Aura tracks the spend.
Wajha causes the spend.
AI-powered shopping discovery across Alshaya's Kuwait portfolio — fashion + retail + food. Real catalog data, real embeddings, real cross-brand search.
English text search
“white t-shirt under 8 KWD” — cross-brand AI search across the Alshaya catalogue, in English
Arabic text search
“فستان” (dress) — multilingual embeddings, no translation layer, Arabic queries handled natively
Visual similarity
Tap a product → find visually similar across H&M / Foot Locker / Mothercare / BBW
Complete the look
Pick an outfit → AI bundles matching items across all the OTHER Alshaya brands
Price ladder
“love this but it’s outside my budget” — find visually similar at discounted price
Fashion vertical — what's in the demo
673 real SKUs scraped from Alshaya-operated KW storefronts on the Adobe AEM Edge platform. Real CLIP (512-dim) image embeddings, real Cohere multilingual (1024-dim) text embeddings in Supabase pgvector. Cross-brand kNN via Postgres HNSW.
What this demo is — and is not
- ✓ Real catalog data, real embeddings, real cross-brand kNN
- ✓ Stage 2 cart + Concierge checkout walkthrough (mocked — no payment, no real orders)
- ! Checkout writes nothing to brand sites; success page is illustrative
- ! Stage 1 is the pilot ask; Stages 2 and 3 are visible here for context only