nqrah

01 Chief complaint

Egyptian merchants lose money on orders that never arrive. Most refused cash-on-delivery orders aren’t a shipping problem. They’re a trust problem, and the store builders on the market weren’t built for it.

02 History

While studying medicine, I co-ran an e-commerce business with a family member, handling media buying and most of the operations. We passed EGP 500K in sales in a short time with strong margins, and I met the market up close: shipping companies, fake delivery attempts, customers who didn’t answer or didn’t know when their order would arrive.

03 Diagnosis

A small automation that tracked every Egypt Post shipment and messaged the customer at each step took our delivery rate to 97%. Merchants don’t need a prettier store. They need a whole system that works the way the market works, because how a store speaks, confirms and follows up decides whether a customer trusts it enough to pay at the door.

04 Key decisions

• No cards, no instalments

  The market pays at the door. Supporting card flows would have doubled the surface area for a minority of orders, so the platform goes deep on cash on delivery, deposits and wallets instead.

• AI composes, it never codes

  Atelier chooses and fills tested sections instead of generating free-form HTML. The output is always valid, accessible and safe for checkout.

• Cost is reserved before every call

  Every AI request reserves its estimated cost first, then settles. Monthly caps per store and a daily platform breaker make runaway bills impossible.

05 Treatment

• Atelier: AI designs a whole store, page or section from a description, using 31 tested section types, never free-form code, so it can never break checkout. Every change is previewed, publishing is the merchant’s call, and undo is always there.
• Cash on delivery at the core: deposits for risky orders, InstaPay and wallet receipt review, Fawry and bank transfer. No cards and no instalments, by design. Prices are calculated on the server and every stock movement lands in an append-only ledger.
• AI with a budget: product pages from a photo, translation, and assistants for customers and owners. Works with Claude, OpenAI, Gemini and others. Cost is reserved before every call, with a monthly cap per store and a daily breaker for the whole platform. No customer name, phone or address ever reaches a model.
• Social commerce inbox (in development): AI replies on WhatsApp, Instagram and Messenger from the real catalogue, policies and order status, with one human reviewing and every complaint escalated.
• Shipping, settlement and tax (in development): direct carrier integrations, automatic COD reconciliation, revenue versus net profit after shipping and returns, and e-invoicing for larger merchants.
• Trust and security: database-level isolation between stores, an encrypted key vault per store, and data the merchant owns and can export at any time.

06 Outcome

The public beta opened on 10 October 2026, free for merchants. Store setup, catalogue, orders, cash on delivery, shipping and the dashboard are all live. nqrah is self-funded and was accepted into startup programmes from AWS, Microsoft and Google Cloud.

ready-made themes

58

tested section types

31

business types

25

governorates covered

27

07 Gallery

08 Follow-up

Next come the social inbox and carrier integrations, then e-invoicing for larger merchants, which is why the company registration in Egypt is under way. Full launch is planned for early 2027: Egypt first, then Saudi Arabia and the Gulf. The business model follows the merchant’s success: nqrah only earns when an order is actually delivered.