AI documentary editing
at archive scale
Documentary is a search problem long before it becomes an edit. You have hundreds of hours and dozens of contributors, with the story buried somewhere in the transcripts. EditAssist indexes all of it, every word and every frame, so you can pull the thread, assemble the paper edit, and conform the finish without ever logging a tape by hand. The footage never leaves your machine.
The documentary workflow
Ingest the drives, with their terabytes of interviews, verité, archive and B-roll. The agent transcribes every contributor and indexes every frame, then clusters the rushes so the shape of the material is visible before you cut. Ask a question of the footage, say “every mention of the trial” or “the wides of the harbour”, and it answers across the whole project at once.
From the transcripts and the director's notes it builds the paper edit: selects per character, a string-out grouped by theme, an assembly of the strongest beats. It proposes cutaways from the B-roll for each segment and colour-matches across episodes so the series feels like one piece. Once the offline is locked, it conforms to the camera originals and validates every relink.
Prompts for documentary editors
“We shot 400 hours for this film. Find every mention of the river and build a thread.”
“Build a paper edit of the three-hour assembly: keep only the strongest narrative beats.”
“Transcribe all 60 interviews, identify the key contributors, and build a selects reel per person.”
“Build an assembly from the director's notes: match each beat to the best soundbite.”
“Cross-reference the shooting schedule with what we actually shot and flag missing coverage.”
“Find the B-roll that best illustrates each interview segment and suggest cutaway placements.”
“Colour-match the night exteriors across episodes three and four to the reference grade.”
“Conform the locked picture to the 6K camera originals and flag every relinked mismatch.”
Hundreds more in the prompt library.
What powers a documentary edit
Search across terabytes
Every interview transcribed and every frame indexed. Ask for 'every mention of the river' or 'the moment she breaks down' and the agent surfaces it across the whole archive in seconds, with no logging by hand.
Selects from director notes
Feed it the director's notes, a PDF script breakdown, or a theme. The agent pulls the matching takes, ranks them, and builds a selects reel per contributor or per beat.
Assembly from transcripts
A documentary lives in the paper edit. Build a string-out or assembly straight from the transcripts, grouped by theme rather than chronology, and let the agent suggest cutaways from the B-roll for each segment.
Colour match & conform
Match grades across episodes so the series cuts together, then conform the locked offline to camera originals via FCPXML, FCP7 XML or CMX3600 EDL, relink checks and all.
Sensitive footage stays on your hardware
Documentary footage is often confidential, covering sources, contributors and legally sensitive material. With EditAssist, transcription, search indexing, clustering and vision analysis all run locally on your own GPU. Nothing about the interviews is uploaded. The only data that leaves is the text of your conversation with the agent. It's free to download, local models are free, and a new account starts with £15 of credit and no card. For long-running cloud work, pay as you go, or choose £20/mo Starter or £100/mo Pro.
Getting started
- 01Download EditAssist for macOS or Windows and create an account. You get £15 credit and no card is needed.
- 02Ingest your drives, and it transcribes every interview and indexes every frame, overnight if needed.
- 03Ask questions of the footage and build a paper edit from transcripts and director notes.
- 04Colour-match across episodes, then conform the locked cut to camera originals.
More workflows: podcast editing, YouTube creators, social clips. Or see everything EditAssist connects to and browse the prompt library.