EP 386 AI Is Making Bad Investors Worse

APPLE PODCASTS | SPOTIFY

When we first got into multifamily real estate, we knew absolutely nothing.

And we mean nothing.

We didn’t come from Wall Street.

We weren’t investment bankers.

We weren’t software engineers.

We didn’t go to school for artificial intelligence.

We came from fashion.

Then, over the next seven-plus years, we went from complete beginners to building a real estate portfolio approaching $500 million, raising more than $130 million in capital, helping create more than $110 million in tax savings, and returning more than $45 million in capital to investors.

But here’s the part people don’t always see:

It took years.

Years of looking at deals.

Years of trial and error.

Years of learning markets.

Years of building teams.

And perhaps most importantly, years of learning which questions to ask.

Then AI showed up.

And suddenly, we’re watching technology compress work that once took us hours—or sometimes days—into minutes.

Which sounds incredible.

And it is.

But there’s a catch.

AI Can Make You Faster. It Can Also Make You Wrong Faster.

There are two big misconceptions we keep seeing about AI and real estate investing.

On one side are people who think:

“AI is going to replace me.”

On the other are people who think:

“AI is going to do everything for me.”

Upload an offering memorandum.

Ask ChatGPT whether it’s a good investment.

Receive a beautifully formatted answer.

Congratulations! You’re now a sophisticated real estate investor.

Except…

That’s not how it works.

AI doesn’t replace the investor.

And AI doesn’t magically give you investment judgment.

We like to explain it with ridiculously simple math.

What’s zero multiplied by $950 trillion?

Still zero.

AI works similarly.

If you don’t understand what makes a good deal…

If you don’t know which assumptions should be challenged…

If you don’t understand the risks…

If you don’t know which questions deserve another question…

AI can’t magically give you seven years of investment judgment.

It can multiply what you already know.

But zero multiplied by anything is still zero.

So How Do You Know When AI Is Wrong?

This was one of our favorite questions from a recent Q&A.

Because AI can sound incredibly convincing.

It can give you a number.

Explain the number.

Put the number into a beautiful table.

And still be completely wrong.

We recently experienced this on one of our own properties.

AI flagged roughly $99,000 of additional accounts payable that appeared to have accumulated in just 19 days.

That sounds terrifying.

Imagine seeing that notification:

“Wait… where did an extra $100,000 of expenses come from?!”

Except that wasn’t what happened.

We looked at the actual T-12 and immediately recognized what AI was interpreting as unusual growth.

Those were normal monthly expenses.

The AI had found something.

It simply didn’t understand what it meant.

And if we didn’t know how to read the financials ourselves?

We could’ve confidently accepted the wrong conclusion.

That’s why we say:

AI can support your brain. It cannot be your brain.

And this is exactly why we created The 9 Biggest AI Mistakes Real Estate Investors Make That Cost Them Money, Deals, and Their Future.

Because AI can make you faster.

But used incorrectly?

It can also make you wrong faster.

We put together the nine mistakes we’re seeing real estate investors make with AI—and what to do instead—into one free blueprint.

If you’re already using ChatGPT, Claude, or another AI tool to analyze deals, research markets, underwrite properties, or make investment decisions, you can grab your free copy at JumpstartApartments.com.

It might help you catch the mistake you don’t know you’re making yet.

Can You Just Upload the Entire Deal to ChatGPT?

Technically?

Yes.

You can upload an offering memorandum, rent roll, T-12, market study, and other documents and ask AI to analyze them.

But that’s not necessarily how we’d do it.

Here’s why.

Real estate deals change.

Imagine it’s August and the seller sends you a T-12 through July.

You start underwriting.

A month passes.

Now there’s another month of operating data.

You have an updated T-12.

Something changed.

Occupancy moved.

Collections changed.

Expenses changed.

You don’t want to start from zero every time new information arrives.

That’s why we increasingly think about AI not as a chat, but as an investment team with context.

We’ve built our own system around nine different AI seats that can analyze different parts of an investment while retaining the context of how we invest.

The point isn’t simply getting AI to read documents.

It’s getting AI to understand what those documents mean relative to everything else.

Sometimes What’s Missing Matters More Than What’s There

We were recently analyzing a property when something didn’t make sense.

The property was barely around 90% occupied.

Yet the financials appeared to show essentially no delinquency.

Could that happen?

Sure.

But after operating apartments for years, our reaction was:

Where is it?

Where are they recording the delinquency?

Is it being categorized differently?

Is something missing?

That’s the kind of question experience teaches you to ask.

A beginner might look at zero delinquency and think:

“Amazing property!”

An experienced operator might look at the exact same number and think:

“That seems unusually good. Show us why.”

That’s the difference between reading data and interpreting data.

And it’s one of the reasons we believe AI becomes dramatically more powerful when it’s paired with judgment.

Stop Asking AI, “Is This a Good Deal?”

This may be the biggest change we’d make if you’re already using AI to analyze investments.

Stop asking it to make your decision.

Instead, use it to improve your analysis.

At a large private equity or real estate investment firm, a deal doesn’t usually go directly from someone’s inbox to a yes or no.

People analyze the financials.

Someone studies the market.

Someone looks at debt.

Someone looks for risk.

People debate.

Then the decision-maker considers all of it.

That’s much closer to how we think investors should use AI.

Not:

“Should I invest?”

But:

“What am I missing?”

“Which assumptions aren’t supported by the historical financials?”

“What would need to happen for this business plan to fail?”

“What questions should I ask the broker next?”

Those are very different conversations.

The Best Deals Aren’t the Ones With the Most Reasons to Say Yes

When we first started investing, we loved every deal.

We’re barely exaggerating.

We’d find an opportunity and immediately start seeing everything that could go right.

Our first passive investment?

We didn’t even wait for the webinar to finish.

We were basically saying:

“Here’s our money!”

It’s kind of like owning a dog.

Someone tells you:

“Your dog eats poop, bites people, and destroys the furniture.”

And you’re sitting there thinking:

“No. Our dog is perfect.”

Investors do the same thing.

Once we like a deal, we naturally begin defending it.

That’s why our thinking has changed.

We’re no longer trying to accumulate reasons why we should invest.

We’re trying to eliminate reasons why we shouldn’t.

What could go wrong?

Which assumption breaks first?

What’s the seller not emphasizing?

What happens if rent growth doesn’t happen?

What happens if expenses come in higher?

AI is incredibly useful here because it can become another set of eyes—especially around financials and quantitative analysis—without becoming emotionally attached to the deal.

Passive Investing Shouldn’t Mean Passive Decision-Making

This matters just as much if you’re an LP.

Maybe even more.

A passive investment should become passive after you’ve decided to invest.

It shouldn’t be passive while you’re deciding.

You still need your own criteria.

What returns are appropriate for the risk?

What vacancy assumptions are being used?

What rent growth does the sponsor assume?

What evidence supports it?

Are the comparable properties actually comparable?

Does the business plan make sense in the real world?

And most importantly:

Who is operating the deal?

When we invest passively ourselves, one of our first instincts is to validate the GP—the operator.

Because you’re not only investing in the property.

You’re piggybacking on someone else’s experience and ability to execute.

Being a passive investor doesn’t mean handing over your money and your responsibility at the same time.

The Documents We Want AI Looking At

For a multifamily acquisition, some of the basic documents we want to understand include:

The offering memorandum.

The trailing 12-month financial statement, or T-12.

The rent roll.

Property-tax information.

Insurance information.

Major service contracts.

But here’s where AI gets interesting:

Don’t only ask it to read each document.

Ask it to cross-reference them.

If the rent roll says the property has a certain occupancy level, does the revenue on the T-12 roughly support that?

If the seller says insurance costs $X, does the historical expense actually show $X?

If the offering memorandum claims rents can increase significantly, what evidence supports that assumption?

We’ve looked at deals where something as basic as receiving a three-month-old T-12 immediately raised another question:

Why don’t we have current financials for a sizable acquisition?

The missing information itself becomes information.

And those are exactly the kinds of questions we want AI helping us surface.

Which AI Should Real Estate Investors Use?

ChatGPT?

Claude?

Gemini?

Grok?

We use different tools for different things.

But we think investors spend far too much time obsessing over this question.

Because the model isn’t the most important part.

Your ability to tell it what good looks like is.

One AI might be faster.

Another might be stronger with certain spreadsheets.

Another might have functionality you prefer.

Those things will also change constantly.

The durable skill is knowing:

What are we trying to accomplish?

What information does the AI need?

What should the answer approximately look like?

And how will we verify it?

Tools change. Judgment transfers.

One AI Project Helped Us With a Real Insurance Claim

One of our favorite examples actually wasn’t underwriting.

We had a fire at a 272-unit apartment community that affected eight units.

We had business interruption coverage designed to compensate us for lost revenue.

But there was a complication.

At the time of the fire, only two of those eight apartments were occupied.

So we needed to build a case explaining why the loss should account for all eight units based on the property’s demand and operating circumstances.

We used Claude to help us build the supporting documentation.

It still required our judgment.

We reviewed it.

Came back to it with fresh eyes.

Reviewed it again.

Then we submitted it.

Eventually, the adjuster approved the full amount we’d requested.

And the part that made us particularly proud?

They’d sent our work to a forensic accountant for review.

That’s the kind of AI use case that excites us.

Not just:

“Write us an email faster.”

But:

“Help us perform work that historically would’ve required significantly more time, people, or outside expertise.”

AI Can Help You Understand Markets You’ve Never Lived In

Nancy and I don’t live in most of the markets where we own real estate.

We live hundreds—or sometimes thousands—of miles away.

So when we’re studying a market, we need to understand things like:

Population growth.

Net migration.

New supply.

Economic drivers.

Landlord friendliness.

And the other factors that influence whether people will continue wanting and being able to live there.

We’ve even built a Claude plugin that helps pull together much of the market information we typically review.

Again, AI isn’t making the investment decision.

It’s compressing the work required to assemble the information needed for one.

The Real AI Advantage Isn’t Knowing More

Here’s what we’ve realized after using these tools across our actual real estate business:

Information isn’t the scarce resource anymore.

You can ask AI what a cap rate is.

You can ask it how DSCR works.

You can ask it to summarize a market.

You can ask it to read a T-12.

You can ask it to compare a rent roll against historical revenue.

That’s incredible.

But information and judgment aren’t the same thing.

AI can hand you an incredibly sophisticated answer that is completely bogus.

And sometimes the most dangerous answer isn’t one that obviously looks wrong.

It’s the wrong answer that looks right.

That’s why we don’t believe the future belongs to investors who simply use AI.

We think it belongs to investors who develop enough judgment to know when to trust it, when to challenge it, and when to ignore it.

AI Doesn’t Replace Investment Judgment. It Multiplies It.

If there’s one idea we hope you remember, make it this:

AI doesn’t magically turn beginners into sophisticated investors.

But it can help investors learn faster.

Analyze faster.

Cross-reference faster.

Research faster.

Challenge assumptions faster.

And potentially catch things they otherwise would’ve missed.

That’s the opportunity we’re excited about.

Because we know how long it took us to learn this business.

Seven-plus years of deals, mistakes, questions, experience, and pattern recognition.

If AI can compress parts of that learning curve for the next investor?

That’s extraordinary.

But don’t confuse compression with elimination.

You still need to understand the fundamentals.

You still need your own investment criteria.

You still need to know what you’re looking for.

And ultimately?

You still have to make the decision.

AI can sit beside you at the investment committee table.

It just shouldn’t be sitting in your chair.

Before You Let AI Analyze Your Next Deal…

There’s one more thing we want to give you.

We created a free blueprint called:

The 9 Biggest AI Mistakes Real Estate Investors Make That Cost Them Money, Deals, and Their Future

Because the biggest danger with AI isn’t that it won’t give you an answer.

It’s that it will give you a very convincing wrong answer.

We’ve learned that AI can help us analyze financials, cross-reference documents, research markets, challenge assumptions, and accomplish work that once required considerably more time and people.

But only when we know how to use it.

So we put together the nine mistakes we believe real estate investors need to avoid as they start integrating AI into the way they invest.

You can grab the blueprint for free at JumpstartApartments.com.

Because AI can make you faster.

But if you use it incorrectly, it can make you wrong faster, too.

And ultimately, that’s the lesson behind everything we’ve shared here:

AI doesn’t replace investment judgment. It multiplies it.

So before you ask AI to tell you whether your next deal is good, learn how to make sure you’re asking it the right questions in the first place.