How Neal Bawa Uses AI, Data Analytics, and Market Cycles to Buy Multifamily Deals

Neal Bawa (00:00)
It has become trivially easy to gather very large amounts of information using AI, but there’s only one caution for people that are using AI. AI tends to lie very confidently, right? So many times it just completely off with a bunch of nonsense and it’ll lie to you. So what we ask is that you use both chat GPT and Claude when asking AI to produce data.

And then you compare the results to each other. If the results are consistent, it’s probably the truth.

Michelle Kesil (02:03)
Hey everybody, welcome to the Real Estate Pros podcast. I’m your host, Michelle Kesil and today I’m joined by someone I’m looking forward to chatting with, Neal Bawa, who is the CEO of Grocapitus and is a data-driven multi-family investor. So excited to have you here today, Neal

Neal Bawa (02:25)
Thanks for having me on the show. Very, very excited to be here.

Michelle Kesil (02:28)
Perfect, let’s dive in. First off, for those not yet familiar with you and your work, can you share what your main focus is?

Neal Bawa (02:35)
⁓ For the moment, data-driven real estate acquisitions and construction. We buy and build large properties, mostly in the multifamily and built to rent arena, but we’ve also dabbled with student housing, with industrial flex, and many other asset classes like self-storage. So that’s our primary focus, but the key is to use data and AI to accelerate. ⁓

you know, our business and also grow our profits.

Michelle Kesil (03:05)
Awesome. And which markets do you operate in?

Neal Bawa (03:07)
We operate in markets throughout the United States, currently in 10 states, 17 markets. And our goal is to continue changing our markets based on market cycles. We find that there are no great markets. There are only great opportunities in certain parts of the cycle. And then you’ve got to move on.

Michelle Kesil (03:24)
Can you expand on what that means and how people can figure out what those opportunities look like?

Neal Bawa (03:31)
Sure, a great market typically is going to have five things and those five things are population growth, job growth, income growth, home price growth and crime reduction. If a market has strong numbers in each of those and when I say strong numbers I’m comparing it to other markets. So let’s say if I were to pick a market and say, know, ⁓ St. George, Utah is a very strong market.

then the right way to say it’s strong is by comparing those five things, population growth, job growth, income growth, home price growth, and crime, and comparing them with other markets. Let’s say I compare them to five other random markets in the US. If St. George is number one or number two in the list, then one can say that it’s strong. That’s a data-driven process. Otherwise, most people who pick markets basically pick them based on a water cooler conversation, or they pick them because somebody that they like or respect.

said that there were great markets. That’s not a data driven approach, right? Even if the other person’s data driven, that’s not data driven. That’s you saying, I think somebody else is data driven and I’ll just accept that. So what we do is that we track markets in the U S there are 323 MSAs or metros in the U S there’s about a thousand cities in those metros. So we track those metros and we measure them on all of these benchmarks, these metrics. And here’s what we found. No city or metro stays up at the top.

Michelle Kesil (04:34)
Mm-hmm.

Neal Bawa (04:52)
No city or metro even stays in the top 10%. They go up, they go down. There are cycles. And those cycles are tied to the fact that when more and more investors realize that a city is a good place to invest in, it becomes a bad place to invest in. Why? Because there’s too many people that know this. When too many people know the prices go up, the cap rates go down, rents are rising at 3%, prices are going up at 10%. Well, that’s obviously not a good thing for an investor.

So at some point the city becomes too expensive to invest in, even though it may be a great city. Even though if you’re very patient and have loads of money and have no money that you borrowed and you stay in there for 10 years, you’d probably make a lot of money, right? So nothing against that. But the point is at some point that city becomes very expensive. We also discovered that certain cities are more likely to give you profit than the others. And speaking from the perspective of real estate, I can tell you that today the two biggest things

that are driving us away from a lot of great cities in the US are property taxes and insurance.

Those are your two biggest expenses, right? So what we found is, and this may shock some of the people here, the last 12 months, there was no rent growth in the United States as a country. And most growth markets had negative rent growth, right? Negative rent growth, Texas, Florida, you know.

Tennessee, I mean, lots of growth markets had negative rent growth, Arizona. So negative rent growth for growth markets and then like just regular markets were flat. Very few markets actually had rent growth and most of those were in either in the Midwest or in the Northeast where it’s, you know, very hard to build and nobody’s building anything. So in the last 12 months, rent growth is not much of a driver of profits, right? Cause you know, that’s not happening. So what is a driver of profits? Well, the driver of profits is if you go to markets that have lower

OPEX or expenses, right? Operating expenses. What are the two biggest components today, today in 2026 of OPEX, property taxes and insurance. So we have moved away from some great markets, very good 10 year growth markets, simply because their property taxes and insurance are too high. And it’s that makes it easier for us to generate profits and easier to break even. So like we don’t have any market in Texas right now. it be consider?

to be investable, even though Texas is probably, probably the best state in America to invest in from a demographics and growth perspective. And kudos to those that are investing there, nothing against them. But what we are finding is the combination of high property taxes and high insurance makes Texas very difficult to invest in to make money and make it consistently. So we’re looking at places that have lower property taxes and lower insurance combined with those five factors of, you know,

home price growth, job growth, population growth, income growth, crime reduction. So if we can get those five factors and we can get low property taxes and low insurance, that’s awesome. There’s an eighth factor, which is only when we build. What if the place also has low construction costs? Then in that market, we can build. But if it doesn’t have low construction costs and has the other seven, then we can buy.

Michelle Kesil (08:46)
Yeah, that’s amazing. I don’t think many people look at it from that much data. And are you using AI to gather that information or what does the process look like to get those results?

Neal Bawa (09:01)
It has become trivially easy to gather very large amounts of information using AI, but there’s only one caution for people that are using AI. AI tends to lie very confidently, right? So many times it just completely off with a bunch of nonsense and it’ll lie to you. So what we ask is that you use both chat GPT and Claude when asking AI to produce data.

And then you compare the results to each other. If the results are consistent, it’s probably the truth. If they’re off,

one of the two AIs is confidently lying to you. So be careful, right? The other thing that we are now beginning to do is we are creating AI agents where one AI agent, let’s say Claude, does the job. It goes out and gathers the information that we’re looking for. And then the other agent, let’s say ChatGPT, checks the work of Claude, right? And gives us a thumbs up. So by doing that,

We get around this issue that sometimes AI lies and lies very confidently. But yes, we are using artificial intelligence a great deal. Now keep in mind, many times AI cannot get access to private data. For example, there’s a company called CoStar, which is used a lot in the multifamily industry. Their data is not really going to be available to CoStar. Maybe a little bit of their data is available, but you know, a lot of people have subscriptions. can, you even you don’t need a CoStar subscription. If you’re working with a broker, they have one.

And so if you’re interested in a certain property or sub-market or area or a city, you can get, you know, CoStar reports by asking your broker and they would be more than happy to send a CoStar report to you or number of CoStar reports to you. So the combination of high quality data like CoStar and AI is probably the most powerful. So for example, if you go into CoStar and pick a multifamily property, let’s say it’s 200 units and it’s in Chattanooga.

you can go type in its address and it’ll give you a report. Well, that report now is unfortunately 120 pages long.

Now what you as a company can figure out is what are the pieces of this that we care the most about? Because if anybody says, I care about everything in 120 page report, then either you’re lying to us or you work 14 hours a day. I don’t. Right? I can tell you, I care most today in a report about something known as absorption, vacancy, right?

Those are the things that I care most about. care about rent growth, obviously. And so there are many, many pages of 107 page GoStar report that I don’t care about. So the first thing that we do is that we use AI to build dashboards out of our GoStar reports of only the things that we care about. And we describe those dashboards to Claude and to ChatGPT, and they build beautiful, extraordinarily gorgeous dashboards for us. And they build charts and graphs in real time of the things that we want to see and compare.

By doing that,

we’re going from 107 page report to a dashboard that might only say, take 90 seconds for me to look at. And that allows me to make almost the same quality of decision, but save an hour reading a report, which I might never even get to. Right. The other thing is we’ve automated the process of going out and getting those reports. Right. So we’re at the point where the data auto magically appears to us when we need it to appear. And I think that’s the power of AI where

It’s all about exercising your decision-making muscles and less about exercising, doing all the work to get to the point where decision-making is required. And that’s the magic of AI. And it’s become dramatically better in the last quarter. ⁓ There’s a tool called Cloud Cowork, which we are doubling down on, which is extraordinary in its capabilities. know, ChatGPT has responded very recently with their own version of Cloud Cowork.

⁓ And so we’re looking at both of the tools, but essentially ⁓ what we’re trying to do is to get to the point where it gets the data for us. And if it cannot get private data, like chat, CoStar, then CoStar gets, you know, it gets the data from CoStar and then analyzes it and presents it to us. So the answer is yes, we’re using data and we’re massively using AI to make it easy for us to get to that data and easy for us to understand it, to interpret it, charts, graphs, you know, major numbers.

Michelle Kesil (13:44)
Wow, that’s amazing that you’re able to harness those tools and support your business in that way. And what would you say, like as far as the execution goes of the construction or the multifamily, like are you guys using AI as well for that process?

Neal Bawa (14:03)
Not quite yet. We are finding that when it comes to construction activities, ⁓ it’s not that AI isn’t good. It’s just not at that critical tipping point where the value is incredible. So we expect that within this year, we will be able to build custom AI development agents, right? So remember, we don’t.

Michelle Kesil (14:15)
Thank you.

Neal Bawa (14:26)
We are now all coders in our company because we speak to a software and it does write the code for us. So we’re writing custom agents that basically join our meetings and are learning about our development calls and development activities and acting as a note taker, publishing stuff. Before you could go in and have ChatGPT listen and you could provide things like ⁓ action items and stuff like that. But now we have.

know, agents that do all of this on their own, right? You don’t have to tell them, they will join the meeting, they will listen, they will, you know, ⁓ process the transcript, they will publish action items, and they’re doing all of this on their own. ⁓ So we’ve gotten to that point. I don’t think it’s a very impressive point. I mean, I still don’t find that when it comes to construction, our AI is like, wow, ⁓ it’s still functioning like an executive assistant or a project manager.

as opposed to really giving us insight into what’s going on in the construction world. But we’ll get there. I think that the tools are getting better at an exponential speed.

Michelle Kesil (15:22)
Yeah.

Yeah, absolutely. And what does your process look like in terms of the operational side of the multifamily world? Because you shared a lot about the analytical side, but yeah, how do you operate things?

Neal Bawa (16:23)
So one of the best things that you can do in this age of AI is basically sit down with your team in what we call a dreaming session. A dreaming session is where you do not have to be realistic, you do not have to be ⁓ reasonable, you do not have to be conservative. You dream that you’re a company

that has a thousand employees whose only job it is to improve the operations of your property. You dream that and you write down or in our case, we just speak it, AI is writing it down. ⁓ And you say things like, I wish that every single day we had an agent that could actually look at how many calls each property is receiving and how many times did they pick up and how many times did they not pick up? And it would provide a report of those things. It would basically say, well, here,

this property, the pickup rate is 50%, this property, it’s only 30%, so this property is basically losing a bunch of phone calls. And how quickly is each property processing those? You might say, well, Neal, there’s software for doing this. Yes, but you have to go to the software. We’re talking about AI going into the software, grabbing the information, putting it into a dashboard, putting it into a Slack channel, and providing an analysis, right?

These are things I know that most people don’t do them simply because of laziness or they think that it’s too hard to do. But now we’ve reached the point where we can do all of these activities. So we use a combination of people in the Philippines wherever AI doesn’t work. doesn’t, browser control sucks with AI. It’s something that we’re waiting and hoping for it to get better. So if we need to basically log into an application and grab files, it seems to do it in a very buggy and unreliable fashion. So we still have people doing that.

and grabbing it and then the AI basically picks it up from there and does the rest of the steps. And so what we’re finding is that we are able to do more things with asset management because we’re not property managers. There’s a property management company that we’ve employed and there’s a regional property manager and there’s on-site property manager. So what is our job? Our job is to use their data, the data that they’re gathering

using their property management system and display it as in high quality dashboard in Slack channels that we are sharing with that property manager’s regional and that property manager’s onsite property manager. By sharing this information in a very visible, easy to read fashion, we basically make them naked. We make it obvious what they’re not doing well. So imagine a dashboard that says,

this property is dead last out of our entire portfolio in terms of its pickup rate or in terms of its ⁓ how quickly it acts to pick up calls or what its renewal rates are. It’s got the lowest renewal rates of ⁓ any property in our portfolio. These are obviously all things that are bad news and the property should do better, right?

The problem is all this gets buried. But what if there was a beautiful dashboard showing all of this stuff? And you might say, we don’t need AI to develop it. Yes, you do. Unless you have an unlimited amount of manpower, an unlimited number of people, and all the money in the world, sure, then you can do it. I think that what we are trying to do is we’re trying to use a combination of people in the Philippines and automation to automate it so that this data is very visible and it’s beautiful. you know, ugly data, I find people don’t read it.

unfortunate because that’s their job to read it. But what I find is that when I provide charts and graphs and dashboards and, you know, speed dials, people get it. And that puts pressure on them without me saying things like, Hey, Michelle, you are not doing your job. Right. I’m simply comparing them to other properties and saying, Hey, you’re dead last in all, you know, four of my five metrics. So you know that

when it’s time to renew the property management contract, that’s probably not going to happen unless you pick it up and you’re in the top two.

Michelle Kesil (20:28)
Yeah, amazing that you’re able to use all these resources to gather data and have such an analytical perspective on real estate.

Neal Bawa (20:37)
I think analytics has nothing to do with AI. Analytics has been around for decades. We’re just, it’s now becoming commoditized. It’s becoming easier. if you’re like, I mean, we’re a company with, you know, dozens of employees, but I think that it’s now possible for everything that I’ve described today to be done in a single person company or maybe one in one person in the U S and one person in the Philippines.

Michelle Kesil (20:40)
Mm-hmm.

Neal Bawa (21:05)
and then just the daily creation one to two hours a day of systems and processes to automate. And if you continue that one hour a day meeting every day for a year, you would end up with the same level of work or close to the same level of work as our 20 person company.

Michelle Kesil (21:06)
Mm-hmm.

Yeah, that’s an amazing way to look at it. Thank you for sharing your perspective.

Well, before we begin to wrap up here, if someone wants to reach out, connect and learn more about what you’re up to, where can people find you and connect with you?

Neal Bawa (21:38)
⁓

I think the best place is multifamilyu.com/club. So don’t forget the U multifamilyu.com/club. So we think of ourselves as a Wikipedia of real estate information. And as you know, Wikipedia is free. There are no subscription tiers, so we don’t have them either. We don’t have an educational product. We don’t sell or upsell you to anything. Roughly 1 % of the people that are in our club have become investors in our projects. The remaining 99

continue to receive high quality information around the year, around the clock on all of the things that we are doing. And not just with AI, mean, things that we’re doing with different kinds of real estate, know, what’s happening in industrial, what’s happening in Airbnb, what’s happening in self storage, what’s happening in student housing, right? And of course, what’s happening in multifamily and single family, which are the two core markets that we operate in. We do high quality 60 slide.

presentations. They are the most gorgeous presentations you’ve ever seen. And I guess we do use AI heavily to build them. But we actually have a human being that improves them as well. And those presentations are viewed by thousands and thousands of people every year. And a small percentage, roughly 1 % of them, invest with us. Everyone else is just there for the learning.

Michelle Kesil (22:55)
Awesome, well thank you for sharing that. I appreciate your time and your story and thanks for coming on here.

Neal Bawa (23:01)
Thanks for having me.

Michelle Kesil (23:01)
For the listeners tuning in, if you got value, make sure you’ve subscribed. We have more conversations with operators like Neal who are building real businesses and we’ll see you on our next episode.