Cite Roselyne — French Real-Estate WhatsApp AI Bot
A French real-estate developer needed 24/7 multilingual lead qualification on WhatsApp without hiring more agents. SkynetLabs shipped a working MVP in 5 days at a $850 fixed bid (50/50 split).
The Problem
Roselyne (real-estate developer, Paris region) was generating qualified WhatsApp inquiries from her listings but losing them after-hours. Inbound questions came in French, English, and occasional Arabic. Her existing process was a single human reading messages on a phone and replying when she had time. Conversion was dropping because response time stretched past 12 hours during weekends and holidays. She did not want to add headcount, did not trust generic chatbots to handle property questions, and had a hard requirement that nothing leak commercial data to a public LLM.
The Stack
The Build
5-day delivery, 50% upfront, 50% on accept. Day 1 was the WhatsApp Business API setup and Meta template approval submission. Day 2 was the n8n VPS provision plus the lead-capture Supabase table. Day 3 wired the GPT-5.4 reply node with a tightly scoped system prompt covering property questions only, with a hard refusal on price negotiation and an automatic handoff trigger to Roselyne for any qualified buyer signal. Day 4 was QA across 30 sample messages in French and English. Day 5 was production cutover plus a 30-minute handoff call.
The Outcome
Bot is live and answering inbound WhatsApp inquiries in French and English within seconds. Roselyne now receives a Slack ping only when a message contains a qualified buyer signal (budget mentioned, viewing request, contact details), eliminating the after-hours reading load. Conversion data is still accumulating — the bot has been live under 30 days at the time of this writing — but qualitative feedback from the first cohort of leads has been positive. Pre-kickoff PDFs in French were sent for client onboarding, kickoff call confirmed.