One of the leading halal brands in the Gulf was refreshing its packaging and catalog for a June launch across MENA. The traditional approach would have taken 2–4 months — an AI pipeline assembled the digital catalog in 3 weeks across two languages.
Sadia holds a significant share of the halal chicken market in the Gulf: ~17,000 points of sale, 111,000 monthly deliveries, and distribution through Carrefour, Lulu, Spinneys, Panda, and q-commerce (Talabat, Noon, Careem). In Q1 2026, the group is closing a deal with the Saudi PIF for $2.07 billion with an IPO horizon in 2027.
The brand is undergoing a packaging and SKU line refresh. By June 2026, new products arrive in MENA warehouses — and an updated catalog must appear on marketplaces by that window: new packaging, current food shots, accurate ingredient information, and claims icons (Halal, pre-cut, protein).
Listing content was outdated: old packaging was appearing on marketplaces, info images were assembled inconsistently, and lifestyle blocks were absent for the majority of SKUs. Launching without an updated catalog would have nullified the marketing impact of the rebrand.
The traditional path for 129 SKUs would have taken 2–4 months: studio brief, shooting each product, retouching, manual info-slide layout, EN and AR localization in separate cycles, sign-off on each slide with HQ. Per SKU: 9 slide types × 2 languages = ~1,140 unique images.
An AI pipeline assembles a digital catalog: data parsing and packaging OCR, generation of 9 slide types per SKU, bilingual EN/AR layout per brand guidelines.
The 24.online LLM parser reads an Excel file with characteristics and simultaneously reads the back of the packaging via OCR (ingredients, macros, storage conditions, some icons), forming a JSON for each product. Data is not entered manually — the pipeline collects it from sources the client already has.
The orchestrator expands the JSON product card, brand book, and references into detailed prompts for 9 content types: packaging angles, ingredient and macro infographics, instructions, hero slides, and product line group collages. One designer handles 10–16 SKUs per day instead of 2–3.
Both language versions are generated in a single pass: the JSON card contains both language blocks, and the typography module at the final stage overlays the text in the correct language with proper text direction — no second production cycle.
Image generation models reliably generate composition but distort fonts, so the visual is generated without typography, and a branded SF Pro font layer (Bold / Medium) is overlaid on top. The result: 100% brand book compliance and consistent typography across all 1,140 images.
The pipeline became a reusable asset rather than a one-off contract project: every assortment expansion, seasonal line, or rebrand runs through the same loop without repeat photo production.
For 24TTL, this is the first production case of 24.online in FMCG MENA — an industry where rebrands and seasonal updates are ongoing and the traditional "photo studio + design agency" model is becoming a bottleneck.
We got an updated catalog in a format we can further adapt for different platforms. The AI pipeline handles the June launch without lengthy PIM integration. The master versions are assembled according to the brand book in two languages, so we didn't have to do the work twice.
The metrics and results presented reflect outcomes of a specific project and depend on its initial conditions. They are provided for informational purposes only, do not constitute a public offer, and do not guarantee similar results in other projects. Supporting materials are available on request.
Generative AI rebuilds SEO content and visuals for product listings: copy, hero images, and rich content tuned to marketplace algorithms — all in a single pipeline.
An AI pipeline assembles lifestyle videos from product photos: decomposition into short artifact-free segments, parallel generation, human in the loop.
A browser agent cross-checks size charts in product listings against official brand websites: auto-search, Vision-based table extraction, and size system normalization.
No cases match the selected filters.
Tell us about your process — we will suggest an audit, pilot, or full deployment.