All projects
Retail

A WhatsApp agent that sells and answers for a fashion store

Based on projects we've delivered. Figures are approximate; client names are withheld for privacy.

The problem

We spent the whole day answering the same questions: is this size in stock, how much is delivery? Night messages got answered in the morning, after the customer had already bought elsewhere.

Before and after

Before

Two staff spent most of their day replying to messages.

After

They focus on special orders and customer follow-up.

What we built

  1. 1

    Understands messages in Darija, Arabic or French and checks stock live.

  2. 2

    Reserves the item and confirms the order — cash on delivery or a payment link.

  3. 3

    Hands special cases to a team member with a ready summary of the conversation.

From a WhatsApp message to a confirmed order: the agent understands the request, checks stock and sends the payment link.

Results

of repetitive questions answered automatically
70%of repetitive questions answered automatically
first reply, any hour
Under 10 sfirst reply, any hour
more confirmed orders from evening messages
+25%more confirmed orders from evening messages

Measured over 30 days after launch, compared with the previous month.

Timeline

10 days

From first call to production

Under the hood
  • WhatsApp
  • n8n
  • PostgreSQL
  • OpenAI / Claude

Next project

Distribution and logistics

A self-running order pipeline for a distribution company

How we built it