Architecture
- 01Supplier link
- 02Scrape product
- 03AI copy & SEO
- 04AI photography
- 05Shopify draft
- 06Slack
The problem
Every product listing was built by hand from wholesale supplier pages, plus a photographer — around 30 to 60 minutes of manual work per product.
What I built
- 1
A 1688.com supplier link pasted into a Google Sheet starts the run; the system scrapes colours, sizes, images and pricing.
- 2
AI picks the best-selling colours for the UK market and writes native British-English SEO copy: title, description, meta tags and Google Shopping fields.
- 3
It generates studio-quality on-model photography in four angles for every colour — up to 16 images per product — keeping the real garment faithful to the source.
- 4
It publishes a fully built Shopify draft with colour variants, SKUs, pricing, SEO fields, category and the right photo mapped to each swatch, then posts the result to Slack.
Result
A hands-off run of about five minutes costing a couple of dollars in API calls, running two tailored product lines, with auto-refreshing authentication, retry logic, memory-safe image processing and Slack error alerts.
