A useful ChatGPT product video prompt supplies verified product facts, available assets, a specific audience, length, format and a budget. Ask for a script and shot list you can review before paying for generation. Start with the copyable template below; it works as a written brief without installing Orkas or a CapCut plugin.
Copy this product-video prompt
Help me plan one product video using the brief below.
PRODUCT: [name and what it does]
AUDIENCE AND NEED: [one buyer and one problem]
VERIFIED FACTS: [approved claims + source for each]
DO NOT CLAIM: [unproven benefits, discounts or testimonials]
ASSETS AND RIGHTS: [photos, clips, logo, dimensions, permission]
CHANNEL: [product page, ad, Reel, Short, etc.]
LENGTH / ASPECT RATIO: [seconds] / [9:16, 16:9 or 1:1]
LANGUAGE AND TONE: [language, reading level, brand voice]
CTA: [one next step + exact URL or wording]
BUDGET: [approved paid services and maximum extra spend]
First flag missing facts, assets, permissions or capabilities.
Do not invent product features or imply you inspected an absent file.
Then deliver:
1. A short production brief and a claim-to-source list.
2. A timed script: hook → product evidence → CTA.
3. A shot list with start/end, asset, action and on-screen text.
4. Caption text; only provide timed SRT when the cut is fixed.
5. A review checklist and a list of missing or generated assets.
Use real product visuals where accuracy matters. Do not alter labels,
shape, colors or included accessories. Mark proposed generated visuals.
Ask before spending beyond the approved budget.
If you cannot render or export a video, deliver the production packet
and say clearly that no video has been created.Download the prompt as plain text. For a physical product, add dimensions and the exact pack contents. For software, add real screens or use clearly labeled explanatory graphics. A service needs evidence of the deliverable rather than invented customer reactions.
Worked example: a 15-second Orkas product video
The real product in this example is Orkas. We used its own logo and public product facts to make a short, silent reference cut for this guide. The visuals are editorial title cards, not captured app screens. This is a production example, not a customer testimonial or a benchmark of ChatGPT, CapCut or the Orkas app.
Input: Orkas; independent creators and small teams preparing a video; English; 15 seconds; 16:9; one CTA to the video-workflow page. Approved facts come from the video workflow, VideoStudio page and macOS and Windows downloads. Assets: Orkas’s owned logo and original text graphics. No supplied customer footage, generated people, music or voiceover.
Boundaries: no CapCut-in-Orkas claim, no invented customer results and no guaranteed time saved. The example makes one landscape version only. It does not demonstrate product photography, lip sync, translation or a full app execution.
| Time | Script / on-screen copy | Shot and source |
|---|---|---|
| 0–5 s | Make your next product video. | Orkas logo and original title card; a task invitation, not a performance claim. |
| 5–10 s | Bring your brief. Review the plan. | Text-led explanation of the brief and proposed-cut workflow described on the video-workflow page. |
| 10–15 s | Review video, captions and cover. Explore the Orkas video workflow. | Text-led deliverables and a CTA to orkas.ai/use/make-videos/. |
What was actually produced: this 15-second H.264 MP4, a timed on-screen-text track, a shot-list CSV, a filled brief and a claim review. The script and graphics were prepared in this Codex session and assembled locally with FFmpeg. They were not produced by running the prompt inside ChatGPT, the CapCut plugin or the Orkas app. The example shows a checkable production packet and reference cut; results from another tool will vary.
Costs and revisions: no additional paid image, video, voice or music service was called for this cut. Existing model-access and computer costs are excluded. The review rejected simulated app screens and a claim of measured time savings; the final version uses labeled text graphics and sourced capabilities. It still needs a brand and channel review before use in a campaign.
Download the filled brief and review notes, shot list, timed text as SRT or reference MP4. The SRT follows these fixed title-card timings; it is not a speech transcription.
Turn your own product facts into better shots
Show evidence for the promise. If your selling point is a removable lid, supply a real opening shot. If it is a software workflow, supply an accurate screen recording. A generic cinematic shot cannot establish a feature or a measured result.
Keep the first version small. Choose one audience, one need, one format and one CTA. Ask for unavailable assets as a separate list. For a video that must show the actual product, use real footage or photos rather than accepting a generated substitute with the wrong label or shape.
Fix the cut before timing captions. A script is not a transcript. Time subtitle files against the final cut or voice track, then check the beginning, the end, spelling, reading speed and placement. After changing scene lengths, review those times again.
Specify what can cost money. If your approved budget is zero for new services, ask for a plan using supplied assets and local tools. If you approve generated footage or voice, record the provider, model, requested quantity and charge before running it. Do not assume a text subscription covers every media-generation service.
Where to make and finish the video
You can take the production packet to an editor you already use. For the official connected CapCut route, see the CapCut ChatGPT / Codex workflow guide. Availability, outputs and charges depend on that route; this example does not test it.
If you want product research, scripts, assets and reviewable video files in an independent desktop project, explore Orkas’s video workflow or VideoStudio. Orkas runs on macOS and Windows. Media generation may use configured services and incur charges; Orkas is not a plugin that runs inside ChatGPT.
Before publishing, play the whole export, inspect the product and every claim, check asset permissions, captions, audio if present, aspect ratio and CTA. Keep the source files and the approved brief alongside the final version so a later change has a clear starting point.
Have several videos to edit?
The batch video editing case shows three existing clips turned into nine exports with VideoStudio and Claude Code in Orkas. Use its task template when your starting point is footage and repeatable format instructions.
Product-video prompt questions
What should I include in a ChatGPT product video prompt?
Include verified product facts and sources, the audience, available assets and rights, channel, length, aspect ratio, language, one CTA and a budget. Request a timed script, shot list, captions and review checklist, and ask the assistant to flag missing information.
Will the prompt create a finished video in ChatGPT?
A text prompt can produce a production packet. A rendered video requires available video tools or a separate editor. Ask for the actual export and inspect it; do not treat a written completion message as proof that a video exists.
Can I use the template without Orkas?
Yes. The template and downloads are plain text and ordinary media files. You can keep using ChatGPT for the written plan and choose an editor separately.
Is the reference video a test of Orkas or CapCut?
No. It is an editorial example about the real Orkas product, prepared in this Codex session and assembled locally with FFmpeg. It does not establish that the CapCut plugin works inside Orkas or that either app completed the workflow.