OpenClaw cuts YouTube Shorts from a transcript
Chen reports OpenClaw downloading his video, transcribing it and cutting a vertical short; a poor first result improved after his framing correction.
Original prompt
Shared by the source author. See source notes for context.
This is awful. Redo #1. No subtitles. My original video is already vertical—just crop the center.
Reach people exploring AI agents
See placements →What happened
Chen reports OpenClaw downloading his video, transcribing it and cutting a vertical short; a poor first result improved after his framing correction.
Reported by Wisely ChenHow the agent did it
- Send a link to the creator’s own video through Telegram.
- Download the recording and generate a timestamped transcript.
- Select candidate moments with distinct hooks.
Where a person stepped in
Chen inspects the first result and supplies a framing correction. The second version still contains subtitles despite his request, so instruction adherence remains imperfect.
Wisely Chen describes asking OpenClaw to edit a weekly-summary YouTube video while he was away from his computer. He used Telegram to give the agent a video link. The assistant reported downloading the 17-minute recording and producing a timestamped transcript, then selected three candidate segments with clear hooks.
The proposed outputs were vertical MP4 Shorts. OpenClaw started converting the first segment while asking Chen how to handle the others. The first attempt looked poor. Chen corrected the brief: his original was already vertical, so the agent should crop the centre and redo the segment. He also asked for no subtitles. He reports that the second attempt was nearly right, but still had subtitles, an important sign that visual improvement did not equal complete instruction adherence.
The usable detail is to make the transcript the selection layer and review one export before applying its template to the remaining clips. Chen reports about ten minutes of interaction, not a measured general editing benchmark. His broader setup uses a GCP VM and Mac mini, but the article does not identify the exact cutting tool. The case is helpful for creators repurposing their own footage who can inspect framing, text overlays and duration before publishing.
Prompt
Original prompt
Source: Blog · Wisely Chen
Frequently asked questions
How were the segments selected?
The assistant used a timestamped transcript and proposed three moments with a clear hook for short-form clips.
What did the first attempt get wrong?
Chen says it looked poor and clarified that the original video was already vertical and needed a centre crop.
Did the revision follow every instruction?
No. Chen reports a nearly correct second export that still included subtitles despite asking for none.
More details
Remaining steps
- Export a first vertical clip and inspect it.
- Correct the crop and review the revised export before continuing.
About this account
OpenClaw downloads, transcribes, selects segments and starts conversion without an additional prompt for each production action, then adapts after user feedback.
Firsthand account by Wisely Chen; original publication date verified against Primary page publication date. Outcomes are the author’s reports, not independently reproduced. Original prompt excerpt transcribed from the primary account; excerpt only, not the complete prompt. Preserve the source-specific context when adapting it.
Credit: Wisely Chen
Source published 4 Feb 2026
Source checked 14 Oct 2026
Tools and topics
Telegram · YouTube · Timestamped transcript · MP4 export
Video editing · Shorts · Visual review
Editorial completeness: 100/100. Measures documented fields, not agent performance or verified outcomes.