Design studyERP for a trading companyCustom ERP with WhatsApp and AI assistant

One system for orders, stock, invoices and customer messages, replacing a set of spreadsheets

A design study of an operations system for a UAE trading company: orders, stock by location, VAT invoices and a WhatsApp inbox in one place.

PlatformCustom ERP with WhatsApp and AI assistant
SectorERP for a trading company
ScopeWorkflow discovery, Data model
StatusDesign study
OpsDesk · Overview
The brief

Imagine a trading company running orders from WhatsApp, stock in Excel and invoices in a Word template. The brief we set ourselves: one place for orders, stock and invoices with 5% VAT handled correctly, customer messages arriving in the same system, and an assistant that answers routine questions but hands over to a person when it should.

What we built

A web-based operations system: order and customer records, stock by location, quotations and VAT invoices, payment tracking, a WhatsApp inbox linked to customer records, and an AI assistant that answers from the catalogue and stock data. Designed to demonstrate design.

OpsDesk · Invoice
OpsDesk · WhatsApp
maisonverre.ae
Decisions

Choices we made, and why.

  1. 01

    Model the real workflow before the screens

    We map how an order moves from enquiry to quotation, invoice, delivery and payment, and who touches it. The data model follows that flow, so the screens match how staff already think and training is short.

  2. 02

    Stock as movements, not a number

    Stock is the sum of receipts, sales and adjustments, each with a user and a date. A single editable quantity cannot be audited, and an audit trail is what lets a manager find where a count went wrong.

  3. 03

    VAT and invoices built in, with the accountant in the loop

    Invoices carry 5% VAT as separate lines with sequential numbering. Tax rules are confirmed with the client's accountant and, for e-invoicing, checked against current UAE requirements before we configure anything.

  4. 04

    WhatsApp through the official API

    Messages arrive via the WhatsApp Business Platform, which requires approved templates for business-initiated messages and a business account. We build on that rather than unofficial tools, which risk the number being blocked.

  5. 05

    The assistant answers from data and knows when to stop

    The AI assistant reads the catalogue and stock, answers price and availability questions, and hands off to a person for discounts, complaints or anything outside its data. It cites what it used, and staff can review every conversation.

  6. 06

    Roles and permissions from day one

    Sales, warehouse and finance see different things, and price or cost fields are hidden where they should be. Adding permissions after launch is harder than designing them in.

Planned timeline: Planned as a 10 to 14 week operations system build: workflow discovery in weeks 1 to 2, design in weeks 2 to 4, core modules in weeks 4 to 10, WhatsApp and assistant in weeks 9 to 12, data migration, training and launch after.

Next.jsTypeScriptPostgreSQLPrismaWhatsApp Business Platform (Cloud API)Anthropic Claude APIRedisDockerPlaywrightGit with GitHub

Designed and engineered by our studio.

Start a project

Your brand, built to this standard.

Tell us what you sell and when you want to launch.

Fixed quote in AED, in writing. Reply within one working day.

Get a fixed quoteWhatsApp