Every product ready for sale carries a workload the shopper never sees: photography, copy, categorisation, creative production and publishing. Across thousands of products, that work consumes both budget and attention.
At Joolyoo, Mollkom’s own pilot store, we examined how AI could carry this workflow from a real product catalogue through storefront presentation and published marketing. We checked execution records against saved outputs, current product fields and publication receipts.
The challenge: giving a large catalogue a consistent presence
Joolyoo held 2,044 products at the time of review, including fragrances, personal care and cosmetics. Products need accurate images and useful descriptions. The catalogue needs a structure that helps shoppers find them. Marketing needs a steady supply of materials based on those same products.
Our aim was to reduce repeated preparation between these tasks. Product knowledge should move with the work, from catalogue content to storefront design and marketing execution.
What the records show
Scroll the table horizontally to see all columns.
| Workstream | Recorded output | What it delivers |
|---|---|---|
| Product content | 2,006 unique products | Applied improvements, with current SEO titles, SEO descriptions and short descriptions matching the latest applied results |
| Product imagery | 2,008 products / 4,016 images | A studio image and a lifestyle image present for every product in the successful group |
| Storefront | 19 sections and 9 banners | A catalogue-based layout with 3 hero banners and 6 category banners |
| Marketing creative | 117 saved designs | Distinct design URLs across campaigns and social plans, excluding the deleted item |
| Video | 9 completed videos | Saved video assets, including 2 approved assets |
| Publishing | 58 publication receipts | 29 each on Facebook and Instagram, with provider post IDs |
These workstreams overlap. A product may appear in the content, image and marketing groups. Unique-product counts prevent retries and repeated improvements from inflating the reported catalogue coverage.
From product input to usable catalogue content
Mollkom’s tools prepared product content and imagery. For 2,006 products, current SEO titles, SEO descriptions and short descriptions match the latest applied AI results. All 2,008 products in the successful bulk-image group retain both studio and lifestyle images.
This gives the merchant thousands of outputs to use in product pages and marketing. Checking product fidelity and reviewing product information remain part of accepting the output, particularly for personal-care and cosmetic products.
Current images of the same catalogue product, shown as outputs rather than platform screenshots.
Production at catalogue scale, measured in hours
Scroll the table horizontally to see all columns.
| Operation | Recorded output | Elapsed time |
|---|---|---|
| Image-enhancement cycle | 2,008 products with two images each | About 11 hours 39 minutes |
| Main content batches | 2,005 products, plus one completed earlier | About 18 hours 10 minutes |
| Latest full smart-store-manager run | 2,037 products analysed; 19 sections produced | 32 minutes 35 seconds |
These are separate stages. The windows include gaps between batches and retries; they are not a claim that the entire workflow finished within 24 hours. They show the scale of automated production available for a team to review and use.
Turning catalogue knowledge into a storefront
The latest full smart-store-manager run analysed 2,037 products and produced a 19-section design with 9 banners. All 19 section identifiers remain in the published homepage at the time of review, with subsequent banner changes.
The category audit also produced 26 naming suggestions and 9 merge suggestions. These are proposals for review, not evidence that every proposed change was applied. Together, the outputs reduce the work of assembling page materials, organising sections and briefing a separate designer.
Marketing moves into the execution workflow
Mollkom’s marketing workflow uses product knowledge to support strategy, creative production, scheduling and publishing through connected accounts. The saved Joolyoo outputs include 117 designs and 9 videos, alongside 58 Facebook and Instagram publication receipts.
The records distinguish ideas, drafts, ready assets and published items. Saved creative is not automatically counted as published content. For the merchant, the benefit is less repeated work transferring product details, preparing design briefs and assembling materials for each publishing cycle.
What does this mean for merchant costs?
We reviewed a newer sample in which completed outputs are linked to settled usage. For this study’s merchant-cost scenario, we apply twice the monetary equivalent of recorded usage. These are modelling estimates, not a change to Mollkom’s published tariffs.
Scroll the table horizontally to see all columns.
| Output | Reviewed sample | Assumed traditional cost | Modelled merchant cost with Mollkom |
|---|---|---|---|
| Product copy plus studio and lifestyle images | 2 completed products | SAR 307.50 | SAR 2.78 |
| Marketing design output | 14 outputs for 13 items | SAR 75 | SAR 3.80 |
| Standalone banner | 3 delivered banners | SAR 75 | SAR 1.63 |
The assumed traditional product cost comprises SAR 75 for studio photography, SAR 225 for lifestyle photography and SAR 7.50 for copy. These are stated planning assumptions. AI generation starts from existing product images and information; it is a different production process from a physical photo shoot. Inputs, required quality and review affect the cost of either route.
One combined budget: 1,000 products, 20 designs and 3 banners
To make the sample useful for planning, we apply its unit averages to a defined production bundle. This projection is separate from Joolyoo’s historical bill.
Scroll the table horizontally to see all columns.
| Production scope | Assumed traditional cost | Modelled merchant cost with Mollkom |
|---|---|---|
| 1,000 products, each with copy and two images | SAR 307,500 | SAR 2,782.06 |
| 20 marketing design outputs | SAR 1,500 | SAR 75.96 |
| 3 banners | SAR 225 | SAR 4.89 |
| Total | SAR 309,225 | SAR 2,862.91 |
Modelled difference for this defined production bundleSAR 306,362.09Using merchant costs set at twice the reference usage value.
The comparison indicates a substantial difference in the cost of producing these materials. It is not a total store operating budget or a claim of realised cash savings. Subscriptions, input preparation, review, paid-media spend, stock, shipping and physical operations sit outside the bundle. We have not invented a matching cost for video production or category management without corresponding billing samples.
More capacity to run the business
Lower production costs can make a broader catalogue and more frequent creative refreshes practical. At larger scale, the benefit compounds when product knowledge can serve the catalogue, storefront and campaign without being re-entered at every step.
This workflow belongs to Mollkom’s wider commerce environment, connecting marketing and customer interaction with orders, inventory, warehouses, point of sale, shipping and driver delivery. This study documents the Joolyoo workstreams we reviewed; the cost impact of other functions requires evidence from their own use.
Joolyoo demonstrates AI doing daily execution work: thousands of product images and content fields, a storefront, and marketing materials carried through to publication. That is the direction of Mollkom’s pilot: helping merchants and their teams accomplish more while reducing repeated work around each product.