While I’m not an expert on AI (even the designers of AI don’t fully understand it), I’m growing weary of the AI hype cycle and thought I could contribute somewhat after seeing this awesome research piece by Luis Gariçano, Jin Li and Yanhui Wu shared by Apollo’s Chief Economist Torsten Slok (he always provides great real estate insights) that I found really fascinating: The Mess Is Your Moat. It provides a refreshing way to think about the AI revolution during the current AI hype cycle by brokerage firms about the future of real estate agents.
The research behind their book suggests a saner way to think about the AI revolution in progress that is coming to the real estate brokerage industry. I try to suss it out here in simple terms.
It’s not complex, relationship-heavy work that’s most at risk from AI, but routine, transactional, “clean” tasks. As tasks become messier, the cost of having AI operate without human oversight rises faster than AI’s capability does, so that mess effectively becomes a competitive moat protecting the job from replacement. I think of the negotiating skills of an experienced real estate agent and the negotiation buffer they provide between the parties as something that won’t be replaced by AI anytime soon.
This chart below illustrates something I wish I had in my back pocket as a teenager when my mom insisted I clean my room.
The chart traces the point where AI shifts from complementing a human to fully replacing them, something that seems to terrify modern workers in the current hype cycle.
What skills remain scarce after AI is implemented?
Their insights are part of a new book that is available. I just picked up a copy but haven’t finished it yet; I just relied on their interesting website for part of this post.
After AI absorbs repetitive tasks while the scarce skills tend to shift from execution to judgment, context and accountability. The bottleneck is no longer “can this be done?” but “who can decide what should be done, interpret the output and own the consequences?”
Deloitte talks about this as a broken skills ladder:
As AI takes on more entry-level tasks, organizations risk losing the pathways that turn novices into experts.
Taking away the mundane tasks will make more complex skill categories more valuable with more automation.
AI strips out the routine real estate brokerage layer
In real estate brokerage, AI will likely fully out the routine brokerage layer: lead qualification, listing copy, basic CMAs, scheduling, document assembly and first-pass market research. What becomes scarce (and more valuable) is the human work that still wins listings and closes deals: pricing judgment, negotiation, local context, client counseling and broker supervision.
“While not yet reducing headcount, AI is expected to strengthen operating platforms and enhance capacity across teams.”
The scarcest skills shift in messy situations is where the answer is not in the data. I told my grad students that a huge part of their value will be finding proxies to make judgments on a development plan. In real estate brokerage, that effort includes reading buyer and seller motivations, setting strategy in multiple-offers or price-cut scenarios, timing and knowing when a transaction is in trouble.
AI should expose weak skills first
AI will probably widen the gap between top producers and everyone else. AI should raise the efficiency of the average agent, but it also exposes weaker competence faster, so firms may need fewer people handling routine work and pay a lot more attention to people with messy skills who can convert complex client situations into closings.
For a real estate brokerage firm to find a winning AI model, I think AI investment will result in fewer entry-level agents and more towards advisors who can leverage AI. That is, firms will invest more in local market expertise, negotiation training, compliance oversight and client relationship management, while using AI for marketing, admin and research.
The biggest risk to the industry will be figuring out where the stronger advisors are going to come from since dependency on entry-level agents and their training is probably going to be reduced.
Final thoughts
Commissions rebounded after the NAR settlement rather than collapsing, suggesting the market already pays for the messy work agents do. Megateams may end up solving the “broken skills ladder” problem, with juniors handling AI-assisted admin while a senior agent owns the client relationship and passes down real experience. And in markets more exposed to rising climate and insurance complexity, it adds a fresh layer of “mess” that AI can’t easily untangle, likely insulating agents even more than their counterparts in more commoditized suburban markets.
But anyway you look at it, it’s still a mess.
The actual final thought — Something to think about.
Leave a comment