Edit a keynote take into two finished video formats

Andreas Fuchs used Claude Code to convert one keynote recording into a polished edit with captions, graphics, branding, music, and both widescreen and vertical exports.

Done with Claude Code· Advanced· Oct 2026

What they did
Andreas Fuchs put together a folder-based Claude Code setup for keynote post-production. They fed it one raw speaking take along with brand assets and a music track, then had it generate a transcript with word-level timing, remove the rough sections with ffmpeg, and create a separate web-page-style visual for each sentence. HyperFrames handled the final render, and the agent had to surface proof after every stage, like the transcript or a frame grab.
How it went
Starting from a single take, the agent output a complete video with motion graphics matched to every sentence, plus captions, branding, and music. It also produced a vertical Reel. The later prompts and steps are available only after a free signup.
Tips
Plan for about an hour of setup time. You need a Claude plan, and frequent use may run into Pro limits; a music licence may also be needed.

Try it yourself with Claude Code

Work on the raw video in [path]. Clean it into a finished edit by transcribing it with word timings, removing mistakes and pauses without chopping off any words, then adding captions, lightweight motion graphics for the main points, my branding from [logo/color/style files], and background music from [music file or folder]. Create both a [horizontal format] export and a [vertical format] export in [output folder]. Before you render anything final, show me the transcript and the proposed cut list, then wait for my approval before any step that could overwrite, delete, send, or publish files.

Fill in the [brackets]. Agents can make mistakes, so review anything before it is sent, paid or deleted.

Read the original on futureofmarketing.de ↗

Get 3 agent uses like this every morning

With the day's top AI news. Free, and one click to unsubscribe.