Jenny – An AI agent that answers WhatsApp customers

Jenny replies to WhatsApp customers by herself, gives prices, books apartments and cars, and hands hard cases to the owner.

Private deployment
A WhatsApp chat with Jenny; the customer's voice note appears as text.

Overview

Jenny is an AI agent for a business in Dubai that rents furnished apartments and luxury cars. She chats with customers on WhatsApp in their own language, and she understands voice notes too. The owner follows every chat from a web dashboard and can take over at any time.

The challenge

What was not working

The business is run by one owner, who had to answer every WhatsApp message alone. Customers write in Arabic, English and other languages, and ask about prices, free dates and bookings. An AI helper that might guess a wrong price could not be trusted.

The solution

What I built

Jenny reads each message, finds the right facts and replies. She has 14 tools, such as checking free dates, giving a quote and confirming a booking. Prices come from fixed rules in the code, not from the AI, and every number in a reply is checked against real data. Complaints and requests for a person go to the owner, and Jenny steps back when the owner takes over. The dashboard shows the inbox, leads, bookings, a calendar and what Jenny knows, and it updates live.

Mobile app

In your pocket

A WhatsApp chat on the phone: voice note, price, and handover.
A WhatsApp chat on the phone: voice note, price, and handover.
The home page: what needs you right now.
The home page: what needs you right now.
All customer chats in one list.
All customer chats in one list.
Today's and tomorrow's viewings, check-ins and car deliveries.
Today's and tomorrow's viewings, check-ins and car deliveries.
Every customer with their stage and price.
Every customer with their stage and price.

Stack

06 parts

The tools behind it, from the screens people use to the services that store data, think and send messages.

NestJS
React
OpenAI
Evolution API (WhatsApp)
PostgreSQL
Qdrant
Screens

Inside the product

01 / 06
01A WhatsApp chat with Jenny; the customer's voice note appears as text.
02Home page: what needs the owner today and what Jenny handled alone.
03See why Jenny gave each answer, and which replies need checking.
04The week calendar: apartment viewings, check-ins and car deliveries.
05All leads in one list, with their stage and the price quoted.
06What Jenny knows: rules, common questions, apartments and cars.
Next case study · 05 · Business SystemLahloba – One system to run a home-cleaning companyOpen case study