The Build Lab · Case study · Video
A talking-head clip goes in, a captioned, cut and branded reel comes out.
A yoga teacher films short tips on her phone but never has time to edit them into reels.
An editor that transcribes the clip, cuts the pauses and retakes, adds captions in her brand style, drops in on-screen cards at the right words and exports a vertical reel ready to post.
An evening in an editing app for every reel.
Coaches, teachers, salon owners and anyone who films quick tips on their phone but never gets round to editing them.
Reels that look edited without the evening in an editing app, so filming a tip turns into posting it the same day.
Demo video
Live demo
Everything below works. The business and data are made up; the real version is built around your business and your tools.
Transcribing IMG_4182.MOV
00:00
4.0s Hi, I'm Maya from Salt & Stone. um If your hips feel tight after a day at the desk, this one's for you.
3.5s RetakeSo pigeon pose is, no, hang on. Let me start that again. Pigeon pose is the best hip opener you can do at home. 2.5s Here are three tips.
uh One: keep your back foot pointed, not sickled. 6.0s Two: square your hips to the front of the mat. um, like Three: breathe into the stretch for five slow breaths.
5.5s Save this for after your next long day at the desk. 13.0s
the best hip opener
Exporting hip-openers-3-tips.mp4
0%Rendering captions and cards
Ready to post
I built this with Claude as a single page, laid out like the desktop editors people already know.
The real version takes a clip from your camera roll or a Google Drive folder, transcribes it word by word and marks the pauses, filler words and retakes. You check the cuts in the transcript, not by scrubbing through a timeline. Captions go on in your brand fonts and colours, and cards drop in on the words that matter, such as a pose name or a list of tips.
The finished reel exports at 1080 × 1920 with a caption drafted for Instagram and TikTok, ready for you to post.
Here it walks through one sample clip for a made-up yoga studio, so nothing is really rendering.
Questions
Yes, for talking-head reels AI can do most of the editing. It can transcribe your clip, cut long pauses, ums and retakes, add captions and place on-screen text at the right moment. What it cannot do is decide what is worth saying or whether a take feels like you, so you still choose what to film and approve the final cut. Off-the-shelf tools like CapCut, Descript and Instagram's Edits app already handle auto-captions and filler-word removal. A custom editor like this one adds your exact brand style every time.
It works from the transcript. The audio is transcribed word by word with timestamps, then the editor finds silences over a set length, filler words and repeated sentences where you started again. Those sections are cut from the video, and the remaining clips are joined so the speech flows. Captions come from the same transcript, so they line up with every word. Because it all runs off text, you can review the edit by reading the transcript and restoring anything it cut by mistake, which is much faster than scrubbing a timeline.
Start with a free app. CapCut and Instagram's Edits app both offer auto-captions and simple cutting, and Descript lets you edit video by editing the transcript. Set up one caption style in your brand fonts and colours, save it as a template and reuse it for every reel. That alone saves most of the time. A custom editor built with Claude makes sense once you post often and want every reel to match your brand without fiddling, and it usually needs a small bit of code rather than just the chat window.
Not on its own in the chat window. Claude cannot open your video and trim it the way an editing app does. What it can do is write and run the code that does the editing, for example in Claude Code, using a transcription model to get timings and a video tool like FFmpeg to cut, caption and export. That is how a custom editor like this is built. For most small businesses, a free app with auto-captions is the right first step, and a custom build comes later if the volume justifies it.
Yes, if the brand rules are set once and applied every time. That means your fonts, colours, caption position and style, logo placement and the look of any on-screen cards. Consistency is the main advantage over editing by hand, because every reel matches without you remembering the settings. The part to watch is caption accuracy: names, product terms and anything with an unusual spelling are where transcription slips. Read the captions once before posting, especially your own name and your business name.
Only with their permission, and ideally in writing. Video of a person is personal data under UK GDPR, so get clear consent before filming anyone else, and before posting it. Check where your editing tool stores uploads and whether it uses them to train its models. Some tools let you switch that off. For your own talking-head clips the main risk is posting something you did not mean to, so keep a human review before anything goes live and never set an editor to publish automatically.
Yes. My 1:1 AI implementation is done-with-you support over six weeks: we build your reel workflow together around the tools you already use, so you understand every step and can keep it running without me. It is not a done-for-you service, and it suits people who want to learn the skill rather than hand it off. If you would rather go it alone, the answers on this page are enough to get a first version working, and starting small is the right call either way.