What happens when one of the real estate industry’s early AI pioneers says artificial intelligence may be moving too fast for its own good?
That’s one of the most interesting moments in my latest Inside AI conversation with Dominik Pogorzelski, President, MLS at Restb.ai. His prediction about the future of agentic AI is remarkably candid:
“I think it’ll get crazy before it settles down.”
Coming from someone who has spent nearly a decade helping bring artificial intelligence into real estate, that observation deserves our attention.
Restb.ai is marking an unusual milestone in the world of AI: it’s in its 10th year of bringing AI to real estate. Today, more than 1 million real estate agents across North America have access to its technology through their MLS, often without realizing they’re using it.
The company processes more than 2 billion images monthly, helping MLSs automate compliance, improve listing information and make property searches smarter.
But what I found especially fascinating about our conversation wasn’t simply how far Restb.ai has come. It was Dominik’s perspective on where AI is taking us next, what the industry still doesn’t understand about it and why the greatest advances may introduce entirely new problems.
The listing mistake hiding in plain sight
Here’s a question every real estate agent should consider: What if your listing is missing the very feature that would attract its ideal buyer?
One of my favorite moments came when Dominik shared research Restb.ai conducted on property listings. Using computer vision, the company compared what was visible in listing photographs with what agents had entered in the listing data. The findings were startling.
Restb.ai discovered that one in four listings where a water view was visible in the photos failed to mention it in the listing information.
Think about that. If a buyer searches specifically for a home with a water view, but the listing doesn’t identify it, that property might never appear in the results. The home could be a perfect match, yet the buyer may never know it exists.
Dominik shared another example involving outdoor kitchens that was even more surprising. I won’t give away that finding, but it raises an important question: How many potential buyers never see a property because its features weren’t properly identified?
As Dominik put it, missing those details is “money left on the table.” That’s a compelling use of AI because it isn’t about generating prettier listing descriptions. It’s about identifying information that could make a difference in how buyers discover homes.
A decade in AI is practically ancient history
What makes Dominik’s perspective particularly interesting is that he was there when Restb.ai was just a small startup in Barcelona, Spain. His journey with the company, including why he left and eventually returned, offers a fascinating look at how one of real estate’s most widely used AI technologies developed long before ChatGPT became a household name.
At a time when few real estate professionals were talking about AI, Restb.ai was developing computer vision technology that could interpret and understand what appeared in property photographs.
Today, its technology can recognize more than 450 property details from photographs. But Dominik explained that the real breakthrough isn’t simply recognizing what’s in an individual image. It’s understanding the property as a whole. That distinction is far more important than it sounds. Dominik gives some terrific examples of why analyzing individual photos without understanding their context can lead to mistakes.
He also reminded me how dangerous it can be for an agent to judge their experience with Restb.ai if it’s based on a year or two ago. “What was true 18 months ago is a whole other world today,” he told me.
That’s a message I wish every agent who tried AI once, got a disappointing result and gave up would hear. The technology is evolving so rapidly that yesterday’s limitations may no longer apply.
MCP isn’t a magic wand
Our conversation also ventured into territory every MLS executive and real estate technology leader should be watching closely. With AI advancing so quickly, there’s enormous pressure to connect existing data and systems to the newest AI tools. Model Context Protocol, or MCP, has become part of that conversation.
MCP provides a standardized way for AI applications to connect with outside tools and information sources. That’s appealing in real estate, where data is scattered across MLS platforms, brokerage systems, public records and technology vendors.
But Dominik pushed back against the idea that simply implementing an MCP server solves the industry’s AI integration challenges. “MCP servers are good for some experiences, but…it doesn’t solve everything,” he noted.
His point is that technology decisions still require careful thought about data standards, information quality and the experience an organization wants to deliver.
The AI race needs brakes and an accelerator
As someone who teaches AI safety and best practices to real estate professionals, I was particularly interested in Dominik’s thinking about agentic AI. We’re moving from AI that answers questions and creates content to AI that can make decisions and take actions on our behalf.
That’s a significant change. Think about the difference between asking AI to draft an email and allowing it to decide who should receive it, when it should be sent and what information it should include. The first keeps a human in control. The second introduces an entirely different level of risk.
I’ve been warning agents about giving autonomous AI access to sensitive client information, financial accounts, communications and transaction systems without adequate safeguards. Dominik shares those concerns, especially as AI moves closer to transactions and personal information.
What struck me was how he described the challenge facing companies racing to adopt this technology: “We need to slow down fast,” he said.
That may sound contradictory, but it perfectly captures the problem. The industry cannot afford to ignore innovation, yet moving too quickly without proper security, governance and human oversight could create serious consequences. AI can do good…and bad. Dominik offers a thoughtful perspective on why getting that balance right may be one of the biggest challenges facing AI.
Could AI make real estate more human?
Near the end of our conversation, I asked Dominik where he sees all this heading over the next few years. His answer surprised me, particularly after our discussion about the potential risks of agentic AI.
Rather than predicting a world where technology replaces more human interaction, he sees the possibility of something quite different. “I think the analog is where humanity lives,” he said.
I love that observation. Think about what makes a great real estate agent valuable. It’s not just access to listing information or the ability to complete paperwork. It’s judgment, trust, empathy and understanding what matters to someone making one of the biggest financial decisions of their life.
AI can improve how agents work, but that doesn’t necessarily mean consumers want less human interaction. In fact, the more automated our digital experiences become, the more valuable a genuine personal relationship may feel. When anyone can generate polished communications or professional-looking marketing materials, those capabilities alone no longer differentiate a great agent.
I’ve always believed the best use of AI is to make agents more effective, not less human. If technology takes repetitive work off an agent’s plate, that should create more opportunities to spend time with clients and deliver personal service no chatbot can replicate.
Dominik’s vision of how technological progress could bring us back toward more authentic human connections is one of my favorite parts of the interview. And it reinforces something I’ve been teaching for years: The real opportunity with AI isn’t to replace the human. It’s to make the human better.
There’s much more to this conversation, including surprising findings about listing data, the evolution of computer vision, AI integration challenges and why agentic AI could be both extraordinary and unsettling.
Watch Episode 10 of Inside AI featuring The REAL AI Guy for my full conversation with Dominik Pogorzelski, President, MLS at Restb.ai.
# # #
Note: Kevin Hawkins, a partner with WAV Group, is editor and co-founder of REAL AI, real estate’s No. 1 weekly AI newsletter and companion podcast, and host of the YouTube series Inside AI featuring The REAL AI Guy. Kevin is the Amazon bestselling author of The REAL AI Guide for Real Estate Agents. Known throughout the industry as The REAL AI Guy, he teaches real estate professionals across the U.S. and Canada how to use AI effectively and safely.
The post Inside AI – Episode 10 with Dominik Pogorzelski: Is AI Moving Too Fast for Real Estate? appeared first on WAV Group Consulting.
Leave a Reply