Use case · Live pilot

AI-first Retail POS — proof from RetailOS @ Mr. Donut.

A small bakery in Cairo used to track everything on paper. We deployed RetailOS — multi-tenant POS + inventory + customer engagement — and wired the daily summary straight into the owner's Telegram. The result: zero-login operator experience, live numbers, auditable receipts.

The problem

Every small retailer in Egypt we've met has the same stack: a WhatsApp, a book of customers, paper receipts, and an end-of-day pile that someone reconciles on Thursday. Western POS vendors show up with Shopify + Stripe + English UIs. None of that matches the ground truth.

The owner doesn't want a dashboard. The owner wants to know, before bed, whether today beat yesterday. That's a Telegram message, not a web app.

What we shipped

  • Arabic-first POS (RTL, Arabic digits, colloquial copy).
  • Real-time inventory with rolling demand forecasting.
  • Customer engagement workflows over Telegram + WhatsApp.
  • Multi-tenant architecture — one RetailOS, many shops.
  • Per-product tax + returns/RMA shipped in Apr 2026 audit sprint.
  • Owner gets the daily summary as a Telegram message.

Numbers that matter

Live
Daily operations at Mr. Donut
18 PRs
Shipped in one week in Apr 2026
EG£500/mo
Pilot price — fully operational

FAQ

How does this work for a shop that still uses paper receipts?

We replace the paper with a tablet-first UI. The owner sees the day's numbers on Telegram without logging in. That's the whole design goal — zero-login operator experience.

What's "AI-first" about it?

The inventory forecasting uses a rolling demand model, the admin summary is LLM-generated from raw events, and we can triage product categorization from photos. None of those are sidebar features — they're the core loop.

What do you cost for a single-location retailer?

Pilot: $500–$2,000. Custom AI extensions on top of RetailOS: $2,000–$5,000. If you're a multi-location chain we'll scope differently.

Is the UI in Arabic?

Yes. Arabic-first, not Arabic-translated. Digits, RTL, colloquial vs MSA — all handled at the component level.

Related:Product: RetailOS →Service: MENA automation →Service: AI agent development →

Running a retail shop and want the same setup?

20-min fit call. Bring your current workflow and one metric you want to know without logging in.