I'm Joni Freeman, PharmD, Principal Broker and Founder of JL Marsaw & Co Realtors in Frisco, Texas. I've been licensed since 2015, and I hold broker licenses in both Texas and Georgia. I also sit on the Texas REALTORS® Board of Directors and chair the MetroTex Young Professionals Network, which means I'm often watching industry-shaking stories before they reach mainstream coverage.
This one went mainstream not too long algo.
CNBC ran a story recently from real estate reporter Diana Olick with a headline that should stop every buyer, seller, and investor in their tracks: "How AI May Be Messing With Home Prices." At the center of it: celebrity broker and real estate CEO Ryan Serhant disclosed that ChatGPT almost blew up a $50 million deal.
Let that sink in.
A professional at the top of this industry, surrounded by experienced agents and advisors, nearly lost a $50 million transaction because an AI tool produced information that was wrong or incomplete or simply not contextualized for that specific market, property, and moment.
If it can happen at $50 million, it is happening right now in the $500K, $800K, and $1.2M range across Dallas-Fort Worth. The dollar amount is different. The risk is not.
Why Buyers and Sellers Are Trusting AI With Decisions It Isn't Equipped to Make
AI tools like ChatGPT, Perplexity, and built-in valuation estimators on real estate portals are genuinely useful for certain things. They can aggregate public data fast. They can pull comp summaries, neighborhood history, general market trends. For someone doing early-stage research, that has real value.
But there is a significant gap between "useful research tool" and "reliable basis for a pricing or offer decision."
Here's where AI breaks down in real estate:
It works from historical data. The DFW market moves faster than most AI models update. A comp pulled from six months ago in a shifting rate environment can lead a buyer to overprice an offer or underprice it and lose the deal entirely.
It doesn't know what isn't listed. An agent who has been active in a specific zip code knows which sellers are motivated, which properties have had deals fall through, which HOAs are financially strained, and which developments have pending permits that could affect value. None of that is in a public database.
It may not tell you the unflattering parts. The CNBC article notes that AI might not be inclined to surface the negatives. That is not a small caveat. If you are making a purchase decision and your information source is optimistically skewed by design, you are negotiating from a position that isn't real.
What Ryan Serhant's $50 Million Near-Miss Actually Reveals
Serhant didn't share the specifics of exactly what went wrong, but the fact that he's talking about it publicly tells me something: this is a pattern he's seeing across the industry, not a one-time anomaly.
What I believe happened and what I'm seeing echoes of in DFW transactions right now, is that AI produced a valuation or market read that a party used to anchor their position. When that anchor turned out to be wrong, it created conflict that nearly killed the deal.
In a $50 million transaction, there are enough attorneys and advisors in the room to course-correct before it collapses. In a $650,000 purchase, you may not have that safety net. The buyer or seller who anchored on bad AI data may dig in, lose trust in the process, or walk and never know why.
What AI Can Do and What Your Agent Does That AI Cannot
I want to be clear: I'm not telling you to ignore AI tools. I use data-driven analysis in every transaction. My background in pharmacy, where precision and evidence-based decision-making aren't optional, has shaped how I approach real estate pricing, negotiation strategy, and market reads.
The difference is knowing what the data means in context.
AI can tell you what sold. I can tell you why it sold for that number, whether that outcome will repeat in current conditions, and what the buyer and seller on that transaction had to give up to get there.
AI can tell you the days on market. I can tell you why it sat, was it overpriced, was there a disclosure issue, was the seller unrealistic, or was it just a slow period for that specific floor plan in that specific micro-market?
AI can generate a number. I can tell you whether that number will hold at the appraisal, survive the inspection negotiation, and actually close.
The expertise is not in the data. It's in the interpretation.
What DFW Buyers and Sellers Should Do Right Now
If you are preparing to buy or sell in the Dallas-Fort Worth market or in Houston or Atlanta, where I'm also licensed; here's my honest guidance:
Use AI tools for general orientation. They're fine for getting a feel for an area, understanding broad price ranges, or generating questions to ask your agent.
Do not use AI to set your offer price, establish your listing price, or make a decision about whether to accept or counter an offer. Those decisions need to be grounded in real-time, hyperlocal intelligence that an AI tool is not able to provide.
Work with an agent who can show you their reasoning, not just their number. Any agent can give you a price. An agent who walks you through the logic, the comps, the market timing, and the risk factors is the one protecting your investment.
The Broader Takeaway for the Real Estate Market
What the CNBC story and the Serhant disclosure signal is something I've been watching develop for over a year: AI is entering real estate transactions not as a support tool for agents, but as a direct decision-making input for buyers and sellers who don't know what it's missing.
That is a market risk. Mispriced offers, inflated or deflated listing prices, and deals that collapse because parties anchored on AI-generated data, these are not hypothetical scenarios. They are happening now.
The agents, brokers, and clients who understand this early have an advantage. The ones who don't are going to find out the hard way.
I've built JL Marsaw & Co on the principle that clients deserve to understand exactly what's happening in their transaction, not an algorithm's best guess. Real intelligence. Real advocacy. Every time.
If you're navigating a purchase or sale in DFW and you want a second opinion on what you're being told, I'd welcome that conversation.
214-770-7762 | [email protected] | www.jlmarsaw.com
Joni Freeman, PharmD, is the Principal Broker and Founder of JL Marsaw & Co Realtors, based in Frisco, Texas. Licensed since 2015, she holds broker licenses in Texas and Georgia, and is a Licensed Loan Advisor (NMLS 2536026). She serves as a Director on the Texas REALTORS® Board of Directors (2026–2027), Chair of the MetroTex Young Professionals Network (2026), and mentor at the NASDAQ Entrepreneurial Center. Her designations include Accredited Buyer Representative (ABR), Seller Representative Specialist (SRS), Pricing Strategy Advisor, and Short Sale & Foreclosure Resource. JL Marsaw & Co serves buyers, sellers, investors, and commercial clients across Dallas-Fort Worth, Houston, and Atlanta. Contact: 214-770-7762 | [email protected] | www.jlmarsaw.com
