Plan YouTube clips from a session transcript

A webhook starts it from a session record in Airtable. It reads how many clips you want from the number_youtube_clips field and creates that many placeholder rows in your YouTube clips table, marked AI Writing.

An OpenAI model reads the session transcript and picks that many separate sections worth clipping, each 5 to 10 minutes long, with no overlap. For each it writes the clip details, a title, a description, a first comment call to action and thumbnail text. A code step pairs each clip with a placeholder row and fills it in, set to To Do. If the model fails, the rows are marked Error. The thumbnail text feeds the YouTube clip thumbnails workflow.