The script is the video: why we made editing script-first

Most video editors start from the timeline. You drag clips into place, then fit words around them. Launch videos work the other way round. The message comes first, and the visuals exist to make each sentence land.

So in Twelve Studio, the script is the source of truth.

One document, every scene

Open the Script view in the editor and you see the whole video as a single document. Each scene is a block that starts with a heading:

## Opening
Meet TwelveAI. AI infrastructure for conversational banking.

## Built for fintech first
Transfers, bills and balances through natural language.

Edit a sentence and, about a second after you stop typing, the narration in that scene updates, the captions re-time, and the timeline grows or shrinks to match. Add a new block and a new scene appears. Delete a block and its scene goes.

Visuals stay where you put them

The hard part of script-first editing is not losing work. If you spent ten minutes choosing a screenshot for a scene, a rewrite should not throw it away.

Each block is matched back to its scene by title first, then by order. Rename a heading and the scene keeps its screenshot, motion and styling. Only the words change.

Let the assistant write the first draft

Write with AI takes a short brief and returns the full script, already laid out: a scene type for each beat, on-screen text, and your uploaded screenshots placed by matching file names to what each scene says. Lay out scenes does the same for a script you wrote yourself, without changing a word of it.

The assistant only writes. Voice, captions and rendering happen on our own pipeline, and nothing it produces is final until you say so.

Timing follows the words

While you write, scene lengths are estimated at about 2.6 words per second, so the timeline is close but not exact. When you render, every line is voiced and timed precisely, and the cut follows the real audio.

That is the whole idea: write the story once, and let the production follow it.