Make a Product Ad with AI: A One-Evening Plan

Start with your product facts, not with the tool

Most generated ads that look cheap did not fail on voice or visuals. They failed because what got handed to the model was vague: "skincare product, write me a catchy ad." The model does not know your product, so it fills the gap with generic lines that sound like a thousand other ads.

Before you open any tool, write five lines on a piece of paper:

• What the product is, in one sentence a person who has never heard of it would understand.
• Who exactly it is for, and the moment they need it.
• Three concrete selling points. Not "high quality," but "lasts two months" or "works with no installation."
• The first objection that usually stops the sale.
• What you want the viewer to do the second the video ends.

Those five lines are the ad. Everything after that is execution. And the real payoff is that you write them once and reuse them for every ad you make after this one.

The structure of a short ad nobody swipes past

A short ad that works is not a condensed product explainer. It is a very tight sequence:

1. First two seconds: a situation or a result. No tagline, no logo. "Losing hair every time you shower?" beats "Introducing..." every time.
2. Quick definition: what the product is and what it does, in one or two sentences.
3. Visual proof: the actual product on screen. A real photo from your phone is enough, and it beats any generated shot.
4. One objection, handled: price, or time, or doubt that it works. Pick one. Only one.
5. A direct ask: where to buy, what to type, what happens next.

A sane runtime for that sequence is 20 to 40 seconds. Go longer and you need a story, not an ad.

The practical steps: from five lines to a finished video

Once the facts are down, production itself is shorter than you expect. At MaqtAi, for example, Presenter ads run like this:

• You pick a ready-made presenter (the options include Middle Eastern faces and presenters in hijab) or your own clone from Clone Studio.
• You paste in your product facts and selling points. You do not have to write a script: the ad gets written from those facts at launch, and it will not claim anything you did not write. If you leave the product fields completely empty, the video turns into general talk about the topic with no sales path in it.
• You upload up to six product photos, shown full screen while the presenter keeps talking, in the order you uploaded them. The opening and the call to action always stay on the presenter, and shorter videos automatically show fewer photos.
• You review the six lines and edit exactly what you want said, then hit Launch. Nothing gets generated and nothing gets charged before that click.

The full walkthrough of the flow is in the three-step explainer, and the credit cost of each video is shown before you launch on the pricing page.

Two technical notes everyone asks about: the presenter ad has a fixed six-scene format, so there is no duration picker and no quality tier. It always runs on the best engine. And it outputs a clean voiceover by default, meaning no background music unless you pick a track yourself.

"Is it going to look fake?" The honest limits

This is where I have to be straight with you, because it decides whether you use the tool at all.

Character cloning is still in Beta. The likeness is close but not perfect yet, and the two hardest parts are mouth movement and hands. The biggest lever is in your hands: a clear front-facing photo with good lighting, and a calm voice recording in a quiet room for about 25 seconds. Free accounts get still images with a zoom motion plus a watermark, and the animated presenter is a paid feature.

There are things this method does not do. Once a presenter ad is finished, you can only edit the music and the captions, because changing the words or the voice would break lip sync. For the same reason, presenter videos cannot currently be translated into another language.

And bigger honesty: if your product needs real sensory proof, like fabric texture or taste or how it fits on a body, a clip from your own phone beats any generation. Use AI for the presenter and the structure, use your real footage for the product. That mix is what actually works.

Does an ad get less reach because it was made with AI?

No. No platform demotes a video simply because AI made it. TikTok, Instagram, and YouTube rank short videos on viewer behavior: watch time, completion rate, saves and shares.

Three things do actually hurt you: near-duplicate posting of identical files across multiple accounts, weak first two seconds, and ignoring the disclosure rules for realistic synthetic content. The disclosure itself does not reduce distribution, but skipping it can get the video pulled. Read YouTube's policy on disclosing altered or synthetic content and TikTok's guidance on AI-generated content before your first post, and then you never have to think about it again.

Where to publish it, and the real limits on each platform

Before you build a posting plan, know what is actually possible today:

• YouTube Shorts: direct publishing, and any channel works.
• TikTok: the video lands in the drafts inbox in your app, and you open the app and hit post to finish. That is normal right now, not a bug.
• Instagram Reels: requires a Business or Creator account linked to a Facebook page. Personal accounts cannot publish through the API.
• Facebook Reels: publishes to a page you manage, not to your personal profile.

If you run more than one store or brand, you can connect several channels and pick the destination for each video at publish time. Details on the features page.

The smallest step you can take today

The real cost of waiting is not money. It is that you sit there with no ad to test while your competitor posts their third version and already knows which opening stops a thumb and which one does not. Your first ad is not supposed to be perfect. It is supposed to give you data.

So keep the step small: take the five lines you wrote at the top of this article, pick one product, and make one ad tonight. New accounts that verify their email get 450 free credits, no card, and that is enough to see with your own eyes what your product looks like in a vertical video before you pay anything.

FAQ

How long does making the ad actually take? The prep (writing your product facts and selling points) is the longest part, and it might take you fifteen minutes the first time. The video itself usually renders in under three minutes after you hit Launch.

Can I use my own face and voice instead of a stock presenter? Yes, through Clone Studio: a guided voice recording of about 25 seconds and a clear photo of your face and shoulders, with explicit consent. There is also a voice-only mode with no image, which narrates any video without your face appearing.

Can I use the video in paid ads? The videos you produce are yours, and you can publish and use them commercially on your channels. Just follow the synthetic content disclosure rules on whichever platform you are advertising on.

Can I make a version of the ad in another language? Presenter videos cannot be translated right now, because the lip movement would not match the new words. Image-and-motion videos do support other-language versions with a new voice and new captions over the same scenes.

Results vary from product to product, and nothing here is a promise of sales or income.