Can AI Make Commercials? How Do You Build a Full Ad Series? NadouPro Ad Production: From One Product to Four Ad Formats

A commercial usually needs four stages from idea to finished video: product reference, format selection, content generation, and batch testing. In a traditional workflow, each stage uses different tools, and information is lost during handoffs. NadouPro's Canvas Agent brings all four stages into one conversation—starting from the same product reference and producing multiple ad formats at once.

Four Ad Formats AI Can Create

Choosing the right format matters more than choosing the flashiest one. The following four formats cover most needs in an ad series:

UGC-style video: Builds trust in feeds because it looks like something posted by a real person—a single performer, native subtitles, and a hook in the first two seconds. Suitable for placements that need to establish trust, with a length of 15-30 seconds.

Cinematic product video: Sells the product itself rather than someone's review. This format is needed for a product detail page hero image or the opening of a TV-level commercial, with a length of 8-15 seconds.

Social media ads: A short, fast format designed for algorithmic feeds—carousels can use still frames with motion effects, quick cuts, and copy that can be read at a glance, with a length of 8-15 seconds.

Explainer video: Clearly explains a product or concept by turning a script, link, or feature list into a video with voiceover. Suitable for landing pages, feature demos, and onboarding, with a length of 15-30 seconds.

Running all four formats from the same product reference, instead of shooting each format separately, changes the economics of ad production.

How NadouPro Canvas Agent Works

The Agent entry point in the upper-right corner of the canvas is the core of conversational ad creation. Once connected, choosing formats, writing briefs, and running generation all happen in the same conversation.

Canvas Agent supports referencing existing canvas nodes, importing from the asset space, and local uploads. A single call can support multiple reference inputs: product images, spokesperson faces, voiceover clips, and shot-style references can all be provided at the same time.

The working pattern is always consistent: attach reference images, describe the need in one or two sentences, and confirm the output stage by stage. Product videos can first generate still frames to confirm the direction, then generate motion—catching issues before they are locked into a complete clip.

Step-by-Step Workflow

Step 1: Attach the product reference. Upload a product photo or product link in Canvas Agent, and add a one-sentence ad objective: key selling point, placement channel, and deadline. This reference is reused across all subsequent formats, so you do not need to upload it again each time.

Step 2: Write the requirement description. The short version only states the angle, selling point, tone, and what needs to be shown, while Canvas Agent fills in the details. In the full version, you write your own shot-by-shot brief: camera movement, actors, setting, material realism, audio, and consistency rules.

Below are example prompt structures for the four formats.

Key points for a UGC video prompt (using a honey product as an example):

UGC-style vlog video, 10 seconds, vertical 9:16, smartphone-shot feel, natural daylight. Character: a woman in her twenties, copper-red curly hair, freckles, thin metal-frame glasses, green transparent trench coat, gray-yellow bucket hat. Scene: alpine meadow, wildflowers, old wooden fence, snowy mountain background. Product: a jar of golden raw honey, blue lid and label. Timeline: 0-3 seconds handheld selfie while walking, lifting the honey jar to cover half the frame, "POV: you brought a jar of honey up the mountain"; 3-7 seconds cut to a fixed tripod side shot, she sits on the fence, twists open the lid, scoops honey with a wooden spoon, tastes it, closes her eyes and smiles; 7-10 seconds grabs the phone and returns to selfie, honey jar beside her face, "slow morning, literally".

Key points for a product video prompt:

Dreamlike 3D animated brand film, vertical 9:16, flat color-block animation style. Color palette: cream, honey amber, butter yellow, with blue as the only accent color. One continuous shot: slowly rising from the ground to the sky. Opening on stylized grass with flat flowers and doodle bees; the camera rises past a bridge and a flat train, a stylized city appears, and a giant honey jar tower stands in the center with a clearly readable blue label; continue rising through cream-colored clouds to the top, where a bubble-like brand logo appears.

Key points for a social media ad prompt:

15-second UGC-style vertical video, 9:16, handheld iPhone feel. Character matches the reference image: woman in her early twenties, long dark-brown wavy hair, black octagonal glasses, light-blue floral headscarf. Product prop: five oversized helium balloons shaped like honey jars. Scene: uptown Manhattan block, red-brick apartment buildings, fire escapes, yellow taxis. Timeline: 0-3 seconds crosses the street holding five glowing honey balloons; 3-6 seconds quick-walk selfie under a fire escape while looking at phone; 6-9 seconds poses against a red-brick wall; 9-12 seconds balloons tilt in the wind and she turns around laughing; 12-15 seconds stops, looks at camera, then laughingly pushes the balloons into the lens to end.

Key points for an explainer video prompt:

Low-poly geometric 3D style, warm amber monochrome tone, one continuous pull-back shot. Opening close-up of a bee perched on the edge of a honeycomb, with glossy honey flowing in the nearest honeycomb cells; the camera pulls back and rises, revealing the full interior of a dome-shaped hive, while two background bees circle slowly. Voiceover: "Back at the hive, nectar turns into rich golden honey." Music: warm strings, flute melody, soft glockenspiel, 78 BPM, a content and calm rhythm.

Step 3: Confirm still frames before generating motion. Canvas Agent first returns a starting frame—confirm the face, scene, and composition. Finding a wrong direction in a still frame costs far less than finding it in the finished video.

Step 4: Generate the complete clip. After the still frame is confirmed, motion, voiceover, and lip sync return together, with no separate voiceover step or manual syncing required.

From One Variant to Batch Testing

After one ad works, Canvas Agent becomes a tool for scaling. Describe the batch requirement in natural language, and the agent plans tasks, selects models, and runs multiple variants at once. Canvas Agent preserves context, so the brief and brand reference carry through the batch without needing to be specified again for each item.

For larger-scale batch needs, the CLI command-line tool supports scripted calls, allowing the ad creation workflow to be embedded into an automated pipeline. Combined with team collaboration features (up to 100 people online at the same time), this enables studio-level scaled production.

A Few Practical Tips

• Lock the product reference at the start of the conversation. Canvas Agent automatically reuses it in subsequent formats. Describing the product again each time is a major source of inconsistency.

• Write the hook first for UGC videos. What it needs from you is the angle: the problem, selling point, and tone—this is where variants differ from each other.

• Make one of each first, then batch. Batch tasks amplify any direction you give, including mistakes. Confirm that a single result is right before scaling up to five variants.

• Select the versions for placement from the finished outputs. First shortlist the clips you want to run in the canvas, then export them.

Frequently Asked Questions

Q: Can the four ad formats start from the same product reference?

Yes. After uploading the product reference once in Canvas Agent, all formats share that reference. UGC videos, product videos, social media ads, and explainer videos are produced from the same asset source.

Q: Do I need to write detailed prompts myself?

Not necessarily. A short description only needs to state the angle, selling point, and tone, and Canvas Agent will expand it into a complete prompt. Of course, the more detailed the prompt, the more precisely you can control the visuals. The prompt examples for the four formats show the structure of production-level prompts.

Q: What does still-frame confirmation mean?

Before generating the video, a starting-frame still image is created first to confirm the face, scene, and composition direction. Catching directional errors at the still-frame stage costs much less than discovering them in the finished video and regenerating it.

Q: How do you create batch variants?

Describe the batch requirement in natural language in Canvas Agent. When testing hooks, keep other variables fixed and change only the opening line, producing 3-4 versions at once. For larger batches, use the CLI command-line tool for scripted calls.

To try more features, visit NadouPro: NadouPro

Contains AI-generated information that may be incorrect. Please verify independently.