Transcript
Can ChatGPT help me with the feasibility of a development project? I don't know, let's find out. In my free time I often like to invest in real estate, and my latest interest has really been around the possibility of doing a subdivision of a lot and the potential to build out some of the other parcels. Now I don't know a ton about this, but I know enough and I have enough people in the industry where I think I could figure it out. But I need a jumping off point and I need to figure out if the deals that I'm looking at will actually make sense.
When it comes to development there's a ton of information you need to figure out from the county in terms of building codes, requirements, setbacks, a bunch of stuff that I've never done on my own. And so I thought this would be a good use case for ChatGPT, and specifically I wanted to try out their o1 model given that it's a little bit more powerful, a little bit more comprehensive, and I think it's really good when it comes to deeper research and fact-checking the things that it's actually presenting to you.
So I wanted to run through an example of actually going through this exercise. I was actually amazed the first time I tried this, and so we're going to go through that same flow and see what it gives back to us. Ideally by the end of it we can have an idea of whether or not this could be a viable deal and something to pursue further and actually maybe even put an offer on. This is legitimately a real world thing that I'm looking at maybe pursuing, and so I think it's a useful real world application of how you can leverage some of these models. So let's go ahead and take a look at it.
So this is the deal. This is in Charlotte, North Carolina, and you can't really see much here but it's essentially this little green box -- it's about an acre of land and there's one little house on there in the middle. So the question would be, is this something you could actually subdivide and do something higher density? And then coming down in here they actually give some highlights on the actual building, and so we see that it's N2-B zoning, which I know nothing about. It's bringing in about $2,000 a month and it's got city water and sewers.
So off the bat I know those are good indications. I know that if it's a flat, non-sloping lot, from a development perspective it should be easier to work with. And by the way, the seller is interested in maybe seller financing this. So the question here would be, okay, where do I even start? I have a little high level knowledge on what it would take, but I'm going to actually go to ChatGPT and figure out what would be possible here. They want $600,000 -- is that a good price and what could we possibly do with this?
So we're going to hop over here, and this is what I already drafted up in terms of my prompt. I'm just saying something to the effect of: I'm exploring the highest and best use for a piece of land in Charlotte, North Carolina. Zoning is N2-B, whatever that means. The lot is roughly one acre and here are the dimensions. How do we put together a development plan for the lot and financial model? They're asking for $600,000. There's currently a single family house on the property, three beds, two baths. Let's hit go.
Now with o1, it is a more comprehensive model. It is chain of thought reasoning, and so the idea here is it's going to take a little bit longer because of what it's actually going through on the back end, but it's going to give you a robust answer in terms of what you're actually trying to answer.
So let's see what it said here. Here's a step-by-step breakdown, and what's awesome is it actually goes deep into immediately N2-B zoning. You can see that I should see some citations in here where it's probably referencing the city ordinance known as the UDO. But already you can see how comprehensive it is. So let's just scan through and see what it's telling us to do here.
Main highlights: you can do single family, you can do duplex, triplex, quads even, or maybe a small scale multifamily. That could be interesting. It talks a little bit about the lot sizes into two districts and what the requirements are there. Talks a little about density and it calls out the UDO regulations. And then it starts to go into options of how you might be able to develop this.
Single family -- it says you could do this into four or five lots. Given that this is an acre, that could be a possibility. However, you're not going to get the same level of high density -- it actually calls that out here, lower density means lower overall profit. Then it talks a little bit about the small multifamily approach here and it gives you a couple ideas in terms of what you could do. You could probably do two or three duplexes, or you could do two quads, eight in total. That's pretty interesting.
And then we go into townhomes or row homes. I actually like this concept a lot, but again you're only getting five to eight units. Ideally when you're doing this kind of development, my preference would be I would want higher density and more units. Doing something like build-to-rent would be incredible, but really just depends on what the numbers look like.
Then it talks a little bit about other considerations. Higher density has other concerns around stormwater and management. There's going to be more sort of horizontal development that you're going to have to sort through. And then it starts to get into here -- it's talking about the townhomes and a potential layout, and then it's doing the same down here.
And then this is probably the coolest part. So now it's actually getting into the feasibility and the financial model of what we've given it so far, and it's actually already making some assumptions. Now this is probably something that needs to be dialed in and refined -- it's probably not the full solution. But for someone coming into this without the sort of historical background and actually how to think this through and just knowing what prices should be, what costs should be -- this is at least a starting point of, hey, there could be something here. It's worth sharpening the pencils. Maybe it's worth reviewing with somebody else, and actually I could frame up a proposal or a plan that I could then put in front of somebody to figure out, could we make a deal out of this potential piece of land?
So here it's actually breaking down the rough proforma, and this is taking the townhome scenario as an option. This seems like it feels to me pretty reasonable. So 6 units, 1,800 square feet. It's assuming an average sales price of $250 a square foot. I think that's pretty reasonable for the Charlotte area, especially this particular location -- maybe a little bit lower, $225, somewhere in that neighborhood. But that seems pretty reasonable. Assuming the selling cost, and it's assuming sort of the top line -- so it's already kind of like, what is it that you're after, what is the potential net sales?
And then also helping you contemplate what would the cost be going into something like this. Including the land purchase, it's assuming $150 a square foot to actually develop it. That actually also seems pretty reasonable. I would expect somewhere low-end finishes maybe $125, higher-end finishes maybe pushing closer to $200. Ballpark, I feel like that's in the right neighborhood.
And then it starts to get into soft costs. This seems like some of the site prep pieces that it's assuming. These are the areas that I think you'd probably want to sharpen the pencil on and figure out, are these numbers dialed in or not? There's probably room between the land purchase, the soft costs, and then the financing carry cost. And at a minimum I could actually go and model each of these out to actually start to formulate the model for this particular project.
And here it's already giving me an indication of -- hey, you're going to lose five grand on this overall based on these numbers. So your potential revenue is $2.5 million, your total cost $2.57 million -- versus $2.65 million. So net, this is a losing deal. But based on that, maybe there's some room, or you have to think through some of the other approaches here.
This is something we can actually start to play around and ask some questions with -- ways to improve feasibility. So it talks about adding more units. In their model they're only giving you basically six; maybe if you could push this to eight, this could actually work out. It then starts to walk through the due diligence of this, and you can see it's mapping out essentially: you've got to do a survey and site analysis, you've got to talk to city planning, you need an architect, an engineer -- all of these are the things that we could expect.
And this is the approach I would take. I would probably JV with the builder as long as I could pencil out the numbers and prove out that there's some sort of business case around this overall. And then here it puts a bow on everything and summarizes everything that it just went through. Basically it seems like it is a net loser on this analysis, but let's figure out if there are some other opportunities in here.
Okay, so now I'm going to ask it what is the highest density option we could pursue, and I'm actually going to also include an aerial photo of the property and see if that helps it in its analysis. And let me see if I can go find the screenshot. Let's go ahead and drop this in there and see what it's going to tell us.
Now I don't know if it'll catch the road frontage that it has there, and ideally it can comment on what's going on in the image or what it's surmising to be the situation with this particular lot. So after a couple seconds it's actually spitting out now the answer to our question, and it's starting to think through what is the highest possible approach that we could take in terms of density.
Here it's now referencing the UDO regulations for this particular area. It is now looking at some of the main considerations that you need to be thinking through -- what is the minimum lot width, frontage that's required, setbacks, parking requirements, open space and trees, stormwater -- all the things that as a builder you need to be thinking through. And it's already projecting it's going to be anywhere from 6 to 12 units, and the devil's going to come in the details in terms of actually sussing all of that out.
And then if we scan down some more here, there are some questions around is there an option to rezone and what that would entail as well. It looks like the current zoning would maybe not support this high of density, and so one of the considerations we'd have to think through -- now I actually like this approach. So here it's talking about maybe even subdividing it into essentially three parcels and then putting three to four units on each, so you're getting 12 units total. Which I actually really like this concept -- it's essentially like a build-to-rent model.
And so if roughly our overall cost is going to be call it $2 to $2.5 million, and if I can see rents on a monthly basis around like $25,000 across the 12 units, then that could be pretty material. So overall, 12 is probably going to be roughly the max that we could do here. Outside of that you probably have to rezone, which is probably not something I want to bite off for my first project.
So overall I think it's a really powerful use case -- stepping into something where I don't have a ton of background or experience. The time it would take me to answer all these questions on my own, to do this level of research and get to some of these answers to then help me make some decisions about what steps I want to take next, is incredibly powerful.
And here I just wanted to highlight one of the use cases specifically for ChatGPT's o1 model, which is probably not your everyday model just because it's slower. You can often get the same performance and results out of some of the faster models with getting the same quality of responses. But for something like this, where it needs to go a little deeper, think about a complex problem, and actually work out all of the nuances and details and do some level of research to actually put the full solution together -- this is where I think something like o1 becomes really useful.
I could also imagine this application to something like business acquisition. The YouTubers these days, whether it's Cody Sanchez or someone else out there that's talking about buying boring businesses as an investment -- frankly I could look at six different businesses, they all operate differently, and if that's a path I actually want to pursue, I know nothing about a laundromat or where to start and how to run that business. But you could leverage a tool like this to help bridge that gap, give you a better jumping off point, and at least help you put together a plan that you could then work towards executing and filling in the details over time and move that much faster in terms of getting closer to whatever your goal may be.
Hopefully this is a helpful application of how you might be able to leverage ChatGPT and specifically the o1 model. I love playing with these models, and the more I actually use them and interface with them, the more it becomes ingrained into my everyday workflow. So hopefully that's helpful -- until next time.