It has become a familiar moment at the start of an appraisal. Somewhere between the kitchen and the lounge, the seller mentions, a little sheepishly, that they asked ChatGPT what the house might be worth. Buyers are doing the same, asking AI tools what they should offer, whether a home is overpriced, or whether now is the right time to move at all. Some arrive with sharper questions as a result. Others arrive with a number they have already become attached to.
I should say at the outset that this isn’t an attempt at dismissing the technology, far from it; when used well, AI is a genuinely capable research assistant. It can explain how LBTT is calculated, map out your potential mortgage repayments, summarise what a Home Report covers, or help a buyer organise their thoughts before a viewing. It can even help shape the covering letter we wrote about a few weeks ago, provided the sentiment behind it is your own.
For a process that has historically felt opaque to most people, anything that helps buyers and sellers arrive better informed is welcome. The difficulty lies in mistaking a well-written answer for good advice.
“What the machine knows…”
“What the machine knows…”
An AI tool works from what has already been written down. It draws on published sale prices, averages and general commentary, all of which describe what happened some time ago, and usually at the level of a postcode or a local authority rather than a street. It cannot know that the closing date on the next road last Thursday attracted five offers, or that the home two doors along sat for three months because buyers kept walking away from the layout. It has no way of knowing which buyers are sold and ready to proceed, which have missed out twice and are now prepared to stretch, or which homes are quietly changing hands without ever reaching a portal.
It also answers with the same assurance whether it is right or wrong. That fluency is persuasive, and it has a habit of reflecting the framing of a question back at the person asking it. Ask whether your home is worth more than the figure you have in mind, and you may well find the answer leaning towards agreement.
In property, where the costliest mistakes usually begin with a confident assumption, that is not a small flaw.
“What the machine cannot read…”
“What the machine cannot read…”
Buying and selling a home is one of the few significant financial transactions in life that is also deeply personal. The seller is often leaving behind years of family life, while the buyer is trying to picture a future in rooms they have spent twenty minutes in. Both are making decisions under pressure, frequently against competition and a deadline, and rarely with complete information.
That is where emotional intelligence earns its place. It is the ability to notice that one half of a couple has gone quiet in the second bedroom, and to understand what that silence might mean. It is knowing when a seller is ready to hear a difficult truth about price, and how to put it in a way they can act on. It is recognising, mid-negotiation, that the buyer on the other side is anxious about their own sale, and finding the reassurance that holds a deal together rather than letting it unravel over a few thousand pounds.
None of these moments appear in a dataset, yet any one of them can change the outcome of a sale.
Experienced agents build that instinct over years of conversations, many of them difficult, and it is the one thing that cannot be downloaded. A good valuation, for instance, is a judgement about this home, these buyers and this moment, along with the strategy most likely to turn interest into competition.
The figure an algorithm produces may be a reasonable starting point for curiosity, but it makes a poor foundation for a pricing decision, and sellers who anchor to it can find themselves arguing with the market rather than working with it.
“The risk of the confident shortcut…”
“The risk of the confident shortcut…”
Earlier this year, we wrote about the gap between property fluency and property competence, and how a nation well versed in the language of the market can still struggle to navigate it. AI has the potential to widen that gap considerably. It makes everyone sound well informed, quickly and convincingly, while removing the friction that used to prompt people to seek a second opinion.
The buyer who commits to an offer because a chatbot suggested the asking price was ambitious, only to lose the home to someone who read the situation better, has paid for that shortcut. So has the seller who holds out for an AI-generated figure and watches their listing go stale while the right buyer moves on. In both cases, the information was not necessarily wrong, but it was nowhere near enough.
There is a sensible middle ground. Use AI to prepare, to understand the terminology, to frame your questions and to sense-check what you are being told. Then take those questions to people who are in the market every day, who have walked through the home, who know the buyers, and who will be beside you when decisions start to carry real consequences.
At Corum, we use technology extensively in how we market homes and manage our buyers, and that will only grow. The most important moments in buying and selling a home, however, still happen between people, in kitchens and hallways and on the end of a phone late on a closing date.
Those are the moments where judgement, empathy and experience decide the outcome, and no amount of artificial intelligence has yet learned to stand in for them.