The AI-Assisted Finish
Twelve chapters of tools mean nothing until they're one workflow. So let's finish a real short film together — ingest to delivery — and see exactly where the machines carry the weight and where you do. Spoiler: the part that's still yours is the point.
Picture the moment every chapter in this course has been building toward: the shoot wrapped last night, and there's a drive on your desk with a short film trapped inside it. Twelve chapters ago, what stood between that drive and a finished film was months of grind. Now let's walk what actually happens — the whole course compressed into one honest workflow, with every tool in its place and one section of the calendar fenced off with barbed wire.
Day one is machine day. Ingest the footage with a real folder structure, back it up twice before you so much as scrub a clip, and set transcription running on everything with spoken words. By tonight, your entire shoot is searchable text — the foundation that captions, text-based editing, and cutdowns will all draw from later. None of this needs your creativity. All of it needs your discipline. Do it while the coffee's still working.
The workflow, start to finish
From there, the whole finish runs in a sequence you now know chapter by chapter:
- 1 · Ingest, transcribe, back up (day one). Organized media, two backups, transcripts running overnight. Boring, foundational, non-negotiable.
- 2 · Selects and AI assembly (days two–three). Let the tools flag usable takes and build the first assembly from your script or outline. It will be wrong in interesting ways — that's its job. You now have something to react to instead of a blank timeline.
- 3 · The human cut (most of the calendar). Rough cut to fine cut to locked picture. Rhythm, performance, what the film is actually about. Protect this time ferociously — more below.
- 4 · Audio cleanup pass. Dialogue isolation on the scenes that need it, de-reverb, level match, stems prepped — every processed clip A/B'd against the original.
- 5 · AI color balance, then the grade. Machine normalizes exposure and matches shots; you build the look and the contrast arc on top. First pass, not final grade.
- 6 · Cleanup shots. The boom dip, the reflection, the logo legal flagged — masked, tracked, filled, and reviewed at full speed and resolution.
- 7 · Captions, reframes, exports, checklist. Captions generated and human-proofed, verticals reviewed shot by shot, presets run, every deliverable ticked off a written list.
Look at the calendar that workflow produces, because it's the quiet headline of this course: steps one, two, and four through seven — the parts that used to devour indie post — now take days. Step three still takes weeks. AI compressed everything except the cut. And the trap I've watched real filmmakers fall into is spending the saved time everywhere except there — polishing deliverables, re-running the grade, fiddling with cleanup — while the actual film gets the same rushed attention it always did. The entire point of the compression is to hand those weeks to the cut. Fence off that time. Nothing about the machine's speed obligates you to hurry the one part that was never the machine's to do.
Where this leaves you
Back in the first module I gave you this course's thesis: AI in post kills the drudge work around the cut, and taste stays human. Twelve chapters later you've seen it hold everywhere — in transcription and selects, in the balance pass under the grade, in the isolation under the mix, in the fill under the frame. Every tool, without exception, turned out to be the same shape: a machine handling volume so a human can handle judgment. Nothing we covered decides what your film means. Everything we covered gets you to that decision faster and less exhausted.
So here's your graduation assignment, and I mean it literally: take the shortest project you have — a scene, a spec, three minutes of anything — and run this exact workflow on it start to finish, this month. The tools will change names by next year; the workflow's shape will not. And the filmmakers who thrive in what's coming won't be the ones with the best subscriptions. They'll be the ones who used all this recovered time to develop the one asset no model can generate: a point of view. Go finish something. That was always the assignment.
The last short I finished this way, I kept a diary out of curiosity. Machine work — ingest, transcripts, assembly, cleanup, captions, exports — came to about six days scattered across the schedule. The cut took five weeks, same as it ever would have. But those five weeks were different: no dread, no 2 a.m. logging sessions, no energy burned on anything but the film itself. I wasn't a faster editor. I was a fresher one. That's the real gift buried in this whole course — the machines don't just save time, they save the version of you that shows up to the timeline.
That's the course. The tools are on your machine, the workflow is on this page, and the drive on your desk isn't going to finish itself. I'll see you in the next course — and better yet, I'll see your film.
Recut is this course in tool form — an AI edit assistant that pulls selects from your footage, builds a first assembly, and preps cutdowns, ready to polish in your own editor.