Why We Built Montage in the Collaborate Quadrant

There's a simple framework we use internally to decide what AI should and shouldn't do in video production. Draw two axes: AI leverage (how much can AI actually help?) and human judgment (how much does the outcome depend on taste, context, and editorial intent?).

You get four quadrants. And where you choose to build defines everything about your product.

The automate quadrant is crowded

Most AI video tools live in the top-left: high AI leverage, low judgment. Transcription, auto-captions, format conversion, batch resizing. These are real problems and AI handles them well. But the output is commodity. If your tool generates 30 clips ranked by a generic "virality score," you've built an automation, not a collaborator.

The tell is in how users describe the experience: they use AI clip tools to get rough cuts, then redo most of them in Premiere. The AI did its job. The human still has all the work.

The bottom-right is sacred

High judgment, low AI leverage. This is where human craft lives — color grading, sound design, motion graphics, the instinct for when a cut should land half a beat early. AI can't help much here, and it shouldn't try. This is what your editor does after the editorial decision is made.

We protect this quadrant by design: Montage exports XML to Premiere, Final Cut, and DaVinci specifically so editors can do this work in their native tools.

The best AI tools don't replace judgment. They give judgment better inputs, faster.

We build in the collaborate quadrant

Top-right: high AI leverage and high judgment. This is the hardest place to build because you need both capabilities — AI that's actually good at surfacing signal from noise, and a UX that lets a human with taste make fast, precise editorial decisions.

Concretely, here's what "collaborate" means in Montage:

The AI does the high-leverage work (watching 45 minutes of footage, finding the 8 moments worth discussing). The human does the high-judgment work (deciding which moments serve the content goal, refining them to be client-ready). Neither waits on the other. Neither replaces the other.

Why this matters for agencies

In an agency, the "collaborate" quadrant maps directly to the content producer's role. They're the person who watches footage, understands what the client needs, and makes the editorial call. Today, they note timestamps on paper and hand them to an editor just to hear the clip — a round-trip that costs hours before anyone even approves anything.

Montage gives the producer a tool where they can create, iterate, and approve clip candidates in real time — then hand off only the approved clips. The editor receives XML and applies their craft: effects, music, color. No wasted polish on clips that get killed. No stepping on each other's workflow.

It's a clean separation of roles — one that only works because AI handles the leverage (discovery, scoring, rough assembly) while humans retain the judgment (editorial decisions, quality bar, client taste).

We didn't build an AI that replaces editors. We built an AI that makes content producers 15× faster at the part that was slowing everyone down.

The implication for where AI video goes next

The "automate" tools will keep getting faster and cheaper. That's fine — we use some of them ourselves. But the highest-value work in video will always live in the collaborate quadrant: where AI leverage meets human judgment. That's where the editorial decisions are. That's where the content quality is decided. And that's where Montage will keep building.

If you're a content producer, agency, or events team that has standards — and you're tired of AI tools that generate clips you wouldn't put your name on — we built this for you.