Your first video

Open your iOS project in your connected agent and describe the video. One sentence is enough to start.

What happens next

1. It asks you a few things. Which iPhone, whether there is music, and how to record. See What the agent asks.
2. It reads your feature. It traces the real screens and the order they appear in, straight from your source.
3. It rebuilds the screens. A clean, self-contained copy of your UI — the same colors, fonts, icons, and copy — so the recording is deterministic and has no loading spinners in it.
4. It shows you the screens before recording. Look at them. This is the single biggest lever on quality: anything wrong here is wrong in the video.
5. It records and composites. The feature is driven on a real iOS Simulator with its own animations, then the look is composited on top.

How long it takes

The first run on a project is the slow one — a cold build can take several minutes. Later runs reuse that build. Re-rendering the *look* of an existing recording is fast, because nothing has to be recorded again.

Where the video lands

Wherever you ask. Say the path in the prompt, or let the agent choose a sensible one and tell you.

Then make it better

Watch it, then say what is off in plain language. The agent keeps the work it has already done and re-renders — you are not starting over.

• “Hold longer on the confirmation screen.”
• “Add captions, benefit-led, under six words.”
• “Brand it from our app and make it 9:16.”
• “The zoom is too aggressive — make it gentler.”

Everything up to writing the file is free. If export is refused, the work is not lost — it stays on your Mac and exports the moment you subscribe.