How to Make a Promo Video with AI (From One Prompt)

Making a promo video used to mean a brief, a quote, a shoot, and three weeks of edits. The newest AI video tools collapse that into a conversation: you describe the product, the system does the production, and you approve the work at the moments that matter. This guide walks through the whole process — from writing the first prompt to publishing the final cut — and shows where human judgment still earns its keep.

What “making a video with AI” actually means now

Early AI video tools generated a few seconds of dreamy footage from a text prompt. That is not a promo video. A promo video is an argument: it has a message, a structure, a voice, and a call to action. Modern AI video agents treat it that way. Instead of one model guessing at pixels, an agent runs a pipeline that looks a lot like a small studio:

• Research. It reads your website, product pages, and anything you upload, and works out what the product is, who it is for, and what proof exists.
• Creative brief. It proposes a direction — the one message, the tone, the visual style, which assets to use.
• Script. It writes the narration and scene plan, timed to the length you asked for.
• Voiceover and music. It records an AI voice, picks music, and mixes them.
• Compose and render. It animates scenes, adds captions, and renders the finished file.

The important part is the review gates between stages. On UltraVora, the pipeline pauses twice — at the brief and at the script — so you approve the thinking before any rendering happens. Fixing a sentence in a script costs seconds; fixing it after the render costs a re-render.

Step 1: Write the prompt like a one-paragraph brief

The prompt is the highest-leverage minute of the whole process. You do not need to be a copywriter; you need to make five decisions and state them plainly:

1. Product: what it is, in one sentence. A URL helps more than adjectives — the agent will research it.
2. Audience: who should stop scrolling. “Small e-commerce brands” beats “everyone.”
3. The one message: the single idea a viewer should remember. One. Videos die from carrying three.
4. Tone: two or three words — “warm and unhurried,” “punchy and confident.”
5. Length and destination: a 15-second vertical ad and a 60-second homepage video are different creatures. See our guide on how long a promo video should be if you are unsure.

A prompt that contains all five, in plain language, is enough:

“A 30-second promo for glowmug.com — a ceramic mug that keeps coffee at temperature for two hours. Audience: remote workers who sip slowly. Message: your coffee stays perfect from the first call to the last. Tone: warm morning light, calm confidence. Vertical, for Instagram Reels.”

Step 2: Read the brief like an editor, not a fan

The creative brief is where the agent shows its understanding of your product. Check three things before approving:

• Positioning. Did it lead with the thing your customers actually buy you for? If it latched onto a minor feature, say so in a sentence and let it revise.
• Claims. Anything that reads like a number or a guarantee should be true. You are publishing this.
• Assets. Curate the images it researched — drop the outdated screenshot, keep the good product shots. Bad source assets are the most common cause of a “that’s not my brand” final cut.

Step 3: Approve the script — the contract for the edit

Everything downstream — voiceover, scene timing, captions — is built from the script, so this is the gate worth the most attention. Two practical checks:

Read it aloud. Narration runs at roughly 150 words per minute, about 2.5 words per second. A 30-second video holds 70–80 spoken words. If the script feels crowded read aloud, it will feel worse with visuals competing for attention.

Judge the first line only on one question: would a stranger keep watching? The hook does not need to mention your product; it needs to name a moment your audience recognizes. Our promo script templates break down five hook patterns that consistently work.

Step 4: Voice, music, captions

The voice is your brand’s sound — audition a few before settling. A calm, lower-pitch read suits premium products; a brighter, quicker read suits consumer apps. (More on this in choosing an AI voiceover.) Keep captions on regardless: much of social video is watched muted, and captions are the only narration those viewers get.

Step 5: Render, review, and ask for tweaks

Rendering happens on the server — on UltraVora you can close the laptop and the job keeps going. When the cut is ready, review it like a viewer first (does it hold attention?) and like an owner second (logo, colors, claims, CTA). Then ask for specific tweaks: “slow the second scene,” “swap the music for something quieter,” “end on the product shot.” Targeted notes re-render what changed instead of starting over.

What it costs and how long it takes

The economics are the quiet revolution. On UltraVora’s usage-based pricing, a typical one-minute promo runs about 400 credits — roughly four dollars — and a first cut arrives in minutes, not weeks. That changes behavior: you stop treating a promo video as a quarterly capital project and start treating it like a draft you iterate on.

A 60-second pre-flight checklist

• One product, one audience, one message, one CTA.
• Length matches the placement.
• Brief approved with correct positioning and true claims.
• Script survives being read aloud.
• Voice matches the brand; captions on.
• Final cut checked on the device where it will actually run — usually a phone.

The best way to learn the workflow is to run it once on your own product. Browse real examples in the Product Plaza to calibrate what finished results look like, then start with a free session — the first draft will teach you more than any guide.