I make YouTube videos, and the hardest part for me is not coming up with ideas. It is turning a pile of notes into something I can actually speak out loud without sounding like I am reading an essay.
I built a small workflow called /youtube for that. I feed it whatever I have: bullet points, a rough outline, source material, or a mix. It asks one question before writing — “Anything else I should know before writing?” — and then turns that context into a spoken script.
The script comes back broken into short lines, the way you would want it on a teleprompter. Each line is a breath-sized phrase, not a wall of text. It always opens with a hook and ends with a call to action. The middle sections change depending on the video: explainers, commentary, tutorials. It also adds light cues where they help — a pause, a bit of emphasis — but nothing about camera cuts or b-roll.
Length is inferred from how much material I gave it, usually capped around ten minutes of speaking. If I want a different tone, I can say so. Otherwise it stays neutral.
After the first draft, I revise in plain language. “Make the hook punchier.” “Slow down the middle.” “Cut this section.” It keeps rewriting the full script until I say I am ready to rehearse.
That is the whole point. I am not pulling transcripts from existing videos. I am not getting titles, descriptions, or upload metadata. I get a script I can read aloud while practicing, and I iterate until it feels natural to say.