EP 380 I Let ChatGPT Analyze a $60 Million Apartment Deal

APPLE PODCASTS | SPOTIFY

A few years ago, if someone had told me I could upload a fifty-page apartment offering memorandum into a computer and receive an investment analysis in under two minutes…

I probably would’ve laughed.

Back then, underwriting a deal meant late nights staring at Excel spreadsheets.

Cross-checking rent rolls against financial statements.

Trying to figure out why one number on page 37 didn’t match another number on page 12.

When Nancy and I bought our first apartment community—a 76-unit property in Phoenix—I wasn’t fast.

I wasn’t confident.

And I certainly wasn’t an expert.

I was simply willing to keep asking questions until I understood what I was looking at.

Today, artificial intelligence can answer those questions faster than I ever could.

But after using AI to analyze a real $60 million apartment deal, I realized something surprising.

The tool wasn’t the competitive advantage.

The questions were.

AI Can Read Faster Than Any Human

One of the most incredible things about today’s AI models is how quickly they process information.

Offering memorandums.

Rent rolls.

Trailing twelve-month financials.

Loan documents.

Hundreds of pages that once took us hours to review can now be summarized in minutes.

That changes everything.

Or at least it seems like it does.

Because speed isn’t the same thing as judgment.

The First Conversation Looked Perfect

For this experiment, I opened a brand-new AI chat.

I uploaded the broker’s offering memorandum.

Then I asked what most people would naturally ask.

“Summarize this investment opportunity.”

The response was impressive.

Strong location.

Excellent financing.

Compelling value-add opportunity.

Nine out of ten.

If I had stopped reading there, I probably would’ve felt excited.

Confident.

Ready to keep moving.

The problem?

AI wasn’t analyzing the investment.

It was summarizing the marketing.

And those aren’t the same thing.

Marketing Documents Are Designed To Sell

Offering memorandums exist for one purpose.

To present the property in the best possible light.

That doesn’t mean they’re dishonest.

Far from it.

Most contain accurate information.

But like every marketing document, they emphasize strengths before discussing risks.

When I asked AI to summarize the document, it did exactly what I requested.

It organized the seller’s story beautifully.

It wasn’t wrong.

It simply wasn’t skeptical.

Everything Changed When We Asked Different Questions

Instead of starting over with another opinion…

We started over with another conversation.

This time we uploaded one additional document.

The property’s actual trailing twelve-month financial statements.

Then we changed our first question.

Instead of asking:

“Summarize this opportunity.”

We asked:

“Compare the broker’s projections against what actually happened.”

The conversation changed immediately.

Suddenly AI found inconsistencies.

Vacancy assumptions that didn’t match historical performance.

Property taxes that would nearly double after closing.

Projected rent increases that required far more operational work than the marketing implied.

Same AI.

Same property.

Completely different conclusion.

AI Didn’t Become Smarter

We did.

That was probably the biggest lesson from the entire experiment.

People often ask whether AI is replacing analysts.

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

AI is replacing analysis.

It isn’t replacing judgment.

Judgment comes from knowing which questions deserve to be asked.

And that only comes from experience.

Or learning from someone who already has it.

Every Great Investment Begins With Skepticism

One thing we’ve learned after underwriting hundreds of apartment communities is this:

Good investments survive difficult questions.

Weak investments don’t.

Every sponsor can explain why their deal deserves your capital.

The better question is:

What would make this investment fail?

That’s where underwriting begins.

Not with confirmation.

With curiosity.

The Difference Between Information And Wisdom

One thing AI gives us is extraordinary access to information.

What it doesn’t automatically provide is wisdom.

Wisdom asks:

What assumptions are hiding underneath these numbers?

What happens if occupancy doesn’t improve?

What if taxes increase?

What if renovations cost more than expected?

What if market rents never reach the projections?

Those questions don’t come from software.

They come from pattern recognition.

And that’s exactly what investors spend years developing.

Why We Passed On The Deal

By the end of our analysis, AI estimated the property’s value was closer to $42–44 million.

The asking price?

Nearly $60 million.

That’s a gap of more than $15 million.

Could the seller still find a buyer?

Absolutely.

Would we become that buyer?

No.

Not because AI told us to pass.

Because AI helped us uncover the questions that changed how we viewed the opportunity.

That’s a very different thing.

The Real Asset Isn’t Artificial Intelligence

If there’s one lesson we hope investors take away from this experience, it’s this:

Artificial intelligence isn’t your competitive advantage anymore.

Almost everyone has access to the same models.

The real advantage is knowing what to ask after the first answer appears.

The software is becoming available to everyone.

Judgment is still incredibly rare.

And in investing, judgment has always been the asset that compounds the most.

Because the tool may be free.

But the questions?

Those are priceless.

I’m running a free masterclass — live, with me — where I walk you through the full underwriting framework Nancy and I built over eight years and half a billion dollars of deals. Every category. The exact questions. The exact order. On a real deal, start to finish, so you can see the process end to end.

It’s free. And it is, without exaggeration, the single most important ninety minutes of preparation you can do before you write your first — or your next — check.