← BackAI Development

Poke Price

The problem

A client reselling Pokémon cards online sets prices by hand, one SKU at a time: check the going rate across several marketplaces, compare it to his catalogue, then reopen his store to fix every listing. On a catalogue of several hundred cards, it all has to be redone the moment the market moves.

His store — Japan Pop Mart
The solution

A tool that closes the loop on its own: it identifies the card, fetches its price across the various marketplaces, and updates the store's product listing.

ResultUsed in real conditions to fix Shopify prices; OCR pipeline working end to end.

PythonWeb scrapingCursorOCRShopify
Charizard ex
Booster Pokémon
Drag left or right to tear open
Click to change

0€

Raw

0€

PSA 9

0€

PSA 10

+12%

Charizard ex · 151 · 199/165

Under the hood

How it works

A card identified from a photo, or a Shopify CSV export, goes through an in-house OCR pipeline (Tesseract), a match against a reference catalogue, then multi-source scraping of market prices — to automatically correct the price shown on the store. The whole thing was built with Cursor.

01 — ImportPhoto or Shopify CSV export
02 — OCRTesseract + in-house pipeline
03 — MatchingMatch against a reference catalogue
04 — ScrapingPrices across marketplaces
05 — CorrectionPrice update on Shopify

Nothing in this chain is specific to Pokémon cards or to Shopify — identify a product, read its market price, update the catalogue: it transfers as-is to any kind of product and any platform.