Kit
Realistic AI UGC ads, from avatar to finished edit.
The production half of UGC ads: prompt library, 10 formats, 3 agents.
More real output
- AI-generated performer
Worked example: 3 skincare myths
AI-generated performerPrompt library A: the 6-part base avatar - AI-generated performer
Prompt library B: a talking clip, piece by piece - AI-generated performer
Formats: UGC with product
How the video above was made
The silent montage: three people, five shots, 7 seconds · Start frames in GPT Image 2 at 2K; four clips in Grok Imagine, 1080p, 6 s, silent (48 credits each); cut in a video editor. · Full example page
Five silent shots, each from a 6-part avatar image (lesson 3.2.4), animated in Grok Imagine, then cut to about 1.5 s each.
SHOT 1 (man after a swim, kept 2.7 s at 2x): He turns his head from three-quarters to face the lens and blinks once. No talking, mouth closed. Slow creeping push-in ending on an extreme close-up of the eyes exactly on the key line, then hold. One continuous take, no cuts or scene changes.
SHOT 2 (woman in a parked car, kept 1.5 s): She pushes the bottle toward the lens until the label fills the frame, then pulls it back and grins. No talking, mouth closed. Very slight, natural handheld movement for a human feel. One continuous take, no cuts or scene changes.
SHOTS 3-4 (same woman, one clip, two moments, 0.6 s + 0.9 s): Keep the same person and place. She massages the visible product into her face in gentle upward circles; the product steadily decreases until absorbed, no residue. Cinematic light, no talking, mouth closed. One continuous take, no cuts.
SHOT 5 (woman in a pool, kept 1.8 s): She slowly tilts her head, looks into the lens, then lets her eyes half close. No talking, mouth closed. Very slight, natural handheld movement for a human feel. One continuous take, no cuts or scene changes.What’s inside
- Onboarding: the tools, model marketplaces and how the prompt library is organized
- A model catalog: which image and video model to use when the avatar talks, and when it doesn't
- Fundamentals: imperfect avatars, the 6-part avatar method, the 3 levers of an ad, 12 angles and 5 awareness levels
- Prompt library A (images): outfits, backgrounds, angles, the character sheet, product in hand, phone and laptop screens
- Prompt library B (video): camera moves, micro-actions, voices, accents, and fashion and skincare actions
- Ten formats with worked examples and editing blocks: person to camera, with product, fashion, apps, podcast, voice-over, dualcast, viral trends, static image ads and the silent montage
- Voice control and cloning with ElevenLabs
- Three agents: the Shot Blueprint, The Strategist and the Static Ads Architect
Includes the interactive course + PDF
- 1Start here6 lessons
- 2Model catalog2 lessons
- 3Fundamentals9 lessons
- 4Prompt library A: The image9 lessons
- 5Prompt library B: The video9 lessons
- 6Formats14 lessons
- 7Voice control and cloning4 lessons
- 8The three agents4 lessons
- 9Before you post1 lesson
58 lessons · about 1.5 hours of reading
Refunds
If it isn’t useful, email support@aiofftheshelf.com within 14 days of buying for a full refund. Read the refund policy.
Questions
Do I need the AI UGC Ads Kit too?
No, but they're two halves: the Ads Kit decides what to say, this course makes it real. Both are in The AI Video Playbook.
Can the AI avatar say it used the product?
No. An AI person claiming personal results is a fake testimonial under the FTC rule, even with a disclosure. The agents and examples are written to educate, demonstrate and compare instead.
What do I get?
A 30-page PDF with the full prompt library and all three agents. It's also included in The AI Video Playbook.