AI Broker for Luxury Real Estate

Connect With Us

    Problem Statment

    Pinnacle Estates was receiving 60 to 80 WhatsApp inquiries a day. 31% of them were arriving outside business hours. Every single one was going unanswered until the next morning.

    A top Dubai luxury agency losing serious buyers to silence

    Pinnacle Estates had built a strong position in Dubai’s luxury market, 14 experienced agents, a portfolio across Palm Jumeirah, Emirates Hills, and Downtown Dubai, and an average deal size of AED 8.5 million. In 2024 they closed 68 transactions. By early 2025 their pipeline suggested they should have been closing more.

    The gap was in the first hour of every inquiry. Over 80% of Dubai luxury property buyers now make their first contact via WhatsApp, not a phone call, not a form, a message. Many of Pinnacle’s buyers were international: investors from the UK, India, Russia, and GCC countries researching during their own business hours, which often meant 10 PM to 2 AM Dubai time.

    When those messages arrived at night, they waited. Sometimes until 7 AM, sometimes until a junior agent noticed them at 9. By then, a buyer who had sent the same message to three agencies had often already booked a viewing with whoever responded first.

    Omar Al Zaabi, Pinnacle’s managing director, shared their lead data at our first meeting: “We looked back at Q4 2024. We had 340 WhatsApp inquiries in one month. We responded to 229 of them within the same day. The other 111, we have no idea what happened to them.”

    What the agency was losing before Korvax came in

    What Korvax built for Pinnacle Estates

    We started by auditing 90 days of Pinnacle’s WhatsApp inquiry threads, 847 conversations in total. We identified the 12 most common buyer questions, the 6 qualification signals that separated serious buyers from window shoppers, and the points in a conversation where handover to a human agent made the most sense.

    We then built an AI broker deployed natively through WhatsApp Business API, trained on Pinnacle’s full live inventory, brand communication standards, and qualification criteria. The system knows every active listing,  floor plans, views, payment plans, developer background, service charges, handover timelines, and comparable recent transactions.

    The qualification flow was designed to feel like a concierge, not a form. The AI asks natural questions that guide the buyer toward properties matching their actual situation, budget, timeline, buyer status, preferred location, without ever feeling like an interrogation.

    When a buyer is qualified, the AI checks agent calendar availability in real time and proposes specific viewing slots. The agent receives a calendar booking with the buyer’s full conversation history attached, budget confirmed, property interest noted, nationality and visa status logged. Everything the agent needs to walk into that viewing with context.

    Build time was 5 weeks. The AI broker went live on 14 February 2025.

    What the AI broker manages on behalf of Pinnacle's team today

    What changed at Pinnacle Estates after go-live

    The overnight gap closed immediately. From the first week, every inquiry received a response within 90 seconds regardless of time. Agents arriving at the office found their morning not full of unanswered messages, but full of booked viewings with pre-qualified buyers and structured profiles already in the CRM.

    The quality of those viewings improved as well. Because every buyer had been through the qualification flow before meeting an agent, the conversations started from a higher point, budget already confirmed, property shortlist already narrowed, logistics already arranged.

    Omar’s comment at the 45-day review: “We had a buyer from London who messaged us at midnight on a Tuesday. By the time he woke up the next morning, he had a viewing confirmed for Saturday and a shortlist of 4 properties waiting in his inbox. He told us at the viewing that was the reason he chose us over the two other agencies he had contacted.”

    Results at 90 days post-deployment