Video summary

How I Make High-Converting Dropshipping Ads Using AI

Main summary

Key takeaways

Technology

Overview

The video explains a fully AI-assisted workflow for creating high-converting dropshipping / short-form ads—not by copying existing ads, but by generating new scripts, visuals, and assembled edits using a consistent framework.

The creator claims the process can be automated to handle:

  • Scripting (AI-generated)
  • Talking-head visual generation (AI-generated)
  • Final ad assembly (editing in CapCut)

Core Ad Framework (3 Pillars)

  1. Script AI generates the ad copy using a structured approach.

  2. Visuals AI video generation creates multiple “clips” (segment-by-segment).

  3. Editing / Assembly Done in CapCut, combining A-roll + B-roll, adding captions, and improving cut pacing.

Script Generation (AI + AIDA)

Uses an AI prompt + AIDA structure

The workflow uses an AI prompt (shared in the description) to generate ad copy based on:

  • Attention → Interest → Desire → Action (AIDA)

The creator sometimes adds a conviction/social-proof layer as well.

Step-by-step workflow

  1. Paste the prompt into Claude (or another AI tool).
  2. Provide the product name (example: mermaid skin cream).
  3. Run Prompt #1 to get 10 ad angles, each including:
    • a topic
    • a hook
  4. Choose the most relatable angle for the target audience, then generate the full script.

Timed clip-by-clip script

A second Claude prompt generates a format suited for short-form pacing (about 8–10 seconds per segment), including:

  • A spoken script
  • Clip instructions to paste into the video generator
  • Timestamp / duration targets

Example constraint:

  • Each clip has a max length (e.g., ~8 seconds)

The creator also requests:

  • Detailed visual descriptions
  • A reference image of an influencer so the generated video matches the speaking character.

Video Generation Tool (kai.ai “Clean” + Image-to-Video)

The workflow uses kai.ai and its Clean video creation tool, specifically Clean 3.0 Turbo.

Process per clip

For each clip prompt (e.g., clip #1 through clip #4):

  1. Paste the clip prompt into Kai.
  2. Upload a reference image (the “AI influencer”).

  3. Set duration slightly above the Claude estimate (e.g., Claude says ~8s → set Kai to ~10s).

  4. Generate the clip (described as taking “within a couple minutes”).

Reference image sourcing

The creator finds a suitable person/image style using Pinterest or TikTok-like filming aesthetics.

QA / Iteration Guidance (Avoid Excess Regeneration)

The creator notes that some generated clips may fail or look wrong due to issues like:

  • environment mismatches
  • incorrect props
  • off-looking character details

Recommendation:

  • If a clip fails, regenerate it.
  • But avoid perfectionism or repeated re-runs.

Main takeaway: editing is where the “magic happens”, so it’s better to end up with a workable clip set.

Editing in CapCut (A-roll + B-roll + Auto Captions)

The workflow uses CapCut (described as free and TikTok-owned).

Editing steps

  1. Create a new project.
  2. Import the AI-generated clips into the timeline.
  3. Add B-roll/supporting clips of the product:
    • The creator pulls B-roll from TikTok (e.g., most viewed videos from the last 7/30 days and all-time).
    • Typically compiles ~10 extra clips.

  4. Timeline tip: disable “magnet” snapping to prevent clips from shifting unexpectedly.

  5. Improve pacing:

    • Cut dead space
    • Remove long pauses
    • Make ads snappier
  6. Add AI auto-captions:
    • CapCut transcribes and generates captions automatically.
    • Uses a preset style (stroke, font sizing/position).
  7. Match B-roll to spoken claims:
    • If the script says “shaving your legs every 2 days,” insert a shaving clip.
    • Goal: “show, don’t just tell” by visually reinforcing claims.

Example Ad Included

The video shows a complete before/after example for a razor/shaving pain angle, including messaging such as:

  • “razor blades dull… bumps, nicks, stubble”
  • “cream melts the hair off… wipe it away… no blade/no pain… no cuts…”
  • Ends with: “Ditch the razor, tap the link, and try it once.”

Tools / Products Referenced

Primary tools in the workflow

  • Claude / Claude 3.0 Turbo: ad script + clip prompt planning using AIDA
  • kai.ai (Clean 3.0 Turbo): image-to-video clip generation using the reference influencer image
  • CapCut: assembling clips, removing pauses, adding B-roll, and generating auto captions

Additional stack (mentioned, not central)

  • Shopify AI store builder (creator’s free builder)
  • AutoDS for importing products from AliExpress and automated fulfillment

Mentorship / next steps

  • Coaching/mentorship mentioned
  • A free 2-hour course mentioned as a next step

Main Speakers / Sources

  • Primary speaker: the video creator (teaching the workflow end-to-end)
  • Sources referenced through tools/systems:
    • Claude (AIDA-based scripting prompts)
    • kai.ai Clean 3.0 Turbo (reference-image-based clip generation)
    • CapCut (editing + auto captions)
    • TikTok / Pinterest (reference and B-roll sourcing)

Original video