Making a video
This is the one to read first. It assumes nothing except that the app is installed and open.
The other pages are for people changing the pipeline; this one is for people using it.
The shape of it
You work inside a project. A project is a folder in your library that holds everything belonging to one piece of work: the footage, the scripts, the narration, the renders. Choosing a project once at the top of the window is why no panel asks you which project you mean.
Everything else follows one order:
Library → New video → Cut → Editor → Review
You will not always use every step. A screen recording that needs no titles goes straight from recording to the editor.
Start a project
The Library is the first thing you see. New project asks for a name, a client and a brand, and makes the folder.
Once you are in a project, the sidebar grows a Make a video group. Those panels are about the work; the Toolkit group below is about the brand and the tools.
Get some footage
Three ways, and the New video page is the chooser:
- Record a screen — drive a browser through a script and capture it, or capture a window while you work. The steps you list are what it does; the settings for microphone, cursor and viewport are in the right-hand column.
- Make from a script — write what the video says, and the browser performs it. Good when the "footage" is really a set of designed cards.
- From a test — you already have a Playwright trace, and the video is cut from it. Press Load the example to see a real one run against a public site before you point it at your own.
Footage you already have goes in by dropping it on the project page. The original stays where it is; a copy lands in the project.
Write what it says
The Scripts panel writes narration and outlines as markdown. Prose is what gets spoken; a fenced block is a browser instruction and is never read aloud.
A script saved to a project appears on the project page beside the footage, with its word count and roughly how long it takes to say. Its menu takes it onward — into Voice to be spoken, or into Make a video to become one.
Give it a voice
Voice turns a script into narration and a perfectly synced subtitle file. The voices run on your own machine, so nothing about an unreleased client product leaves it and there is nothing to pay for per word.
The timings are exact by construction: every line is synthesised and then measured, so the subtitles cannot drift out of step with the audio. Edit one line and only that line is re-made.
Put it in order
Cut is where footage becomes a video. The shelf at the top is what the project holds, shown as frames rather than filenames — a recording with no sound says so there, where you are choosing it, rather than in the editor when you wonder where the audio went.
Take clips off the shelf, put them in order, add titles over the top, and hand the result to the editor. A rough in and out here is optional, and nothing is re-encoded either way.
Shape it
Editor opens the real timeline: trim handles, waveforms, zooms, cursor treatment, captions. This is where the video is actually shaped. Cut hands it a first assembly; the editor is where you finish.
Send it out
Review publishes a cut for comment and collects what comes back. Storage is where the finished thing goes — an S3 or Cloudflare R2 bucket you can browse from inside the app, upload to by dropping files, and organise by dragging.
The Toolkit
You need these less often, and mostly once.
- Brand — the marks, the colours, the clay renders, and the wallpapers that sit behind a recording. Drop a client's logo here and it becomes available to every composition.
- Components — the parts a scene is built from — titles, lower thirds, callouts, statistics, browser chrome — each shown working, with the markup under it.
- Storage — the remotes, and what is in them.
- Console — everything the pipeline runs, as it runs. Nothing here shows you a spinner: you get the command and its output.
When something does not work
Look at Console first. Every long step streams there, including the exact command. The most common causes are ordinary:
- A tool is missing. The dots at the bottom of the sidebar are lit for what is installed. A dark dot is the answer.
- The recording has no sound. A capture made with the microphone and system audio off carries no audio track at all. The Cut shelf says "silent" on the clips this is true of.
- A remote will not list. Storage's Test authenticates for real, which is the cheapest way to tell a wrong key from a wrong endpoint.
Where things are on disk
Your library is a folder of ordinary files and you can open it in Finder at any time. Nothing here is stored in a database, and nothing is hidden from you:
<your library>/
<project-id>/
project.json what the project is
media/
Footage/ video
Audio/ narration and music
Stills/ images
scripts/ what it says
renders/ what came out
Brand/ logos and imagery you added