ViralShop.AI
From manual content creation to 300+ users and R$5,000+ in revenue, built in under 30 days by someone who has never written a line of code.
Tech Stack
The Problem
TikTok Shop is one of the fastest-growing e-commerce channels in Brazil, but creating content for it is brutally repetitive. Every product listing needs high-quality images and videos: different angles, lifestyle contexts, movement-driven clips that mimic viral formats.
Pedro Paulo was living this problem firsthand. As a TikTok Shop seller creating AI-powered product visuals, he was spending entire days crafting individual creatives. The process worked, but it didn't scale.
Nothing on the market did exactly this. So he built it.
The Solution
ViralShop.AI is a SaaS platform that automates TikTok Shop content creation using AI. Users upload a product, and the platform runs it through an enrichment pipeline that extracts product details, builds context, and generates production-ready images and videos.
The entire application was built from scratch in less than 1 month. Pedro had no programming background. Every line of code was produced through AI orchestration using Claude Code. No dev team, no freelancers, no bootcamp.
Architecture
Frontend: Next.js 16 with React 19 and TypeScript, deployed on Vercel. 94 components across 27 releases.
Backend: Express.js 5 on Render (Ohio region, migrated from Oregon after Gemini API IP-blocking).
Database: Supabase with 56 SQL migrations for users, credits, projects, and billing.
AI Engine: Google Gemini API across 3 Google Cloud projects for redundancy.
Payments: Mercado Pago with credit-based billing model.
Key Features
Enrichment Pipeline: Four-stage flow: product extraction, enriched context generation, prompt building, and image/video output. Simplified from ~10 JSON fields to a 120-150 word narrative format, producing dramatically better AI outputs.
POV System: Products categorized into 2 display contexts: HANGING (clothing) and SURFACE (everything else). Originally 4, simplified after testing showed no quality improvement from added granularity.
Clone Viral: The standout feature. Users feed in a viral TikTok video, and the platform replicates those movements on a new product using the Kling Motion Control API.
Credit-Based Billing: Three plans (Start R$67/25 credits, Viral R$97/50, Escala R$197/100). One credit = one AI-generated image.
Lessons Learned
Over-engineering is the silent killer. An entire "affordance system" was built and then ripped out completely. Build for the problem in front of you, not the one you imagine having in six months.
Prompt engineering is product design. The enrichment pipeline shifted from structured JSON to freeform narrative. AI models respond better to natural language. The prompt system reaching v85 reflects that prompts are living product features.
Speed is a competitive advantage for solo builders. The entire platform went from idea to paying customers in under 30 days.
The Meta-Narrative
ViralShop.AI was built entirely by a non-programmer. It's a production SaaS with paying customers, real revenue, and a codebase of 94 components across 27 releases.
The question is no longer whether non-programmers can build real software. ViralShop.AI is the answer.
The method behind these projects, in your business.
Two 3-hour sessions on your machine, and your business leaves with a daily dashboard running. You run it on your own afterwards, without becoming a programmer.
See schedule and price