EP 388 How AI Can Make You a Worse Real Estate Investor

APPLE PODCASTS | SPOTIFY

There’s something strange happening with AI right now.

It has never been easier to analyze a real estate deal.

And somehow, it has never felt harder to know whether you’re making the right decision.

AI can do in minutes what used to take an investment team hours. It can read an offering memorandum, analyze a rent roll, compare a T-12, research a market, stress-test assumptions, and write an investment memo before you’ve finished your coffee.

That sounds like an enormous advantage.

And it is.

But there’s a problem we don’t think enough investors are talking about:

The more AI can do for us, the easier it becomes to stop thinking for ourselves.

My sister Nancy and I have spent the last seven-plus years building a real estate portfolio approaching $500 million. We’ve returned more than $45 million to investors and helped families preserve more than $110 million in taxes.

And after all of those deals, all of those underwriting sessions, and plenty of decisions we got right and wrong, we’ve come to believe something pretty strongly:

The biggest advantage AI gives a real estate investor isn’t speed. It’s the ability to multiply good judgment.

But only if the judgment is there first.

And weirdly enough, one of the best ways we’ve found to explain this starts with a fire at NASA almost 60 years ago.

The Fire That Changed How NASA Built Spacecraft

January 27, 1967.

Three astronauts—Gus Grissom, Ed White, and Roger Chaffee—were sitting inside the Apollo 1 capsule during a launch-pad test.

The cabin was filled with pure oxygen.

Then somewhere in the wiring, there was a spark.

In ordinary air, that spark might have smoldered.

Inside the oxygen-rich capsule, materials that normally wouldn’t burn easily became fuel.

The fire spread in seconds.

All three astronauts were lost.

But there’s one detail from that story we haven’t been able to stop thinking about.

Oxygen doesn’t burn.

It wasn’t the spark.

It wasn’t the fuel.

Oxygen simply made everything else burn faster and hotter.

And that’s exactly how we’ve started thinking about AI.

AI Is Oxygen

AI isn’t inherently good judgment or bad judgment.

It’s an accelerant.

If you already understand markets, leverage, underwriting, operators, capital stacks, and risk, AI can help you analyze more information, challenge more assumptions, and potentially spot things you would have missed.

But if you start with a rent-growth assumption that doesn’t make sense and a deal you desperately want to believe in?

AI can accelerate that too.

It can build the model.

Write the memo.

Find supporting arguments.

Create beautiful tables.

And make a questionable thesis look incredibly sophisticated.

We think of it like leverage.

A loan doesn’t know whether your investment is good.

Debt doesn’t have an opinion.

It simply magnifies the outcome.

AI works similarly.

AI × poor judgment = faster mistakes.

AI × strong judgment = leverage.

And that changes how we believe investors should use it.

NASA Didn’t Get Rid of the Oxygen

After Apollo 1, NASA didn’t decide oxygen was too dangerous and stop using it.

Astronauts need oxygen.

Instead, NASA changed the environment around it.

The materials changed. The cabin environment changed. The process changed.

They changed what was in the room before adding the accelerant.

We think investors need to do the same thing with AI.

Before we let AI analyze a deal, we want our investment mandate established first.

What markets will we invest in?

What assumptions are acceptable?

What risks won’t we take?

What has to be true before we’ll deploy capital?

Maybe one of your rules is:

We won’t underwrite more than 2% annual rent growth unless the leases already prove it.

Or:

We won’t buy in a market where we don’t have an operating advantage.

Or:

We won’t stretch the exit cap just to make the IRR look better.

Those rules come before AI.

Then AI operates inside them.

That’s how we’ve designed our own AIREI OS, but the principle matters more than the software:

Judgment first. Acceleration second.

And that brings us to a very small bird.

The Bird That Raises the Chick That Destroyed Its Family

There’s a little brown bird in Europe called the reed warbler.

She builds a nest.

Lays her eggs.

And while she’s gone, a cuckoo can sneak into her nest and lay an egg of its own.

Then the cuckoo leaves.

The cuckoo chick hatches and, within its first few days of life, pushes the warbler’s eggs out of the nest.

Then the mother warbler returns.

And instead of rejecting the intruder…

She feeds it.

For weeks.

Even as the cuckoo becomes several times her size.

Why?

Researchers found that the cuckoo chick’s rapid begging call can mimic the feeding stimulus of an entire brood.

One voice sounds like many.

And the warbler responds.

Now think about what happens when we upload a real estate deal into an AI chatbot and type:

“Is this a good investment?”

One voice suddenly sounds like an entire investment committee.

It sounds like an underwriter.

A market analyst.

An asset manager.

A risk officer.

Maybe even an investment committee.

And because the answer is polished, organized, and confident, we can start mistaking presentation for diligence.

That’s the cuckoo.

AI Can Make a Bad Deal Sound Very Smart

The scary part isn’t that AI will always give you obviously ridiculous answers.

Those are easy to catch.

The dangerous answers are the ones that are almost right.

The spreadsheet looks legitimate.

The memo sounds intelligent.

The assumptions seem reasonable.

And the conclusion happens to confirm exactly what you were already hoping was true.

So how does the reed warbler defend itself?

Not after the cuckoo hatches.

By then, it’s often too late.

Its defense happens at the egg.

Some warblers learn what their own eggs look like—the color, pattern, and size—so they can identify an egg that doesn’t belong.

And we think there’s a huge investing lesson hiding inside that:

You can only recognize what’s wrong when you already know what right looks like.

That’s why we believe most investors are using AI in the wrong order.

They start with AI.

Then maybe apply judgment afterward.

We want to reverse that.

Judgment → AI → Verification → Decision.

Not:

AI → Answer → Invest.

We Built a Fake Deal to See If AI Would Catch the Lies

We wanted to put this idea to the test.

So we created a fictional 205-unit garden-style multifamily property in a Sun Belt submarket we’ll call Riverton.

We built the entire package.

Offering memorandum.

Trailing twelve months.

Full rent roll.

Then we deliberately planted the kinds of inconsistencies we’ve encountered while buying real estate.

On the surface, Riverton looked great.

The OM said occupancy was 94%.

The property was supposedly stabilized.

In-place rents were supposedly 8% below market.

Year-one rent growth was projected at 4%.

Taxes looked reasonable.

NOI supported a purchase price that didn’t seem crazy.

If you simply gave that package to a chatbot and asked:

“Is this a good investment?”

You could get a polished, convincing investment memo.

So we changed the question.

Instead of asking AI to validate the deal, we asked it to cross-check the documents against each other.

And that’s when Riverton started falling apart.

$12.8 Million Disappeared

The OM claimed 94% occupancy.

The rent roll showed 89.3%.

The OM advertised rents 8% below market.

The leases didn’t support the story.

Recurring expenses had been pushed below the line, making historical operations look healthier.

Then there were the taxes.

The tax expense was roughly $200,000 lower than what the county could actually assess after a sale.

Add everything together, and we found approximately:

$672,000 of annual NOI that didn’t actually exist.

At the deal’s cap rate, that represented roughly:

$12.8 million of value that wasn’t there.

Think about that for a second.

The numbers were all there.

The documents were all there.

But the problem didn’t live neatly inside one spreadsheet.

It lived in the gaps between them.

The occupancy in the OM versus the occupancy in the rent roll.

The taxes in the T-12 versus what the county would actually charge.

The “below-market” rent story versus what the leases could support.

That’s where we think AI becomes incredibly powerful.

Not as the person who tells us what to think.

As the system that helps us find where we should think harder.

One of Our Favorite AI Answers Is “I Can’t Verify This”

There was another part of the Riverton experiment that we loved.

On one finding, the system couldn’t confirm the answer.

And instead of pretending it knew?

It said:

Cannot verify.

Then we supplied additional evidence and let it test the claim again.

That’s what we want.

Because confidence isn’t the same thing as accuracy.

A cuckoo imitates certainty.

A good investment process tells you where it can’t see.

Then we can attack the thesis.

What happens if rent growth is 2% instead of 4%?

What happens when taxes reassess?

What happens if actual occupancy is the rent-roll number rather than the brochure number?

What happens if our exit cap expands?

What happens if we’re simply wrong?

For Riverton, our answer became pretty clear:

Don’t proceed at this price.

But notice what AI didn’t do.

It didn’t make the investment decision.

We did.

A Prompt Forgets. A System Remembers.

This is also why we’re becoming less interested in individual prompts and more interested in systems.

A prompt can analyze one moment.

An operating system can remember the standards you used when you bought the property and compare those assumptions against what actually happens after closing.

Every month.

Property by property.

Assumption by assumption.

Did occupancy perform the way we expected?

Did expenses stay where we modeled them?

Did rent growth materialize?

Did the business plan actually happen?

Because due diligence shouldn’t end at closing.

A prompt forgets. A system remembers.

But there’s still another AI trap we see investors falling into.

And to explain this one, you don’t need NASA or a bird.

You just need to look at your own body.

Your Heart Knows How to Be an Eye

Almost every cell in your body carries essentially the same genetic instructions.

Your heart cell contains the blueprint your body uses to build an eye.

Your skin cell contains the blueprint your body uses to build your brain.

The whole manual is there.

But your heart doesn’t suddenly decide:

“You know what? I have the instructions. I think I’ll be an eye today.”

It specializes.

A heart cell beats.

A lung cell does its job.

A liver cell does its job.

Your body works because trillions of specialized cells perform different jobs together.

Having access to all the information doesn’t mean every part should do everything.

And AI has suddenly given investors access to something very similar.

The whole manual.

Underwriting.

Market research.

Asset management.

Capital raising.

Investor relations.

Deal sourcing.

AI can give one person access to knowledge across all of those functions.

And that’s amazing.

But it can also create a dangerous illusion:

Maybe I don’t need anybody else anymore.

AI Shouldn’t Turn You Into a One-Person Investment Firm

One of the biggest mistakes we see aspiring GPs make is trying to become great at everything.

We think about a real estate deal as requiring five major seats:

Deal finding. Underwriting. Asset management and operations. Capital raising. Net worth and liquidity.

Different seats require different strengths.

Trying to occupy all five can leave someone average at several things instead of exceptional at one.

So our goal isn’t to use AI to replace every person around us.

It’s to use AI to become exceptional in our seat and then partner with people who are exceptional in theirs.

That’s an important distinction.

AI doesn’t have a balance sheet.

It can’t personally guarantee your loan.

It hasn’t operated through three cycles.

It can’t build trust with an LP over dinner.

It doesn’t have relationships that took 20 years to build.

And it doesn’t have your scar tissue.

So if you’re a GP, ask yourself:

What’s my seat?

If you’re an LP evaluating a sponsor, ask something slightly different:

Who occupies each seat?

Put an actual person’s name beside each function.

Who finds the deals?

Who underwrites them?

Who operates them?

Who raises and communicates with capital?

Whose balance sheet supports the transaction?

If one person claims to occupy three or four seats because “AI handles the rest,” we’d want to understand that very carefully.

Because your LPs aren’t backing your chatbot.

They’re backing your judgment.

What If You Already Feel Behind on AI?

This is the part we think a lot of people quietly feel.

Someone has a new AI tool.

Someone else built an agent.

Someone automated their underwriting.

Another person is posting about a workflow they built over the weekend.

And suddenly you start wondering:

Are we already behind?

We don’t think that’s the right question.

Moving faster doesn’t necessarily mean moving ahead.

Sometimes it simply means making the same mistakes at a higher velocity.

We’ve been investing long enough to have gotten things right—and to have been wrong enough times to know that judgment isn’t something you download.

It’s earned.

Deal by deal.

Decision by decision.

Mistake by mistake.

That feeling you get after looking at thousands of lines in a financial statement and realizing something doesn’t add up?

AI doesn’t make that obsolete.

If anything, it makes it more valuable.

Because AI is making intelligence abundant.

Which means judgment may become one of the scarcest things in the room.

The Real AI Advantage Isn’t Speed

Yes, AI can do in minutes what once took an investment team hours.

That’s real.

And yes, we think it’s going to fundamentally change how real estate investment firms operate.

But speed by itself isn’t the edge.

Speed without judgment is expensive momentum.

Judgment with speed?

That’s leverage.

Which brings us all the way back to Apollo 1.

AI is the oxygen.

We decide what’s already in the cabin.

That’s why our order stays the same:

Judgment first.

Then AI.

Then verification.

Then a human decision.

AI is the mechanism.

Judgment is the product.

Want to See How We Actually Use AI to Make Better Investment Decisions?

If you want to see how we run this sequence inside our AIREI OS—the specialized seats, the workflows, and how we use AI without outsourcing the final investment decision—we’re going live on September 23rd, 2026.

We’re going to show you how we’re using AI across real estate investing to increase the capacity of our investment team, analyze opportunities faster, pressure-test assumptions, and make better-informed decisions.

Because the goal isn’t to have AI think for you.

It’s to build a system that helps you think better.

And everyone who joins us live will also get The AI Scale Audit, so you can identify where AI could create the most leverage inside your own investment process or organization.

👉 September 23rd. Live. Free.

There’s no replay.

Save your seat and join us live.