An agentic YouTube pipeline built on GPT Image 2.5

Simon Høiberg @SimonHoibergOriginally posted 2026.09.16

Open original on X

Making YouTube videos fully agentic is such a sick experience now. DaVinci Resolve and Fusion handle the edit. Remotion and Blender handle motion and 3D. GPT Image 2.5 handles visual generation. Then agents coordinate research, scripts, thumbnails, and publishing around it. Man, you can make some cool videos with this setup. I still come up with the ideas and record the footage. But I can now produce the kind of videos that would have needed an entire production team a few years ago. My full YouTube production stack is below 👇