← All case studies
E-COMMERCE · RETAILAI-POWERED

The AI Catalog & Content Engine

The goal: Automate the cleanup and translation of massive supplier product catalogs.

WHO THIS WAS FOR

An online retailer selling supplier catalogs sourced from overseas.

PythonOpenAIFlask (control dashboard)PrestaShop

The situation

New products arrived from suppliers with messy listings, often still in the supplier's original language, with clumsy translations, broken pricing, and titles that read like they'd been typed by a machine. Every new arrival meant hours of manual cleanup before it could go live, and the store was losing sales because listings didn't look trustworthy.

What we built

A one-click system that takes the raw supplier feed, cleans up the text, translates it properly into the store's local language, and reformats every listing into clean, readable, on-brand copy, then pushes it straight to the live store. It's controlled through a simple internal dashboard, so staff can trigger a new sync without touching any code.

Along the way, we also fixed a checkout bug in the same pipeline: the store was silently rejecting any price with cents in it, quietly blocking real purchases.

How it works

  1. 01A script pulls the freshest product data straight from each supplier's feed.
  2. 02The raw text, often in another language and sometimes with broken formatting, gets sent through an AI cleanup pass that rewrites the title and description in natural, correctly-translated language.
  3. 03The cleaned listing is checked against existing products by its unique reference number, so nothing gets imported twice.
  4. 04The finished listing pushes live to the store automatically, with the same formatting and quality every time.

The challenges we solved

  • Every new listing needed a permanently unique web address, and long AI-written titles kept breaking the store's strict character limit for that address, so imports were failing outright until this was fixed.
  • The store's live product categories didn't match the ones used during testing, so new products were landing in the wrong sections until the mapping was corrected against the real site.
  • The checkout was silently rejecting any price with cents, unrelated to the catalog work, but actively costing sales, so it was fixed as part of the same project.

The result

The team no longer manually rewrites a single listing. New arrivals are translated, cleaned, and live on the store automatically, and the pricing bug that was quietly blocking checkouts is gone.

100%automated translation & formatting
~22 hrssaved weekly (≈500 products × 2.5 min)
0hours writing product descriptions
0duplicate entries or catalog errors
* Hours-saved figures are estimates based on typical task frequency and time per task, not measured client data.

Want a system like this built for your business?

Book Free Discovery Call
WEEKLY AI SHORTCUTS TO BUY BACK YOUR TIME

Delete one hour of busywork this week.

Get our free "Busywork Elimination Framework" and join our Tuesday newsletter. Every week, we send you one manual task to eliminate and the exact AI tool to replace it.