Phones were ringing constantly, calls were getting missed, and brokers were waiting too long for simple answers, all while the business was trying to grow.
The Problem
Our client, a fast-growing real estate brokerage in the United States with around 13,600 active brokers, had a support team that simply couldn’t keep up. Thousands of incoming calls a day led to missed calls, slow follow-ups, and inconsistent answers depending on who answered. Onboarding new brokers was slow and manual. They needed a way to handle high call volumes accurately, day or night, without adding more staff every time they grew.
What We Built
We built a voice AI system that answers and makes calls on the brokerage’s behalf, sounding natural enough that brokers and prospects feel like they’re talking to a real person.
The AI can both make outbound calls (reminders, scheduled check-ins, follow-ups) and answer inbound calls (questions, appointment bookings, broker support), automatically choosing the right one based on the time of day and the situation. Underneath it, we built a knowledge system, think of it like giving the AI an instant, perfect memory of every company policy, FAQ, and document, so it always answers questions accurately and consistently, instead of guessing or going off-script.
The system plugs directly into the brokerage’s existing CRM and scheduling tools, so when the AI books an appointment or talks to a broker, that information appears immediately where the team already works, without manually copying data between systems. We also built in a feedback loop that lets the AI improve its own phrasing and conversation flow over time, based on what actually leads to good outcomes.
The Result
- Onboarding productivity improved by 61% in the first month
- Missed calls dropped sharply, since the AI can handle hundreds of calls at once with zero wait time
- Prospective brokers got instant, clear answers and could book appointments immediately, instead of waiting on hold
- Existing brokers saved time, too, since routine questions got answered automatically instead of needing a person
- The brokerage now supports around 13,600 active brokers in the United States on a system built to keep scaling without adding headcount
The Honest Bit
The hardest part of this project wasn’t the logic; it was the voice itself. Voice AI was still a genuinely new field when we built this, and our first attempts sounded exactly like you’d expect from an early voice model: robotic, with awkward pauses and unnatural transitions between sentences. The moment a caller can tell they’re talking to a machine, trust drops, and so does the whole point of the system.
We tested across several voice providers, including Hume AI, ElevenLabs, Vapi AI, and Retell AI, before settling on ElevenLabs, and even then, it took real work to find the right voice for this specific client. We layered our own processing logic on top of the raw voice output, tuning pacing, pauses, and emotional tone until callers genuinely couldn’t tell whether they were speaking to a person or an AI. It took longer than we initially expected, but getting that human feel right turned out to be the difference between a system people tolerated and one they actually trusted.
Book your AI strategy call
Please schedule a AI consultation call. We’ll discuss your business needs and explore how AI can enhance your operations and drive growth.
