‍“Hi there, I’m interested in leasing an apartment for myself and my two roommates. Can you tell me about the 3-bedroom units you offer? What are the prices for those? Great, thank you. I’ll consider that. You know, one of the guys I don’t really like that much, so we may just have two roommates. What 2-bedroom units do you offer? And what’s the pricing on those? Uh-huh, yes. Got it, thanks so much. Well, you know, I am also considering just moving in by myself and not dealing with roommates at all. Avoid the drama, you know what I mean! I know, I know, roommates are fun, but also a LOT. Ha, right. Okay, so could you walk me through your studio and 1-bedroom options? Great, and what are the prices on those? Awesome, thanks so much. And these were all for 12-month leases, right? That’s all you offer, got it. Okay, I think I have all I need to make a decision. Sure thing, I’ll let you know what path I decide to go down. Have a great day.”

This is what I heard day in, day out from my two deskmates at my old real estate firm. They were constantly calling apartment complexes to find out pricing, whether it was to evaluate an acquisition or new build or to update our existing properties’ pricing strategy. One time, I even overheard my coworker pretend that he was looking for an assisted living unit for his mother and father, only to say that he didn’t think his father would be around too long, so could she please also walk him through single-bed options. Morbid, right!?

Back then, underwriting meant armies of analysts dialing every property in a 3-mile radius of the target, pretending to be a prospective renter, and then hand-stitching together rent rolls and Excel files. These types of cold call market surveys have been all too common in our industry. That is, until now. 

From cold calls to code

Nowadays, nearly every apartment community publishes its pricing online. Renters do most of their prospecting online and can get most, if not all, of their rental information online. Which means the days of pretending to be a student attending Notre Dame next year are (at least mostly) gone. 

AI can perform a comp analysis for you in minutes. Don’t believe me? Check out my colleague from HelloData, Marc Rutzen’s, viral post on LinkedIn, and his whitepaper on how to automate the comps section of an investment memo using AI. 

And now, the next level of analyst work — underwriting — is on the table for AI to automate.

The question I wanted to test is: Can AI build not just a comps table, but forward-looking underwriting inputs? Specifically, can it predict rents for a value-add renovation?

Spoiler alert: Yes. It’s not perfect (you still need to confirm assumptions and adjust formatting a bit), but it’s significantly faster to edit than to start from scratch. You can even have your AI pull the formulas into Python or Excel so you can recreate the scenario exactly. I’m intentionally not creating an AVM (automated valuation model) because I know how distrustful non-programmers can be of code hidden behind the outputs. Ironically, that’s exactly how Excel works, while other forms of coding, like Python, show your work up front. Regardless, AI is not a replacement for judgment. It’s a power tool for it.

Underwriting is your secret sauce, and you need to understand how the sausage is made. So let’s break it down.

Let’s say I’m interested in acquiring Park Terrace Apartments just south of downtown Austin, right behind Peter Pan Mini Golf, for those of you who know the area. It’s a Class C property that, in theory, could generate even higher rents with renovated units. But is the ROI there? Would a renovation achieve the types of rents necessary to justify the capital investment?

Let’s find out.

What AI just did for underwriting 🤯

I gave it one property name and address (Park Terrace Apartments in Austin, TX) and a set of comps.

14 minutes later → a forward-looking underwriting forecast that normally takes hours, if not days.

What it delivered:

• Auto-detected statistically similar comps.
• Pulled effective rents down to the unit level.
• Fit a model that ties rents to quality (class), floor plan, vintage, and location.
• Forecasted baseline rents vs. renovation rents month-by-month.
• Blended in a renovation schedule (30% units upgraded in 6 months, another 20% in the next 6).
• Produced a full 12-month rent forecast table and chart, exportable to CSV or Excel.

This isn’t some AI black magic. It’s investor-grade analysis with data sourced from 250,000+ property sites updated daily, and with transparent Python + Excel formulas you can rerun yourself. It delivers floor plan-level forward projections, not just the “3% rent growth” shortcut. And most importantly, there’s no manual data entry or scraping headaches.

Early adopters of AI and automated data analysis aren’t just cutting grunt work — they’re running circles around the competition.

Here’s how I tested it with two simple steps:

1. Query the data in HelloData → I used the HelloData Query Builder to pull rents over the last few years for each property in the surrounding urban zip codes. I included criteria like quality score (HelloData’s proxy for property class), year built, and more. Then, I exported the data to CSV.

2. Prompted my favorite AI model → I then uploaded the CSV export to my work AI and gave it the following prompt: 

“Use this spreadsheet to forecast average rents for Park Terrace Apartments over the next 12 months. First, figure out how rent changes with quality, floor plan, year built, zip, seasonality, and time trend. Tell me what you find. Then project rents assuming no renovations, and compare that to a scenario where 30% of units are upgraded to a quality score of 75 in the first 6 months and another 20% over the next 6 months. Show me a line chart that compares no renovations to renovations. Also show the monthly results in Excel with formulas included (so I can change the assumptions), and share the Python code you used so I can re-run it myself.”

And here’s what the model returned:

• Renovated units rent ≈ $400 more per month than unrenovated units (with predicted rents ~$200 higher overall).
• By month 12, property-wide average rents run ~14% higher vs. baseline.
• All delivered in minutes, with the code and Excel formulas I could reuse.

Of course, a $400/month unit uplift means nothing if it costs too much to do — and with interest rates and tariffs, there’s a lot more to the renovation decision today than there was even a year ago. But that’s a post for another day.

So what?

So yes, AI can now do in minutes what your analysts used to do in weeks. That doesn’t mean you fire the analysts — it means your analysts are underwriting circles around someone else’s analysts, and you’re winning deals while the other firm is still pretending to be “three guys looking for a 2-bedroom near campus.” In this market, speed isn’t a nice-to-have; it’s the winning edge.

Want to see how it works? Schedule a full product demo with the HelloData team!