An experienced real estate agent can often tell within the first ninety seconds of a call whether a lead is serious. There’s no script for it. It comes from years of phone calls and closed deals, the kind of instinct that builds up slowly and is never written down anywhere. That gap, between what an agent knows and what’s actually documented, is the problem now facing real estate teams trying to bring AI into their business.
A year ago, team leaders worried that AI simply couldn’t handle the human side of real estate: whether a computer-generated call would sound believable, whether clients would tolerate it, whether relationships could survive being automated. According to Parwaan Virk, founder of the AI startup Kyzo AI, that worry has mostly disappeared. The question agents and brokers ask him now isn’t whether AI works. It’s where to start.
That sounds like a simple matter of picking the right software. Virk argues it’s something deeper. With roughly 200 AI tools competing for attention in real estate, the real obstacle isn’t choosing one. It’s that most brokerages have never captured their own know-how in a form any of those tools could actually use.
The Real Problem
When an experienced agent qualifies a lead, they draw on years of accumulated judgment: what to ask, how to read hesitation, when to push and when to wait. None of that exists in a training manual or call script. It lives only in their head, shaped by hundreds of conversations that were never recorded or written down.
Without that knowledge written down somewhere, AI tools have nothing to learn from. A brokerage buys a tool expecting it to perform, only to find the tool has no understanding of how that specific firm operates, what its buyers care about, or how its agents have learned to close deals.
“Whatever tool we bring to the table, it won’t solve the problem unless we can get that out of their heads and into that AI,” Virk says.
Why Tools Aren’t Enough
When this happens, the tool underperforms, the team loses confidence in AI altogether, and the project gets shelved. The brokerage concludes that AI simply doesn’t work for real estate, when the actual problem was that no one translated the firm’s own expertise into instructions the AI could follow.
Virk describes this as a knowledge transfer problem, not a technology problem. Getting an AI to call leads for a real estate firm in Texas requires first understanding what that firm’s agents actually say on those calls, why they say it, and what outcome they’re aiming for. That takes time, recorded conversations, and repeated rounds of training, not just a software subscription.
The implication reaches beyond any single firm. As AI adoption speeds up, brokerages that skip this step will consistently fall behind those that take the time to do it. The gap between firms may come down less to which technology they bought and more to whether they did the work of making their own expertise understandable to a machine.
Too Many Choices
Beyond the knowledge gap, the sheer number of available tools is creating its own kind of paralysis. Virk says team leaders who are ready to act often freeze once they see their options. Which tools to buy, whether they need outside help, who can configure these systems for their specific needs: these are the questions he hears again and again from people willing to invest but unsure where to put their money.
Choosing a tool and implementing it well are two different problems. A brokerage can pick the right tool and still fail if it lacks the expertise to configure the AI around its own workflows. On the other hand, a firm with strong implementation know-how can often make an average tool perform well, because the real work, transferring its knowledge into the system, has been done correctly.
Most brokerages focus on which tool to buy, Virk says, when the implementation work is what actually determines results. His company’s approach is to study how a brokerage already operates day-to-day, then build the AI around those existing habits rather than asking the team to change how it works. “You don’t need to adapt to how the technology works,” he says. “We will adapt our technology to how you work in your current setup.”
Closing the Knowledge Gap
The fix for this problem isn’t a particular brand of software. It’s a process: studying how a brokerage actually works, recording and reviewing real agent calls, and translating that accumulated judgment into instructions an AI can follow before any tool goes live. Skipping this step is why so many AI rollouts stall. The technology was never missing the knowledge it needed; no one had captured that knowledge in a usable form yet.
Whether the broader industry adopts this more deliberate approach, treating the documentation of expertise as a required first step rather than an afterthought, may determine how much of the current wave of AI investment in real estate actually pays off. The firms willing to write down what their best people know before deploying new tools are likely to pull ahead of those chasing the technology itself. For brokerages sitting on years of expertise that has never left anyone’s head, the first move isn’t buying software. It’s figuring out what their best people actually know and putting it into a form a machine can use.
About the Expert: Parwaan Virk is founder of Kyzo AI, a bootstrapped proptech startup focused on lead response automation for real estate teams, operating across India, the UAE, and the United States. The company launched as a standalone product in June 2025 with a seven-person team.
This article is intended for informational purposes only and does not constitute legal, financial, or investment advice. The views and opinions expressed herein reflect those of the individuals quoted and do not represent an endorsement of any company, product, or service mentioned. Readers should conduct their own due diligence and consult qualified professionals before making any investment decisions.
