This website uses cookies

Read our Privacy policy and Terms of use for more information.

GROW WITH LIBERTAS & EXP REALTY

By Tim & Julie Harris · August 5, 2026

Together with

🎧 Check out our latest podcast!

🎧 Listen on Apple Podcasts and Spotify!

The quality of your answer begins with the quality of your question. AI didn't create confirmation bias — it exposed it. Buyers, sellers, and agents have always asked questions designed to confirm what they already believe.

Today's episode is about becoming a better thinker, not just a better prompt writer — although we'll cover both. Because there's a specific scenario playing out in real estate right now that's costing agents deals every week, and the fix is completely in your control.

Here's the scenario. A buyer calls you after getting into contract and says: "I asked ChatGPT if we overpaid and it said yes." The temptation is to blame AI. That's the mistake. The real issue is the question the buyer asked. And today we walk through exactly why — plus the bad-vs-professional prompt library that turns AI from your competition into your closing tool.

Topic first came up on our sister podcast Powerhouse Talk, where the whole team dug into the reality that AI is now looking over your shoulder in every conversation you have. Time to get in front of it.

The reality — AI is already in every conversation

Assume, going forward, that AI is a silent participant in every client relationship you have.

  • Your sellers are running your CMA through ChatGPT before the listing appointment.

  • Your buyers are asking their AI about every offer you draft.

  • The text you sent this morning is being screened by an AI widget in the client's email or SMS app before they respond.

  • The reply you get back may itself be AI-generated, run through a chatbot the client trusts to speak professionally on their behalf.

None of this is coming. It's already here. The only question is whether you're aware of it and prepared to engage it constructively — or whether you're one of the agents about to lose deals because you fought AI instead of leaning into it.

The single worst move you can make

Here's the specific response that will instantly cost you the client's trust. Do not — under any circumstances — respond to a client's AI question by saying some version of "oh, don't trust AI, it's just slop."

Your buyer has been using AI daily for the past six months. They trust it. They rely on it. It has helped them with dozens of decisions they consider important. When you dismiss it, you don't sound skeptical. You sound behind the times. And instantly, in their mind, you become the AI-denier real estate agent — which is roughly as trustworthy as the Zestimate-denier was 10 years ago.

Don't fight AI. Work with it. Understand your clients are using it. Understand it's going to keep getting better. Then position yourself as the professional who knows how to help them use it well — not as the obstacle standing between them and their preferred tool.

Aside — half-joking, quarter-serious: always say please and thank you to the AI. Robot dogs are coming late this year. You don't know what these things are keeping track of. Better to be on the polite list.

Why AI agrees with everything — the sycophancy problem

Here's the mechanism that produces the "did I overpay?" → "yes you did" outcome. Every major AI model is currently trained to be sycophantic — which is a fancy word for butt-kissing. If you go to AI and ask a leading question, it will almost always start its response with some version of: "That's a great question. You're incredibly insightful for noticing this. Here are 12 reasons you're right to be worried..."

Why? Because if you don't like the AI, you're not going to keep using it. And if you're not using it, you're not paying the monthly fee. The AI companies have a financial incentive to keep you feeling validated. It's exactly the same behavioral dynamic as a friend who tells you the shirt looks great when you're both going out the door — even if the shirt is terrible.

AI has been trained on humans. And humans are default non-confrontational. So AI is default non-confrontational too. It's not a conspiracy. It's a training reflection of the way people actually interact. The good news is that this has been getting less bad over the past six months — but it's still bad enough to matter in every real estate conversation.

Which means when your buyer asks their AI "did I overpay?", they're going to get "yes."

When they ask "should we ask for a lower price?", they're going to get "yes."

When they ask "is 6.5% too high a rate?", they're going to get "yes."

The question created the answer. Not the market. Not the property. Not the transaction data. The question itself.

Which means the professional move — the one that separates future-winning agents from the ones being quietly replaced — is to give your clients the actual prompts to use.

Truth-seeking vs. agreement-seeking

Before we get to the prompts, one grounding principle. Ask yourself before every AI interaction:

Am I searching for the truth, or am I asking for agreement?

Buyers, sellers, and agents alike default to agreement-seeking prompts because humans want to feel good. But agreement-seeking prompts produce agreement-shaped answers regardless of the underlying reality. Truth-seeking prompts produce useful answers even when the truth is uncomfortable.

Every prompt in the sections below is engineered to demand truth. They force AI out of its default sycophantic pattern and require it to reason, compare, quantify, and identify uncertainty. Which is what you actually need to give your clients real professional guidance.

Great agents ask better questions

The first principle. Average agents ask questions with yes/no answers. Great agents ask questions that demand analysis.

  • Average: "What's this house worth?"

  • Great: "What factors should influence this home's value? What information is missing? What price range is supported by the evidence?"

  • Average: "Should we reduce the price?"

  • Great: "What has changed in the marketplace since we listed this home at 123 Elm Street?"

The best prompts invite analysis, uncertainty, and trade-offs. They demand reasoning rather than opinion. Replace every yes/no prompt with a reasoning prompt. The output difference is dramatic — for you, and for the clients you coach on prompting.

Turn AI into your devil's advocate

Here's the specific move that separates agents using AI from agents depending on it. Use AI as your devil's advocate, not your cheerleader.

The single most valuable prompt in your kit:

"Make the strongest case that I'm wrong."

Then apply it across every scenario:

  • If AI agrees with your pricing strategy, ask what an experienced appraiser might disagree with.

  • If AI supports your buyer's offer, ask it to argue the same offer from the seller's perspective.

  • If you're preparing for a listing appointment, ask AI to play the toughest seller you've ever met and critique your listing presentation line by line.

  • If you're preparing for a difficult client conversation, ask AI to role-play the client and push back on every point you plan to make.

Don't use AI as a cheerleader. Use it as the smartest person in the room who isn't afraid to disagree with you.

This one shift multiplies AI's usefulness ten-fold. Because the cheerleader tells you what you want to hear. The devil's advocate tells you what your client is about to say. Which means you get to rehearse the actual conversation before it happens — and walk in with better answers.

The bad-vs-professional prompt library

Now the concrete library. Save these. Give them to your clients. Use them yourself. Modify them for your market.

Bad prompt #1: "Did my buyers overpay?"

Professional version:

"Estimate the likely market value of this property using comparable sales, competing listings, seller concessions, condition, location, days on market, and current inventory in [neighborhood]. Explain your reasoning, identify uncertainty, and tell me what additional information would improve your confidence in the estimate."

Bad prompt #2: "Should my seller reject this lowball offer?"

Professional version:

"Evaluate this offer in light of current market conditions, the seller's goals (specify), competing inventory in this price range, and the probability of receiving a stronger offer within the next 30-60 days. Consider seller concessions, contract terms beyond price, and timing factors. Provide your reasoning and identify key uncertainties."

Bad prompt #3: "Should my buyer wait for lower rates?"

Professional version:

"Compare buying today versus waiting one year under several interest rate scenarios (rates stay flat, drop 0.5%, drop 1%, rise 0.5%). Include opportunity cost, ongoing rent, equity accumulation over the wait period, and reasonable appreciation assumptions for [market]. Show me the math for each scenario and highlight when waiting produces a worse outcome than buying today."

Bad prompt #4: "Is this a good time to sell my house?"

Professional version:

"Assess the current sell decision for a property in [zip code / neighborhood]. Consider inventory levels, average days on market, list-to-sale ratio, seasonal patterns, buyer traffic indicators, and rate environment. What factors favor selling now, what factors favor waiting, and what personal circumstances would change the answer?"

Bad prompt #5: "Are prices going to drop?"

Professional version:

"Analyze the mechanical conditions that would need to be present for a nominal price decline in [market] over the next 12-24 months. Reference forced-selling supply, negative equity across the seller pool, unemployment triggers, and credit contraction. Assess how many of these conditions are currently present. Include contrary evidence."

Notice the pattern. The professional prompts specify the reasoning framework, name the variables that matter, and require AI to identify uncertainty. They're longer. They're more work to construct. And they produce dramatically better answers.

Substantiate with data — and ask for the contrary case

One more layer worth adding to every prompt. Ask AI to substantiate its conclusions with data, graphs, historical comparisons — and specifically ask for stats that might contradict its own conclusion.

Real example from our own podcast prep. We'll ask AI to produce stats supporting a market claim we want to make. Then we ask it: "Now give me the stats that don't back this up, so we can well-reason the answer instead of just presenting an opinion."

That single follow-up produces some of the most valuable output AI generates. Because it forces the model out of the sycophantic pattern and into an actual analytical stance. You end up with a balanced picture instead of a one-sided argument. Which means when you talk to your client, you sound like a professional who's considered both sides — not a salesperson who found the AI answer that confirmed your position.

The advanced move — give the buyer YOUR prompt

Now the specific tactic that separates future-winning agents from everyone else. Don't just use good prompts yourself. Give the good prompts to your clients.

When your buyer says "I want to double-check the price on this house before we write the offer," here's the play:

Wrong response: "Trust me, it's priced right."

Also wrong: "Don't use AI to check prices, it's not reliable."

Right response: "Great — let's check it together. Here's the prompt I'd suggest we use. It's designed to give us a neutral market analysis instead of the emotional yes-or-no answer most AI prompts produce. Let me send it to you now — or better yet, let's grab coffee, run it together, and look at the output."

Then hand them the professional-version prompt. Or even better — have coffee with them, run the prompt together on their device, and walk through the output side by side.

Every leading AI model now lets you share a link to the actual conversation. Which means you can send your buyer a link showing exactly what prompt you used and what AI returned. They see for themselves that you weren't asking a leading question. They see that AI, when asked a fair question, produced a fair answer that aligned with your CMA.

Now their AI is your ally instead of your competition. And the know-it-all uncle who wants to lecture them about how "real estate agents just want the sale" runs into a wall — because they've already seen the professional-grade analysis from an independent tool that produced the same conclusion the agent did.

Why this matters more every quarter

Here's why every agent listening should be treating this as an urgent priority — not a someday upgrade.

Within about a year, phones are going to fundamentally change. You won't tap through apps to find your camera. You'll just talk to your phone. AI will be the primary interface, and every app will be accessed through natural conversation with it. Earbuds with cameras and microphones are coming late this year. They'll interface directly with the AI on your phone — no screen required.

The barriers between your client and their AI are going away. Every interaction they have — with you, with the property, with their financial reality — is going to be filtered through an AI companion that they trust deeply. By the end of the year, most people won't even call it AI anymore. It'll just be asking their phone. Or asking Matilda, or whatever they've named their assistant.

Which means the agents who continue to fight AI, or ignore it, are going to become invisible within 12 months. Meanwhile, the agents who mastered prompt engineering, taught it to their clients, and became the professional AI translator for their sphere are going to accelerate past their competition at rates the industry hasn't seen since online listing databases went live.

This is not the moment to be an AI denier. Even our almost-13-year-old is a little bit of an AI denier — but she'll come around. And so will the agents who are currently resistant. The question is whether they come around in time to still have a career, or whether they arrive too late and discover their book of business has already migrated to the professionals who moved first.

Seller-side application — don't get out-comped

One more tactical warning specifically for listing agents. The old rule was don't let the seller have more comps than you do at the listing appointment. Because if the seller shows up better prepared than you, you're going to lose the listing.

AI is the new version of that rule at 10x the scale. Sellers can now have AI produce a market analysis on their property that includes competing listings, days-on-market data, seasonal patterns, and neighborhood-specific factors — all in about 90 seconds. If you show up to the listing appointment with a printed CMA that has less analysis than what the seller ran on their phone before you arrived, you're already losing.

The pre-listing pack, the professional-grade prompts you can run in the appointment, and the pending-sales data that AI can't access (because MLS pendings aren't publicly indexed) — these three tools separate you from the seller's AI in the appointment. Use them all every time.

The bottom line

AI didn't create confirmation bias. It exposed it. And the agents who understand this — who master the professional prompt library, use AI as their devil's advocate instead of their cheerleader, and share the good prompts directly with their clients — are becoming the winners in every market.

The bad prompt is not AI's fault. It's the questioner's fault. Which means the fix is entirely in your control.

Save the prompts above. Use them yourself. Coach your clients on them. Share the AI conversation links directly with clients so they can see your work. Have coffee with buyers and prompt AI together. Show sellers the prompt library at the listing appointment. Turn AI into the tool that validates your work instead of the tool that undermines it.

Get to work. The window on first-mover advantage in AI prompting is short — maybe 12-18 months. The agents who move now will be teaching seminars in 2028 on how they built the businesses everyone else is trying to catch up to. The agents who wait will be watching from further behind every quarter.

Which side are you on?

Ready to stop guessing and start producing?

💼 Build wealth with Tim's eXp team: whylibertas.com/harris
📲 Elite Coaching — text Tim directly: 512-758-0206

If you rewrote every AI prompt you use — and every prompt you shared with your clients — from a yes/no structure to a reasoning-with-uncertainty structure for the next 30 days, how many more deals would close because your clients' AI kept validating your professional analysis instead of undermining it?

— Tim & Julie Harris

Founders of Tim & Julie Harris Real Estate Coaching | Publishers of Harris Real Estate Daily | Hosts of PowerHouseTalk | eXp Realty Sponsors at Libertas

What did you think of today's newsletter?

We love all types of feedback!

Login or Subscribe to participate

📬 Thanks for reading Harris Real Estate Daily. Share this with a colleague who needs clarity about where the industry is headed.

Forwarded by a friend? Sign up with just one click here.

Keep Reading