In this episode, the backend engineering for LandPricer’s AI-powered contour analysis is revealed, achieving 95%+ accuracy without any manual prompting. The system autonomously determines buildable areas and evaluates contour impact through sophisticated decision-making criteria that account for gradient analysis, neighboring property improvements, and property-specific use cases across different regions and terrains.
Key Takeaways:
- 10-Foot Gradient Detection Within Buildable Areas The AI identifies whether anticipated buildable areas have greater than 10-foot elevation changes by analyzing contour line numbers, neighboring structures, and multiple building site possibilities.
- Conservative False Negative Preference Over False Positives The system deliberately overestimates slope severity rather than underestimating it, protecting against the far greater risk of acquiring unbuildable steep properties versus missing flatter opportunities.
- Contour Relevance Varies By Primary Use Case If properties are designated for recreational, agricultural, timber, or hunting purposes rather than residential development, contour analysis becomes effectively negligible in the valuation calculation.
Listen to the full episode to understand the complete AI decision framework behind automated contour assessment.
(Podcast transcript below)
Welcome to get serious wanted to provide some more insights on how we are designing land Pricer and some of the backend engineering, how we’re handling image analysis, for instance, that’s all handled by AI. You don’t need to prompt anything. You don’t need to do anything other than provide some of the images. And those are very simple to do instructions using a mapping software.
relate land ID as the default here. So I’m to be talking about contour today. You know, one of the most important parts of considering value for land. know, again, you have to really boil it down. What’s the most important factor? I think, you know, when I chatted the other day about, you know, road access, that’s probably I would still consider the number one largest factor because if you can’t access
property, regardless of what its use is, pretty much has no value, at least in the short term, unless you can really work some magic, find an easement, or buying it for speculation, additional county-led development over time, or some private developer acquiring land in the area to potentially work with. But if you don’t have that, just…
It’s almost impossible, you know, contour is right up there. It’s pretty much a standard overlay that I will have on when I’m looking at any mapping software. And obviously some areas of the country, this is going to be more relevant than other areas. Like if you’re operating in Florida, contour is almost never an issue. You know, instead it’s like much more of the wetlands, the flood zone situation. But I would say in most other states,
you know, outside of coastal areas. Oftentimes, you know, East Coast tends to be flatter, know, West Coast, even close to the ocean, you know, can be more hilly. So most of the U.S. you’re going to be dealing with some type of contour considerations before purchasing a property. And so this is arguably the most difficult.
yeah, I’ll, yeah, it definitely was the most difficult, image analysis, AI prompt to work out just because there’s, there’s so much nuance to it. Right. So, you know, it’s really two core questions. One, I just want to understand within the anticipated buildable area for the subject property, is there a larger than 10 foot gradient?
crossing that buildable area. So, you know, obviously there’s just some nuance to even consider within that standpoint. How do you determine a buildable area? You know, do you look at the subject property and then if you have a wide enough scope on your aerial image, can you see other neighboring properties and make reasonable determinations where a buildable area would be based on where other
homes or structural improvements are located, that is certainly one clue. But if you don’t have other homes to look at, what else do you need to determine? It’s probably not going to be an area immediately against the street, probably at least a third of the way onto the property to account for setbacks available in most parts of the country. If it’s a much larger rural tract, there could be multiple different building sites to consider. So if the quote unquote primary buildable area is impacted by
larger contour, you have an alternative that might be a little bit more difficult to access. But, you know, could still serve as a cheaper build option when you account for contour. And so you have to have some of these weighted statements that run through different decision making criteria. And, you know, if there’s a clearing, OK, that’s probably more likely to be buildable area. But, you know, forested properties, which is
pretty common in a lot of vacant land. can’t rule out dense foliage from occluding a, or, you know, uh, precluding the ability to consider an area to be a primary buildable area either, because somebody can just go ahead and clear out those trees as well. Um, so, you know, a number of different considerations. I’m probably even forgetting a few, but those are the bigger ones that come to mind here.
And so, you we need to teach the AI how do you determine? Okay, what is that buildable area and then assess, you know the actual contour which needs to take into account the different contour lines and also look at the numbers on the lines because some areas of the country it might only be a five foot increment some will be ten foot some will be 20 some will even be like 40 feet so you need a that the You know depending on what numbers are associated with those lines
the AI needs to be able to determine, okay, is this truly a 10 foot or greater contour change within this buildable area based on these numbers? Or if there’s no numbers at all there, then again, it could make a determination based on neighboring properties or just have a default. Okay, you know, if there’s no numbers here, let’s make an assumption that it’s at least a 10 foot. Increment be more conservative. I’d rather have a false.
false negative than a, a false positive in determining, you know, whether something has, or to be clear, I would rather be more conservative and consider a property to have a steeper contour than it actually has versus thinking that a property is flatter than not, because you think about the risk, right?
The underlying risk of getting caught with a property that has a really aggressive slope is much worse than something that is just flatter. So you always want to wait to your decision making towards how do I avoid the biggest risk for the property if you know you’re not going to be 100 % right all the time. But yeah, we’re really close. We’re like 95 % plus accurate here. So that’s one avenue that we try to determine. And then…
The next determination is, is the buildable area of the subject property flatter or equivalent to neighboring properties that have improvements on them? Or is it a steeper slope? And so this does require some improvements to be
nearby. If there’s not, then we have to label it, hey, there’s no structural improvements. We can’t quite tell. So, you know, this requires the AI to look at all these neighboring properties, determine, okay, is there structural improvement on it in the first place? What’s the contour in that particular area? And then run that entire assessment from the previous prompt for understanding, okay, is there a
you know, the contour within that anticipated buildable area, whether it’s primary or secondary, is again equal to or less in terms of aggressiveness of slope, or is it worse? And obviously we want to be conservative so that if the slope of the subject property in the buildable area is worse than any neighboring
structural improvements, then that gives us a clue that, this may not be buildable. We really need to assess this further. And it’s going to be more of a black mark on the property. But if there’s already a precedent of a neighboring property having a structural improvement built, even if it appears to be a bit more of a structural or aggressive slope, that’s okay because somebody has already figured that out within the area. It’s proving that it is buildable, even if it might be a bit more expensive.
Um, so, you know, as you can tell, you know, this might even be flying over your head a little bit on how we’re making these determinations. know, sometimes like, yeah, this, if you’re just looking at this and you’re more experienced land investor, like you can try to tell this a little bit more quick. I would say a lot of people are not routinely doing this when it comes to assessing contour. Um, but when you’re actually trying to prompt AI to do this instead and, know,
and get it to be really accurate. you don’t need to do this and you can just have a more reliable assessment on a particular property. Like you can see that the prompt gets really, really complex quickly. And I’m not like even reading off all the if then statements and how we design the prompt on the backend there. I’m just like kind of walking you through the logic on how you would.
consider this even from a human perspective. And it gets, you know, again, pretty crazy to consider that. And then all the possible false positives and negatives, like what do you determine to be like a structural improvement that allows for people to live in the property versus, you know, maybe sheds or, you know, non-residential improvements that shouldn’t be considered as,
something to prove out the viability of the contour within the buildable area. So you need to determine that as well. And you also have to account for subject properties that might be significant acreage. And so your resolution might be a bit different. It might be harder to suss out where neighboring structural improvements are in relation to smaller properties. So all of these different factors need to be accounted for.
before making a determination. So just wanted to share that as again, the level of nuance that we go into because understanding contour and what might be buildable or not is just one of the most critical factors of land. Plus there’s also the factor that, you know, for some of the user oriented questions we have to ask, okay, is this property going to be
you know, primarily recreational, in, in, in usage, or, know, for some type of agricultural usage, you know, is it like for timber or is it just for hunting? even hiking, because if that’s the case, if that’s the primary use of the property, then contour really doesn’t matter. Right. So we have to be able to account within our system. All right. If, if it’s not a anticipated.
primary residential use of a property, then the contour issue is, you know, effectively negligible or, you know, much less of a concern. So we have to all weigh these different factors together when we’re considering the underlying value of the property. So, you know, I realize this might have been a little bit complex, but, you know, I’m really trying to walk you through our logic.
you know, both on how we assess properties just by ourselves without software, but also how we’re building this into software. Um, so that you don’t actually have to do this anymore. And it doesn’t matter on what type of energy you have or what team member you have and so forth. Like it will be standard, um, anywhere you operate within the country with, you know, rigorous testing to determine, okay, yeah, this is that this AI image analysis just has it down pat.
and you can trust its assessment in that regard. And that is where we have it at the moment. SeriousLand.Capital for any of your funding needs. Dealer Diligence Facebook Group for zero cost review of your land deals and LandPrizer.ai, which I just talked about some of our feature sets here today. Subscribe and share. Happy Fourth if you’re listening to this when the podcast comes out. Looking forward to talking to you next time. Take care, everybody. Bye.
