Three salespeople on one number, two prices for the same customer

A conversation belongs to one salesperson, and the follow-up leaves in time.

A run through the desk, demo data: a thread taken by one salesperson under everyone’s eyes, the thread she holds, the follow-ups, the catalogue, and the supervisor’s day.
The shared queue, demo data: a thread just taken by a salesperson, the others free or held by a colleague, and the count of threads nobody is on.
A customer thread read by the supervisor, demo data: the salesperson who holds it, the 24 hour reply window, and the piece under negotiation, flagged to follow up.
The open negotiations, demo data: announced price, salesperson, deadline, and the alarm on the one waiting for more than 20 hours.
The shared catalogue, demo data: the available count deduced from sales and reservations, and who is already negotiating each piece.
The supervisor view, demo data: sales of the day, threads waiting for a reply, negotiations in alarm, results per desk and the activity log.

The problem

An artisan jewellery boutique in Morocco sells from a single WhatsApp number. Three salespeople answer from their own devices: two of them send the same customer two different prices, a piece reserved by one is sold by another, and a customer who says « I will tell you tomorrow » is never followed up.

What I built

An internal console installed at the shop. A shared queue gives one salesperson an exclusive lock on a conversation, with an expiry so a closed tab never freezes a thread. Catalogue and stock, pending negotiations that escalate on their own, orders with deposits, and a supervisor view of the day. Text, images and voice notes go out through the WhatsApp provider, voice re-encoded server side.

Technical details

Exclusive lock per conversation, held by a lease that expires
Follow-up reminders capped by the WhatsApp 24 hour service window
Voice notes re-encoded server side to the format WhatsApp accepts
Live updates over a server-sent event stream, filtered by role and scope

Stack

React 19
Fastify 5
SQLite
SSE
Zod
Playwright

The result

The desk is in service and measured on real traffic. 310 unit tests pass, and 67 end-to-end scenarios cover concurrent pickup, stock, escalation and a provider outage.

A similar project in mind?

Tell me about your business and when you need it. You get a proposal by email.

I want this system

More projects

ERP & CRM · ProximaCustomers, stock and invoices live in spreadsheetsView the project