Nobody Wants to Talk to Your AI Assistant

80% of consumers say they feel MORE frustrated after interacting with a chatbot. Not less. More. That comes from the Consumer Financial Protection Bureau, a federal agency, and it should stop every real estate agent mid-purchase on whatever chatbot subscription they are about to buy. The CFPB also found that 78% of those consumers ended up needing a human anyway. So the chatbot did not solve the problem. It just added a frustrating step before the problem got solved.

Nobody calls a real estate agent hoping to talk to their assistant. And they definitely do not want to talk to a real estate chatbot that is probabilistically guessing at the right answer and sending it back with confidence. Sounds harsh right. But lets walk through why this matters, because I think most agents have the right instinct about AI and the completely wrong application of it.

I should say upfront: I am probably one of the most aggressive AI adopters in residential real estate. 19 years as a broker, 2,000+ transactions, and I have built more production AI systems than I probably should have (my wife would agree). Automated CMA systems. An AI-powered content pipeline. I use Claude AI every single day to run my brokerage. But talking to my clients? That is the line. That is where AI stops and Ed Neuhaus starts.

What a Real Estate Chatbot Can Actually Do

Lets be honest about the ceiling here. A chatbot can look up a phone number. It can tell you office hours. It can maybe pull a listing from the MLS feed if the integration is wired up correctly.

That is it. That is the entire capability.

If the answer lives in a database and the question is simple enough to match against a script, the real estate chatbot works. “What is your office address?” Great. “What are your hours?” Perfect. “Can you send me the listing details for 123 Main Street?” Sure.

But nobody is calling a real estate agent to ask what the office hours are. The people reaching out to you have real questions that require judgment. Should I waive the option period on this offer? Is this neighborhood going to appreciate? My inspector found foundation cracks, how worried should I be? What is the best strategy to compete against three cash offers?

A chatbot cannot answer any of those questions. Not even close. And a real estate agent could never realistically produce a chatbot accurate enough to handle them. Think about what you would need. The bot would need to understand your local market at a granular level, know which neighborhoods flood, which HOAs are problematic, which builders cut corners, which school zones are about to get redistricted. It would need to understand Texas contract law well enough to advise on option periods and contingencies. It would need the emotional intelligence to tell a nervous first time buyer that everything is going to be ok and mean it.

No one is building that. No one can build that. And the vendors selling you a real estate chatbot for 00 a month are not even close to trying to.

AI Chatbots Are Worse Than Scripted Ones

So here is the thing that really bothers me about the current wave of AI chatbots for realtors. A traditional scripted chatbot at least knows what it knows. You ask it a question it has a script for, it gives you the right answer. You ask it something off script, it says “let me connect you with an agent.” Honest. Limited, but honest.

An AI chatbot does something different and worse. It probabilistically guesses at the right answer and sends it back with confidence. It does not know what it knows. It does not know what it does not know. It just generates a response that sounds plausible.

Daniel Kahneman wrote extensively about how humans are terrible at distinguishing confidence from competence. We hear a confident answer and assume the person (or machine) actually knows what they are talking about. An AI chatbot exploits that exact cognitive bias every single time it responds. It is a junior employee who fabricates answers but delivers them with the polish of a 20 year veteran.

And you are putting that between yourself and your clients.

If you are a realtor and you are allowing something to speak for you, to guess at answers and send them back to people making six figure decisions, that is a bad look. That is a liability. I wrote about this dynamic when I built our automated CMA system. The AI never talks to the client directly. Scripts do the math. AI writes the narrative around the data. But there is always a human checkpoint before anything reaches a client. That distinction matters more than people realize.

The Data Is Brutal

This is not just my opinion. The numbers tell a pretty clear story.

The CFPB study documented what they called “doom loops” where chatbots trap consumers in circular conversations with no way to reach a human. If you have ever tried to get past a chatbot to talk to a real person and just kept getting routed back to the same menu, that is a doom loop. A federal agency named it and documented it because it was happening at scale across financial services. Real estate is a financial service.

The Invoca survey of 1,000 US and UK consumers found numbers that should keep every real estate chatbot vendor up at night. 60% of consumers say they feel forced to use a brand’s AI. Not impressed by it. Not excited about it. Forced. Only 14% of Baby Boomers said they had a positive experience with AI when making a high-stakes purchase. Fourteen percent. And if you are working in residential real estate, Boomers are your seller demographic. They own the houses. They are the ones deciding whether to list with you or the agent down the street. 86% of them do not have fond memories of talking to your chatbot.

It gets worse. 53% of all consumers say AI performs worst at solving complicated problems. Buying or selling a home is probably the most complicated financial transaction most people will ever do. So the majority of consumers already believe AI is bad at exactly the type of question they need help with.

And here is the one that really stands out. Since 2022, the preference for calling a business on the phone has increased by 12%. Not decreased. Increased. In an era where we can text, email, chat, DM, and use whatever platform we want, people are choosing the phone more often for high-stakes purchases. They want a human voice. They want to hear someone think through their problem in real time.

As Rex Software put it, “Your clients don’t trust software. They trust you, especially when they’re investing thousands (or millions).” And “When buyers feel like they’re talking to a script, they may vanish.” Not complain. Not ask for a human. Vanish.

The Companies That Tried It and Failed

I keep hearing agents say “but the technology is getting better.” Ok. Lets look at the companies that went all in on AI chatbots and what actually happened.

Air Canada deployed a chatbot on their website and it told a customer he could get a bereavement fare discount on flights he had already purchased. The customer booked based on that information. The discount did not exist in the way the chatbot described it. When the customer asked for the promised discount, Air Canada’s defense (and I am not making this up) was that the chatbot was “a separate legal entity responsible for its own actions.” They literally argued the robot was not their responsibility.

The Canadian tribunal ruled against them. The finding was blunt: “It should be obvious to Air Canada that it is responsible for all the information on its website.” They were ordered to pay 12 in damages. Small dollar amount. Massive precedent. If a chatbot on your website gives wrong information, you are responsible. Not the chatbot vendor. Not the AI model. You.

Now think about that in real estate. Your chatbot tells a buyer that a property does not have an HOA (it does). Or that the option period is 10 days (it is 7 in the contract). Or that the seller will pay closing costs (they will not). You are on the hook for that. Your license. Your E&O insurance. Your reputation.

Then there is Klarna. They went the other direction. Instead of deploying a chatbot alongside humans, they replaced 700 customer service agents with AI. CEO Sebastian Siemiatkowski was initially thrilled. Then quality dropped. His own words: “As cost unfortunately seems to have been a too predominant evaluation factor… what you end up having is lower quality.” And: “Really investing in the quality of the human support is the way of the future for us.”

Klarna is now actively rehiring humans. Let that sink in for a second. A publicly traded fintech company with some of the best engineers in the world tried to replace human customer service with AI, realized it made the product worse, and reversed course. If Klarna could not make AI customer communication work, a solo real estate agent with a 00/month chatbot subscription is not going to crack it. I mean come on.

Talking to People Is Literally the Job

Ok lets take a step back from the data and talk about something more basic. What does a real estate agent actually do?

You talk to people. That is the job.

Your judgment. Your knowledge of the market. Your ability to read a situation, calm someone down during a stressful inspection, negotiate on their behalf when there are three competing offers, explain to a seller why their Zestimate is wrong without making them feel stupid. None of that is automatable. Not now. Probably not ever.

The National Association of Realtors put it simply: “At the heart of it all remains the trusted relationship between the agent and client.” That relationship is built through conversation. Real conversation where a human is listening, processing context, reading emotions, and responding with judgment.

Nobody wants to talk to your junior analyst. Nobody wants to talk to your assistant. They want the person who makes the decisions. When you put a real estate chatbot between yourself and your client, you are automating the one thing they are paying you for.

Think about it this way. If an agent automates every client conversation, what exactly is the client paying commission for? The MLS access? They can get that from a dozen apps. The paperwork? A transaction coordinator handles that for 00. The negotiation, the local knowledge, the human judgment in high-stress moments, that only comes from the conversation. And you are trying to hand it to a robot.

Seth Godin has this concept about being a “purple cow” in a field of brown ones. The remarkable thing, the thing that makes people notice you and trust you, is your human judgment. Your experience. Your willingness to tell someone the truth even when it hurts your commission. A chatbot cannot be a purple cow. It is the most brown cow imaginable. Same generic, personality-free interaction at every brokerage that deploys one.

The Speed Argument (And Why It Is Wrong)

I know what the chatbot vendors will say. “But Ed, most buyers work with the first agent who responds. Agents take forever to reply. The chatbot responds in seconds.”

Ok, lets unpack that. Speed matters. I am not arguing that. What I am arguing is that the real estate chatbot industry is conflating the problem with the wrong solution.

The problem: agents are too slow to respond to leads.

The chatbot industry’s solution: remove the agent from the response.

See the issue? You have identified a real problem (agents are slow) and proposed a solution that removes the thing the client actually wants (the agent). That is like saying “patients wait too long to see a doctor” and solving it by having an AI do the diagnosis. The wait time goes to zero. But you have also removed the thing the patient came for.

The actual solution is simpler and less expensive than a chatbot subscription. Fix your operations. Set up notifications on your phone. Hire an assistant (a real one, a human being) if you need coverage. Build a team where someone is always available. Use systems that route leads to whoever can pick up right now.

A fast human response beats a fast bot response every time. Every. Single. Time.

The client who gets a real person on the phone in 3 minutes is infinitely better served than the client who gets an AI chatbot in 3 seconds that says “Thanks for reaching out! I’d love to help you find your dream home. Can you tell me more about what you’re looking for?” That template response fools nobody by the way. People can tell. And remember, the CFPB found that 78% of people who interact with a chatbot end up needing a human anyway. So your chatbot is not solving the speed problem. It is just adding an extra step.

Where AI Actually Belongs in Your Brokerage

So if I am so down on the real estate chatbot, what do I actually use AI for? Great question. The answer is: everything behind the scenes.

At Neuhaus Realty Group, AI touches almost every part of our operation. But it never, ever speaks directly to a client.

Here is what AI is genuinely good at in real estate:

Market analysis and data processing. AI can crunch comps, analyze market trends, and identify pricing patterns faster than any human. Our automated CMA system generates property valuations that are genuinely useful. The scripts do the math. AI writes a readable narrative around the data. But I review every single report before it goes to a client.

Content creation. You are reading an article right now on a site with hundreds of data-driven articles about Austin real estate. AI helps me produce more content, better content, and do it consistently. That is a legitimate use of the technology because the content goes through my review and reflects my actual positions. It is not guessing at what I would say. I am telling it what to say and it helps me say it faster.

Transaction management. Tracking deadlines, organizing documents, summarizing inspection reports, drafting emails that I review and edit before sending. All of this makes me more efficient. None of it pretends to be me.

Research and analysis. Neighborhood data, school ratings, market forecasts, investment analysis. AI can compile and synthesize information that would take me hours to gather manually. It makes me smarter and faster when I sit down with a client.

The pattern is simple. AI does the work that makes me better at my job. It never does the job itself (which is talking to people and exercising judgment). If you want to understand why I chose Claude AI as my primary tool, that philosophy is exactly the reason.

The Bottom Line

You got into real estate because you are good with people. Maybe that sounds cheesy but it is true right. The ability to sit across from someone, understand what they actually need (which is often different from what they say they need), and guide them through one of the biggest financial decisions of their life. That is why people hire you. That is the entire value proposition.

Stop trying to automate it. Use AI for the things that free you up to do more of it. Build systems behind the scenes that make you faster, smarter, better prepared. Then pick up the phone and be the person your client hired.

Nobody wants to talk to your assistant, much less your AI assistant.

If you are an agent trying to figure out where AI fits in your business without crossing the line into client communication, I have been down this road. Lets talk.

Be safe, be good, and be nice to people.