# AI UGC three-level ladder
A framework for making AI UGC (user-generated-content-style) product ads. It treats the failure of AI UGC as a problem of inputs and workflow, not of fighting the model with cleverer prompts.
| Level | Output | Core move | |---|---|---| | Beginner | A single clip of someone holding the product | Use a top video model with a reference-image creator: pinterest-reference-image-creator-build, briefed as in all-models-platform-sora-2-pro-creator-brief | | Intermediate | High-performing content | Ads built on native formats and customer research: native-format-customer-language-ad-strategy, supported by tiktok-format-to-prompt-pipeline and the logic in scroll-behavior-native-ads-attention-diversion | | Expert | A proven, repeatable system | Libraries of formats and angles, plus a Claude-to-Max-Fusion MCP link, so ads can be produced in volume |
The expert end state is a stocked library of ads that can be downloaded and handed to an editor without the founder spending hours. The source only outlines this level and defers the build to a possible follow-up video, so there are no concrete expert-level parameters to record yet.
The source shows ad results and points to heavy spend behind them. It infers that a brand wouldn't keep spending that much unless the ads were profitable. That is circumstantial support, not a demonstrated causal result.
Product-in-hand image upload, Turning image variations into talking UGC video ads, Audio realism as the last step that makes AI UGC pass