UGC Video Generation Pipeline
Engineered an end-to-end system that automates the generation of User-Generated Content (UGC) promotional videos. It uses an n8n orchestration layer as the central processing unit to route workflows to AI models like Sora, Veo, or Nano Banana.
Client
Service Provided
The Goal: Zero-Touch Content Production
The objective was to eliminate manual video production by extracting raw input data from a centralized database (Google Sheets). The system takes a base product photograph, the Ideal Customer Profile (ICP), and key product features to initiate the workflow autonomously.
1
The Architecture: Dynamic Routing & Synthesis
For workflows requiring enhanced realism, an LLM agent generates a detailed image prompt. This is sent to the Nano Banana model to synthesize a realistic image of a human interacting with the product. An advanced LLM agent then evaluates the product features and reference image to autonomously generate a natural, conversational dialogue script tailored for a UGC-style format.
2
