Skip to main content
Live in Production

A Kuwait pharmacy chain,
14 branches, one system.

A 14-branch pharmacy operation running on WhatsApp groups and spreadsheets, replaced end-to-end with production-grade automation. The system is live in production, handling around 32 orders a day at the busiest branch, with the same architecture ready to scale. We frame it here as evidence of what our stack can run day to day.

The engagement
14-branch pharmacy chain, Kuwait
A Kuwait pharmacy group running online, delivery, and in-store channels across 14 branches. We operate the delivery and automation stack that ties those channels together, in production every day.
723
Drivers managed
60-90
Days to launch
By the numbers

Real production. Real verifiable numbers.

14
Branches synced
723
Drivers managed
11
Production workflows
488
n8n nodes shipped
$114K
Saved a year
~32
Orders / day, busiest branch
What we built

Six systems, one operational backbone.

The deliverables sit across delivery, ERP, automation, customer comms, intelligence, and e-commerce. Each is wired into the others; the operational layer is the integration.

01

The delivery app

Delivery fleet management platform with real-time GPS tracking, auto-assignment by proximity and queue, driver scoring, and route-aware batching.

02

Phoenix ERP Integration

Bidirectional webhook sync between Phoenix and the fleet system. Inbound order events (insert / update / delete) plus outbound driver assignment and delivery confirmation. Idempotent on billguid.

03

N8N Workflow Suite

11 production workflows (488 nodes) covering order processing, sales automation, customer comms, multi-branch data sync, and inventory bridges.

04

WhatsApp Business Gateway

Automated customer messaging via Evolution API. Order confirmations, delivery updates, customer-service triage, plus invoice OCR and command-channel integration.

05

Operations Dashboard

Real-time operational intelligence with KPI tracking, role-based access, fleet stats, and live driver map. Sub-second dashboard latency on 14-branch traffic.

06

Multi-Platform E-Commerce

Shopify (8 pharmacies), Talabat, and Deliveroo integrations. Branch-level stock sync, price-after-discount logic, expired and damaged exclusion.

Before and after

From WhatsApp chaos to operational discipline.

Before

Dispatchers calling drivers on WhatsApp, one order at a time

After

Auto-assignment scores by pickup proximity (60%) and queue position (40%), batches 5-minute windows for same-area orders

Before

Manual reconciliation between Phoenix, branches, and customer chats

After

Phoenix webhooks fire from a dedicated server, idempotent on billguid, with order events landing in the delivery app inside seconds

Before

Branch managers chasing inventory across WhatsApp groups and Excel

After

Ending Inventory endpoint pulled into Shopify, with package SKUs override-list and expired/damaged branch exclusion

Before

End-of-month manual close, multi-day reconciliation

After

Real-time finance dashboard reading the operational stack; close converges in hours, not weeks

Architecture

A production stack, not a workflow demo.

Every box below runs in production. Each is wired to the next, and each has been live in production and documented from day one.

Phoenix ERP(client-managed)webhookFastAPI Backendasync, idempotent on billguidCeleryPostGISRedisMinIOn8n suite11 workflows, 488 nodesReact Admindispatch + analyticsFlutter Driver AppGPS, accept, deliverWhatsApp GatewayEvolution API + LLMINTEGRATION PLANEShopify (8 pharmacies)TalabatDeliverooKimi K2.5 (OCR)Telegram botPostGIS geofencing

Every box runs production traffic today. The diagram is the system, not a target state.

Outcome

What changed for the business.

Dispatch no longer depends on people staring at WhatsApp groups. Auto-assignment routes the next order to the right driver based on pickup proximity, queue position, and motorcycle-vs-car economics. Order batches are consolidated in 5-minute windows when the same area and same pharmacy align.

The Phoenix ERP and the delivery app now share state through a dedicated webhook server. When a pharmacist approves an order in Phoenix, it lands inside the delivery app in seconds, idempotent on billguid so retries do not double-bill.

Across the operation, this work removes roughly $114K of cost a year that used to be carried by dispatchers, manual reconciliation, and platform-specific staff. The systems run continuously, scale linearly to additional branches, and are live in production today.

Take this with you

Get the 12-page case study, in your inbox.

Full architecture, before / after KPIs, and lessons learned. Forwardable to your CFO or COO before our first call.