Some of my job deserves to be automated. That is a strange sentence for a real estate agent to publish, so let me explain.
A meaningful slice of what this industry has historically billed as service is actually clerical work wearing a nice jacket. Scheduling showings. Chasing signatures. Sending the reminder about the reminder. Assembling documents that follow the same template every single time. Software already does most of this at the brokerages that have modernized, and clients rarely notice the handoff, which is exactly the point. NAR's latest technology survey found 79 percent of agents using electronic signatures and roughly seven in ten touching AI tools with some regularity. Nobody misses the fax machine. Nobody will miss the rest of it either.
For a long time, our industry padded its sense of importance with those clerical tasks. The stack of paperwork became a prop: look how much I handle for you. If a machine can do a task better, faster, and without dropping it during a busy week, that task was never the value. The value was always somewhere else, and automation is doing agents the strange favor of forcing us to locate it.
Consumers are running the same experiment from their side, and their results are worth reading closely. Realtor.com found last year that 82 percent of Americans shopping the market now use AI for housing information. Meanwhile, a large international survey released this spring found trust in AI to actually help find a home fell from 30 percent to 16 percent in a single year, and 44 percent of consumers said they would pay a human being to verify what AI tells them. Hold those two findings together and you get the moment we are living in: everybody uses the robot, and almost nobody trusts it with the decision.
They are right to hesitate, and the industry's most famous algorithm proves it. Zillow publishes its own accuracy numbers for the Zestimate: a median error around two percent for homes actively on the market, and around seven percent for homes that are not. On a $500,000 house near the bay, seven percent is $35,000. An estimate that can miss by the price of a decent boat is a conversation starter, and that is all it is. Pricing a home on Old Mission or a cottage on a private road in The County with no comparable sale since 2019 takes judgment, local memory, and a willingness to be accountable for the answer. The algorithm offers none of those.
The clearest way I can explain where the line sits is medicine. Pew Research found 60 percent of Americans are uncomfortable with their own doctor relying on AI for diagnosis and treatment. Notice what people are objecting to. The scan can be brilliant. You still want it read by a person who sits down, looks at you, and explains what it means for your life, and who stays in the room while you decide what to do about it. A home purchase runs on the same wiring. The alerts can be automated. The 9pm call after an appraisal comes in low, when a first time buyer needs to hear their actual options from a voice that has been here before, cannot.
So here is my operating rule, and I hold my own business to it: technology gets adopted in small pieces, and every piece has to pass one test, which is whether the client's experience improves. Automation that buys me back hours I then spend on strategy, negotiation, and picking up the phone is good for you. Automation used to run more volume with less attention is good for me at your expense, and you should not tolerate it.
Which suggests a question worth asking anyone you might hire: how do you use AI in your business? A good answer names specific tasks and where the human takes over. The two bad answers are a proud refusal to touch any of it and a fog of buzzwords with no line drawn anywhere.
The tools will keep improving, and I hope they hurry. Let them take the paperwork. What remains when the busywork drains away is the part I would do anyway: sitting with people through the biggest financial decision of their lives and telling the truth as well as I can. No update ships for that.
