I read the New York Times article about a homeowner selling his house with help from AI, and I thought it was fascinating. Not threatening, not offensive, not some shocking end-of-the-industry moment. Just fascinating.
The basic story is that the writer used AI to help prepare a listing, communicate with agents, understand documents, evaluate offers, negotiate, and ultimately sell his home without hiring a traditional listing agent. He still used other services along the way, including a flat-fee MLS option and a lawyer for closing, but the AI helped him handle a lot of the thinking and communication that people usually associate with an agent.
My first reaction is pretty simple: good. AI is here, it’s useful, and pretending otherwise doesn’t help anyone.
Part of why this article resonated with me is that I’m not anti-technology or anti-DIY at all. I have a background in tech, including time at Amazon, and I built and deployed my own real estate website using code and AI tools. I use AI regularly, and I think people who ignore it are going to fall behind.
I also like doing things myself when it makes sense. I change the oil in my own cars, I’ve built fences, and I generally appreciate self-sufficiency. There’s a real satisfaction in understanding how something works instead of immediately outsourcing it.
I’m writing this while drinking a coffee from Hotwire in West Seattle, which feels like a fitting little detail. Hotwire started as an internet cafe back in 2002, when that meant something very different than it does now. Most of that old tech element has faded into the background. The coffee shop is still here. The neighborhood is still here. The technology changed, people adapted, and the useful parts became normal. There’s still a printer and a place to charge your phone, but nobody thinks of sitting in a coffee shop with a laptop as some futuristic thing anymore.
That’s probably a decent way to think about AI and real estate too.
Technology changes industries. Sometimes slowly, then very quickly. The old version of a service doesn’t always disappear, but it has to justify itself in a new way. Real estate is going through that now. AI can explain things, draft things, organize things, and help people feel more confident moving through a process that used to feel much more opaque. That’s a good thing.
But I also think there’s a point where DIY starts to have diminishing returns. Sometimes doing it yourself saves money and gives you more control. Other times, the time, stress, risk, and potential downside start to outweigh the savings. Real estate sits right on that line because the dollar amounts are large, the contracts matter, and the condition of the actual home can change the entire equation.
I use AI. I know how to use AI. I also work in real estate, understand Seattle homes, understand the process, and spend a lot of time looking at actual properties in actual neighborhoods. To me, that combination is where this gets interesting. The future probably isn’t “AI replaces every agent” or “agents ignore AI and nothing changes.” The better version is agents and clients using better tools, with a lot less mystery around the process.
The article makes a fair point about the parts of real estate that have become easier to replace. Listing descriptions, basic email responses, explaining paperwork, organizing offer terms, writing polished messages, and helping someone think through a negotiation are all things AI can help with. Some of that work used to feel more specialized because the average person didn’t have an easy way to generate it quickly. Now they do.
That means agents have to provide real value beyond sounding professional.
A lot of the comments on the article seemed to fall into a few camps. Some people were thrilled because they’ve had bad experiences with agents and feel the fees are too high. Some were worried about AI replacing human relationships. Some pointed out, correctly, that the writer had time, tech confidence, and enough judgment to catch a bad AI recommendation before it became a problem. Others said the whole thing sounded exhausting and that they’d rather pay someone to manage it. All of those reactions make sense.
That’s probably the most honest answer here: it depends.
If someone is highly organized, comfortable with technology, has the time to manage showings and communication, understands risk, and is selling a fairly straightforward property in a strong market, AI can probably help them do a lot. That doesn’t mean it’s easy. It means the tools are getting good enough that motivated people can take on more of the process themselves.
But there’s a big difference between generating competent language and understanding what’s actually happening.
In Seattle real estate, the details matter a lot. A house in West Seattle near a busy cut-through is not the same as one tucked into a quieter pocket. A Ballard townhouse from one builder may perform very differently than a similar-looking one a block away. A Queen Anne Craftsman might have valuable original details, or it might have layers of deferred maintenance hiding under fresh paint. A condo might seem like a good deal until the HOA dues, reserves, rental cap, or upcoming assessments change the picture.
AI can help organize the questions, but someone still has to know what matters.
That’s where real estate experience still counts. Not because agents are special by default, but because houses are physical, local, and situational. Someone has to walk through them. Someone has to look at the drainage. Someone has to notice if the basement smells damp. Someone has to understand whether the yard is private because of thoughtful landscaping or because a laurel hedge is about to become a maintenance problem. Someone has to know whether a sewer scope is a nice-to-have or something that should be taken very seriously.
The article is also a good reminder that not every agent brings the same value. If the service is mostly opening doors, forwarding forms, and writing generic listing copy, AI is absolutely going to make that feel less valuable. It already does. The industry should be honest about that.
But a good agent should be doing more.
A good agent should understand pricing, preparation, negotiation, offer structure, inspection strategy, neighborhood context, buyer psychology, and the physical condition of homes. They should know when a house is sitting because the market missed it and when it’s sitting because buyers are rightfully concerned. They should understand what matters for resale, what matters for daily life, and what is just noise.
That’s especially true in Seattle, where the market is very block-by-block. Zoning changes, new construction, side sewers, old foundations, drainage, slope, tree cover, noise, parking, transit, and neighborhood feel can all affect value. AI can explain those concepts, but it doesn’t automatically know how they apply to a specific house on a specific street.
For sellers, I think AI can be a great tool. It can help brainstorm listing language, organize prep lists, summarize inspection issues, draft communication, compare offer terms, and think through negotiation options. I’m all for that. The question is whether the person using it knows what to ask, what to ignore, and what to verify.
For buyers, AI can also be useful. It can explain terms, compare neighborhoods at a high level, summarize documents, and help organize questions. But I wouldn’t want a buyer relying only on AI to understand a Seattle inspection report, evaluate a pre-inspection, judge a remodel, or decide whether a waived contingency makes sense. That’s where experience matters.
One of the best points in the broader discussion was about time. The writer didn’t just press a button and sell a house. He did the work. He coordinated, communicated, uploaded, reviewed, stressed, negotiated, and managed the process. AI helped him, but it didn’t physically carry the whole transaction. That’s an important distinction.
That’s also where the commission conversation gets more interesting. Some people want a full-service agent because they don’t have the time, confidence, or desire to manage the process. Some people want a more limited service model. Some people may want flat-fee help. Some may want AI-assisted consulting. I think we’ll see more of all of that.
I don’t think that’s bad. I think it’s probably healthy.
Real estate has been too opaque for too long. Consumers should understand what they’re paying for. Agents should be able to explain their value clearly. AI will force that conversation, and I’m fine with it.
My own view is that the best version of real estate going forward is not anti-AI. It’s AI plus judgment. AI plus local knowledge. AI plus someone who has walked through enough Seattle homes to know when something feels off. AI plus someone who understands how buyers actually behave, how sellers think, and how deals fall apart.
That’s where I feel very comfortable. I know how to use the tools, and I also know that the tools are not the whole job. They can make the process faster, clearer, and more organized, but they don’t replace the value of understanding homes, neighborhoods, risk, and people.
So yes, AI is going to change real estate. It already is.
The agents who only provide boilerplate are going to have a harder time. The agents who combine technology with real judgment, local expertise, and actual care for the outcome are probably going to become more useful, not less.
That’s the part I’m interested in. Not defending the old way. Not pretending AI isn’t powerful. Just using better tools while still doing the part that requires experience, taste, local knowledge, and a real understanding of what people are actually buying or selling.
If you're buying a home in Seattle, visit my Seattle buying guide. If you're thinking of selling your home, start with my selling roadmap. Browse Seattle neighborhoods or learn more about me.