re approving whatever the algorithm decided. We wanted something different. Every selected clip in AI Clip Cutter comes with its scores and reasoning. And the trim points are editable. If the AI picked 00:17:32–00:18:08 and you think the real clip starts two seconds earlier? Change it. The captions are rebuilt around the new timing. You remain the editor. The AI is your ridiculously fast assistant. We also made a slightly controversial decision We don’t add b-roll. We don’t add random zooms. We don’t add transitions. We don’t try to make your podcast look like a MrBeast video. And we don’t pretend that every creator wants the same editing style. The output is deliberately simple: Your content. A clean cut. A vertical frame. Captions. That’s it. Because sometimes the most powerful thing you can do to a good piece of content is not get in its way. Captions are more important than people think There is a difference between having captions and having good captions. A sentence-level subtitle appearing every few seconds is technically a caption. But short-form video moves much faster than that. So we built the caption system around word-level timestamps. The captions know when each word is spoken. That allows us to highlight the active word, synchronize the visual rhythm with speech, and create different caption styles without guessing where the words should appear. There are currently 18 caption presets, ranging from minimal and cinematic styles to more energetic formats with active-word highlighting. And there’s a small technical detail here that matters: When you change a clip’s trim points, we don’t simply slap the old captions onto the new video. We regenerate the caption groups against the new timeline. It sounds like a tiny feature. It isn’t. It’s the kind of detail that determines whether an automated editing product feels genuinely useful or feels like a demo. We built for creators in India first There’s another part of this product that I’m especially excited about. AI creator tools are becoming global. But the pricing infrastructure hasn’t always caught up. A lot of creator software is priced in dollars, even when the creator is earning and spending in rupees. That creates unnecessary friction. So AI Clip Cutter is priced in INR, with payments through Razorpay. Our current paid plans start at ₹399, with higher tiers at ₹899 and ₹1,999. Credits don’t expire, and the free signup includes 10 credits with no card required. This isn’t just a pricing decision. Subscribe to the Medium newsletter It’s a product philosophy. If you’re a creator who publishes three videos this month and none next month, your unused credits shouldn’t disappear just because a calendar flipped. The economics are intentionally simple We don’t want you calculating complicated upload quotas. There are basically two things you pay for: Analysis: 1 credit per minute of source video. Export: 1 credit per finished clip. That’s it. So if you analyze a 30-minute podcast and export four clips: 30 credits for analysis. 4 credits for exports. 34 credits total. The model is transparent before you press the button. No surprise invoice. No mysterious “AI processing units.” No pretending that a 30-minute video and a 2-minute video cost the same to process. And yes, there are things our AI can’t do This is probably the most important thing I can say about building AI products: Don’t hide the limitations. AI Clip Cutter currently doesn’t have face tracking. It doesn’t automatically add b-roll. It doesn’t add transitions or zooms. It doesn’t have speaker diarisation. Vertical reframing currently tops out at 1080×1920. And although captions can handle many languages, our clip scoring and hook-writing are currently strongest for English-language content. We’re not embarrassed by that. In fact, I’d rather tell you exactly what the product doesn’t do than promise an imaginary future version of it. Because the goal isn’t to build the longest feature list. The goal is to build something people actually use. The bigger opportunity isn’t “AI video editing” I think that phrase is too broad. The interesting opportunity is something much more specific: Turning existing knowledge into more opportunities to be discovered. Think about how much valuable content already exists. Podcasts. Interviews. Founder conversations. Courses. Webinars. Livestreams. Customer calls. Conference talks. Educational videos. There are millions of hours of long-form content being created. Most of it will never be turned into short-form content. Not because it isn’t valuable. Because somebody has to sit there and find the moments. That’s the bottleneck. And that is exactly where AI is useful. One podcast can become an entire content engine Imagine recording one 60-minute conversation. Traditionally, that becomes: 1 long video. With a good clipping workflow, it can become: 1 YouTube episode plus 5–10 short videos plus multiple hooks plus captioned social posts plus ideas for future episodes The original conversation doesn’t change. You’re simply extracting more value from something you already created. That’s the part of AI that excites me. Not replacing creativity. Compounding it. We are not trying to make creators obsolete This distinction matters. I don’t believe the future of creative work is: Humans create → AI replaces them. I think it’s closer to: Humans create → AI removes the repetitive work → humans make better decisions. The creator should decide what they stand for. The creator should decide what feels authentic. The creator should decide what gets published. The AI should handle the parts that humans shouldn’t have to spend hours doing. Finding timestamps is not creativity. Generating 30 caption variations is not creativity. Reframing a video from 16:9 to 9:16 is not creativity. Searching through a one-hour podcast for the strongest 40-second moments is useful work — but it’s also incredibly repetitive. That’s where we want AI. We’re building the boring parts of content creation That might sound like a strange startup pitch. But I think it’s a powerful one. The future of creator software isn’t necessarily about adding more effects.