The words explain it. Your own footage shows it.

The input is one talking-head take, up to three minutes, filmed on a phone.

It transcribes the read, drafts the cut, places b-roll against your words and returns a 9:16 MP4 with captions burned in.

A batch is 100 credits. Rendering is 20 credits per output minute.

So a thirty second demo is 10 more, about 110 in total.

Basic is $39.99 for 2,500 credits, roughly 22 demos a month. Premium is $79.99 for 5,000, roughly 45.

Your product goes in the ad through your workspace library.

Film the shot where the thing visibly works, upload it once, and highlight the phrase where you describe it.

The clip lands over exactly those words, and the free b-roll search covers everything around it.

That makes this a demo where the explaining is spoken and the proof is yours.

There is no screen capture built in, so software walkthroughs are recorded elsewhere and uploaded the same way.

The reason most product demos underperform is structural.

They are organised the way the product was built, which is the order that makes sense to the person who made it.

A viewer is running a different sequence.

Is this for me. Will it work with what I already own. How long until it is set up.

What happens when it breaks. Is there a catch in the price.

A demo that answers those in that order converts better than a demo that shows more capability.

The objections are the actual purchase path.

That is a script decision, and it is why the transcript being the interface matters here.

Reordering the argument is reordering sentences rather than rebuilding a video.

Delete the feature nobody asked about and the cut rebuilds around what is left.

The practical workflow for a physical product.

Spend twenty minutes filming the one shot where the product visibly does the thing.

Upload it to your workspace library and it is available to every demo you make afterwards.

Then film reads. One about setup, one about who it is wrong for, one about the accessory people always ask about.

One about what happened after six months of use.

Each read is a separate project at about 110 credits, from the same three minute ceiling.

Ten of them is roughly 1,100 credits, inside one month on Basic at $39.99.

Roughly 5 to 8 percent of creatives carry spend, per Motion's analysis of 550,000+ Meta ads.

Ten demos is a real test where one hero demo is a guess.

Coverage caps at 40, 45 or 52 percent depending on pace, so the person explaining stays on screen. That is what a demo needs.

A demo watched in silence is a person mouthing something unless the captions carry the specifics.

So the numbers, the model name, the setup time and the price all need to be spoken.

Then emphasised so they land visually. Per-word emphasis exists for exactly that.

Six caption packs are available, plus colour, weight, size and position.

Position matters because the interface of whichever app it plays in will cover part of the frame.

You can also fix a mis-heard word in the transcript.

That matters when the product name is unusual and the machine spelled it the way it sounded.

Write the opening headline yourself, or take the one it drafted from your own words.

One habit is worth building. Say every number out loud in the read rather than assuming a caption will carry it.

The caption is generated from what you said, so an unsaid number never appears.

Exports are 9:16 MP4 only, which is the shape every vertical placement takes.

Nothing widescreen for a website hero or a trade stand comes out of here.

Source takes cap at three minutes and one project makes one finished video.

There is no timeline, no keyframes and no layers, which is why a tenth demo costs minutes.

There is no screen capture built in. Record software walkthroughs with your own recorder and upload them to the library.

From there they land under the words that describe the step, exactly like a filmed product shot.

Nothing of yours is stamped on the render, so the caption treatment carries the brand.

Somebody has to be on camera, because the transcript is the interface.

If nobody will, the avatar module builds a presenter from a photo and a voice-over at 320 credits per minute of render.