Summary
This tutorial from Youri van Hofwegen walks through a complete five-step AI-powered workflow for producing YouTube Shorts, from idea and script generation through voiceover, image creation, and final video assembly — with the explicit goal of completing the entire process in under 15 minutes. The video is aimed at content creators looking to systematize short-form video production using AI tools.
The workflow begins with finding an existing viral short in any niche, extracting its transcript via youtubetotranscript.com, and feeding it to Claude with a prompt that instructs it to replicate the structure and tone for a different topic — in this case, fitness content. Claude then generates both the script and a set of image prompts optimized specifically for GPT Image 2, with the creator noting that prompts should be model-specific since different generators have different prompting preferences.
For voiceover, the tutorial uses ElevenLabs (accessed through the OpenArt platform) with the 11v3 model, filtering by accent, gender, and use case to select a natural-sounding voice. Images are generated in GPT Image 2 at 4K resolution with a 1:1 aspect ratio, then assembled into the final short using CapCut. The video is a practical demonstration of how Claude, ElevenLabs, and GPT Image 2 can be combined into a repeatable content production pipeline, with specific tool settings and prompt strategies shown at each step.
📺 Source: Youri van Hofwegen · Published July 09, 2026
🏷️ Format: Tutorial Demo
