AI-Assisted Narrative Assembly from Transcripts — Script to Rough Cut Inside Premiere Pro

I'm a documentary director currently working on a film with approximately 13 hours of interview footage. I've been testing a workflow that I think points to something important for the future of editing inside Premiere Pro.

Here's what I did:

I used Premiere Pro's Text-Based Editing to transcribe all 13 hours of material. The resulting transcript runs to tens of thousands of words across multiple sessions. I then exported that transcript, combined it with my written script, and fed both into an external AI tool.

The AI did two things: first, it helped structure the film into chapters and narrative arcs based on the script logic. Second, it matched specific transcript fragments to each section — identifying which moments from the footage correspond to which part of the story.

The result was a working assembly blueprint: a structured map of the film before a single cut was made on the timeline.

The workflow works. But it requires exporting from Premiere Pro, switching to an external tool, working through the material in chunks (because the files are too large for a single session), and then manually bringing everything back. It's slow, fragmented, and breaks the creative flow.

None of this should need to leave Premiere Pro.

What I'd like to see built natively:

— Script import + transcript matching: load a script, let AI identify which transcript segments correspond to each beat or scene
— Chapter and act structuring: let AI propose a narrative structure from the raw transcript before the director even touches the timeline
— Intent-based clip suggestions: describe what a scene needs emotionally or narratively, get relevant clip recommendations from the transcript
— Reference-based assembly: analyze the pacing and structure of a reference video and apply similar logic to your own material

Premiere Pro already has the transcription layer. The text is there. The missing piece is the ability to reason over it narratively — not just search it, but understand it structurally.

This would change editing from a process of finding and placing, to a process of directing and refining. For documentary work especially, that's a fundamental shift.

Thanks to the team for Text-Based Editing — it's what made this entire experiment possible in the first place.