Case Study - 1000 products imaged in one day
The e-commerce catalogue landed in the system without a single product photo. An AI pipeline with twenty parallel agents delivered 100% catalogue coverage in under a day.
- Client
- Marion Mizzi Wellbeing
- Country
- 🇲🇹Malta
- Year
- Service
- AI Automation
Overview
I've built dozens of AI automations for Marion Mizzi Wellbeing. I'm describing this one because it shows what current AI can do - and it's easy to measure.
When the e-commerce catalogue landed in the system I run for them, it turned out the products had no photos at all. Searching for packshots by hand, across dozens of manufacturer catalogues, would have taken weeks. The AI pipeline did it in under a day.
How it works
- Twenty AI agents in parallel searched for official packshots via EAN codes and manufacturer catalogues.
- Every photo found went through resolution and format checks.
- A separate step visually verified that the photo actually showed the right product, not a different variant or packaging.
- Approved photos went to production through the system's API - every change audit-logged, every change reversible.
The result: 100% catalogue coverage in under 24 hours - 911 products with photos, including 736 official manufacturer packshots. The rest got neutral AI-generated placeholders, swapped for real photos as they become available.
I've done dozens of things like this for this company - this one just happens to be easy to show in numbers.