Summary
This full tutorial from the Nate Herk AI Automation channel walks through using GPT Astra — OpenAI’s agentic model — combined with the Codex desktop app and a motion-graphics plugin called Hyperframes to automate professional-quality YouTube and short-form video editing from start to finish.
The workflow is structured around four stages: transcription (using either local Whisper or ElevenLabs at roughly $5/month), silence and mistake cutting, beat planning where Astra reads the transcript and maps scenes to visual beats, and a verification loop where the agent watches its own output, takes screenshots, and iterates until the result meets spec. The instructor demos how a 14-minute raw recording was automatically cut to 9 minutes with dynamic scene transitions, rounded face-cam crops, animated title cards, and 3D motion graphics — all driven by natural-language prompts rather than manual timeline editing.
The video also covers short-form content generation, showing reels about Claude and Nvidia built entirely within the same pipeline. Setup instructions include how to get free API access to over 80 models through Nvidia’s developer program, and how to install Hyperframes inside the Codex environment. The tutorial is aimed at creators who want to reduce editing time without learning traditional video software.
📺 Source: Nate Herk | AI Automation · Published September 08, 2026
🏷️ Format: Tutorial Demo
