227.5 Bonus Episode: AI in Homebuilding: A Practical Approach - Bill Gelbaugh and Al Trellis Jul 29 2024

Kevin Weitzel: [00:00:00] All right. Well, hey, everybody. Thank you for joining us today on OutHouse's webinar. We've got a good one for you today. Um, I get to be just, I'm not actually bringing any content. I'm just, I get to be the pretty face today on this webinar, uh, with these two gentlemen, and I'll introduce them right now. We have Bill Gelbaugh, a founding partner here at OutHouse who is with us.

We also have Al Trellis. He is a principal with the Home Builders Network as well as a founder and a consultancy company called [00:01:00] Values That Matter. Amongst other things, uh, one, he's a multiple published author. Uh, he is a habitual speaker at IBS for the NAHB as well as other speaking engagements around the country.

Bill, can we just get a quick hello from you?

Bill Gelbaugh: Hello.

Kevin Weitzel: That was very brief.

Bill Gelbaugh: You said quick.

Kevin Weitzel: I did say quick. Uh, Al, how about yourself, sir?

Al Trellis: Hey, I'm ready to rumble. Let's go.

Kevin Weitzel: Alright, well, there's no secret of what we're talking about today. We're talking about AI. Uh, and rather than giving a detailed description, because, hey, I'm a mouth breathing knuckle dragger myself, uh, we're going to go into just starting the program with a fun example that Al has. Al, let's go ahead and check out that fun little example you have. What AI can and cannot do.

Al Trellis: Right. So, if you put that slide up, there you go. AI can do this. Give me four great bass lakes. And by the way, I fished three of those four. And, but it can't do this. It can't go out and catch the fish. [00:02:00] And I think it's always important to remember that. Now, someday we might have a robot that can do that. But for the moment, what it's going to do is talk to you, give you ideas, give you feedback. help you stimulate your thought, but in the end it can't really do a lot. It can only help you do a lot. That's really the key.

Kevin Weitzel: So Bill, uh, let me get a question out to you first. There's a lot of recent buzz about large language models. Can you talk to us a little bit about them and why they are important to the framework of AI?

Bill Gelbaugh: Yeah, large language models are really the beginning. We're at the beginning stage of AI, and it really encompasses a couple different things, and I like to break it down. You know, you have all the servers in the network, and then you have the data, which has been being put together for years and years, so that AI has the content that it needs to be [00:03:00] able to answer the questions.

But the real breakthrough came, you know, this past couple of years, especially with OpenAI where they have the natural language processing and the pre-processing of all of that data and information so that you can have natural language and interaction, you know, with humans.

And so what it does is it understands context. It understands emotion. It understands grammar. And at that point, you know, both verbally and through, you know, typing and text, you can communicate back and forth with a seems like has a very human interaction. It has a way to go, and I'm hoping at the end of the conversation, I can tell you a little bit more about the direction that [00:04:00] it's gone. But that's really where we're at right now.

Kevin Weitzel: All right. So, Al, there are plenty of GPTSs out there. What are some more interesting ones that you want to rattle off for us?

Al Trellis: Well, so DALL E used to be separate, but now if you buy a paid subscription to ChatGPT, which everybody should do, it's 20 a month, the DALL E image generator is part of that. And I'll show you some images that it made. There are, so what happens is there are now like, like the app store, you can, you can get different baby GPTs that are sort of very specific uses, tools built around the big GPT around ChatGPT, and these are some of the ones that are available. You can find them and I can show you later how to search and find them, but you can't get them easily without a paid subscription, but you have a free subscription, which is ChatGPT 3.5 [00:05:00] you're going to just get the regular things.

And the real key to understanding large language models is the power of iteration, because you can ask it a question, get an answer, and then either ask it for more information or a clarification of that information over and over and over again and keep zeroing in on what you want, and there's no, there's no price. There's no charge. If you draft something, send it to an editor, they send it back. Then you tell them, I want you to change this. You got to send it back. The machine will just keep doing that as many times as you want for no charge. That's the real, iteration is the real power of AI, in my opinion.

Bill Gelbaugh: Yeah, I agree.

Kevin Weitzel: So, there's practical applications for LLMs. Many of us can use in our day to day business. Can you share a few

Al Trellis: For me personally?

Kevin Weitzel: Yeah, Al.

Al Trellis: So, I one of the things I use it for a lot is to stimulate my thinking on a particular subject. So, I also use it [00:06:00] like I don't read business books anymore. My wife, my wife says I should never tell people that makes me sound ignorant, but I don't read business books because I just go to Chat and tell it, give me a 300, 500, 1,000-word summary, give me the 20 key teaching points, and give me 10 ways I could use those in my home building business of Sioux Falls, North Dakota, where I build approximately 12 houses a year at an average price of 350, 000.

And you'll notice that my prompt didn't say, tell me about the book. It said, the more you can be specific in the prompt, the more powerful the answer is that you'll get.

Kevin Weitzel: So, wait a minute, you're meaning you're gonna stand there and tell me that I can put CliffNotes out of business?

Al Trellis: You, CliffNotes is out of business whether they know it or not. I don't have any idea. .

Kevin Weitzel: Uh, all right. So, here we go. So, you can also use it for like job postings as well.

Al Trellis: This is a company that I work with called Contractor Staffing Source. I refer them to my clients. They have a whole bunch [00:07:00] of AI methodologies put together to write the job description, post the job description in 37 different places, get the responses, screen the responses, filter them, rate them, and then give them to my clients with questions to ask at the interview. They can do it for a discounted price compared to a regular placement service because there's very little human involvement. So, it's an example of how AI makes things more efficient, which eventually will lower the price of a lot of different things.

Kevin Weitzel: So, all home builders could use it to help them with their logo design. They can help, they can use it for languages. They can use it for job postings. What else, what else do they have available at their fingertips?

Al Trellis: Well, we can use it to help organize your company. You can organize your files with it. You can use it for marketing. I mean, there's four ways to make more money in a home building business. In fact, there's really only four ways to make more money in any business. And you guys should, since you guys have a business, you should understand this. What are the [00:08:00] four ways? You raise the price, you lower the cost, that's the direct cost. You reduce your overhead or you sell more. And by the way, that's the order in which I believe you should think of them. Most people, most builders, which of those do they think of first, Kevin?

Kevin Weitzel: They're always talking about if they can raise their prices because

Al Trellis: They're always scared to death to raise their prices. They're always talking about what? Everybody wants to, but everybody's afraid to. Everybody talks about which one first.

Kevin Weitzel: They always want to lower the cost.

Al Trellis: No, they don't. That's not how you're going to grow your business. I'm going to sell more stuff. Think about it. If you're Heinz ketchup, I'm going to sell relish too. If you're selling SUVs, I'm going to build an electric car. They all want to like increase their lines. AI can help you with all of these and we have a couple of examples. We'll just pick a couple of them. look at raise.

Kevin Weitzel: Not to side skirt too [00:09:00] much, but just so you listeners know, one of my thorough joys in life is to derail either Bill or Al in any of our meetings. So, uh, so just kind of come along for the ride on that one.

Al Trellis: So, but you, here's the fallacy that you believe we're actually derailed. We're just playing along,so you have a job. So, so when all of these things AI can help you with in raising the price, we can use optimized pricing or dynamic pricing. Now, dynamic pricing has gotten a bad name in a lot of people like it, but the reality is they see it every day in their life.

I'm going to work in Alabama next month with one of my clients and the Auburn football game is in town and the same hotel that I usually pay 165 for is 355. And Al, if you don't like it, don't come to Auburn that week. That's an example of dynamic [00:10:00] pricing. Well, we can't do it that much, but let me give you an example.

I have a subdivision with 40 lots. Every lot has a unique price. It's got a lot premium. Some are zero, some are fifteen thousand dollars. I start to sell them. The machine tracks what's selling and comes back and recommends to me you should raise the price of these kind of lots because that's obviously what most people want. That's a form of dynamic pricing. It's we change the price based on feedback.

Variable margin pricing. Every good builder understands this. A good example would be options and upgrades. Any builder who's listening, can you charge the same margin on a washing machine that you can charge on a four foot extension out the back of the house? Of course not, because the whole world knows what a washing machine costs. It's all over the place. You go online, but nobody can tell you how much a four foot [00:11:00] extension at the back of your house costs, except you.

So, on the washing machine, we typically make 15%, and on the four foot extension, we make 50%. That's variable margin pricing. And many builders make a huge mistake. They price all their plans with the same margin. Every plan should have its own unique margin, depending on the perceived value of that house versus what it costs to build it. Price elasticity, bundling, competitive studies, AI can help you do all of these things.

And in fact, one of my builders, we use it heavily for competitive analysis. And we have an AI company that I've invested in that I'm involved in. We do competitive analysis all the time. And we use the multiple lists. We get the listings. We compare the sales. We compare old houses to new houses. We got, there's no problem with getting data.

The question is, how do you analyze it? So that's, those are just some examples of how you can get more money from the same [00:12:00] number of houses. with the assistance of AI. And you can do some of those things without AI, but AI makes it easier. These are some ways you can sell more with AI. You get better, better customer analytics.

Um, by the way, does anybody know what lead scoring is? So, lead scoring is where the AI helps you score the quality of the lead based on certain things that we put into the system. It's fantastic for training. One of the things I love about AI, whether it's ChatGPT or Claude, which are the two I use the most, is it's fabulous to write training materials, and even the tests that go with those materials.

It can teach you how to overcome objections. We can ask it, like, here's an objection, give me some ideas about how to overcome it. Now write me some scripts to overcome it. Now take the best part of script number one and this part of script number four, put it together in a new script. It's fabulous to help you with frequently asked questions and answers to those questions.

This [00:13:00] is the one that frustrates a lot of people. Automated customer engagement at its most primitive form is the idiot thing you have to deal with when you call the cable company and you got to go through 400 questions before they'll let you talk to somebody. But when it's done right, it can actually be a positive experience. Most people, when they have automatic automated customer engagement, it's a negative experience, so.

Kevin Weitzel: Well, well, Al, what are some real life examples of using AI to benefit the client?

Al Trellis: Some real life examples to benefit the client? Well, first of all, anything that the client being the home buyer or these, these, a lot of these examples are for my builders, but benefit the client, the number one benefit to clients everywhere of AI is eventually lowers costs. It lowers, it makes productivity more efficient, lowers your cost, some of which, you notice I didn't say all of which, some of which we can pass on to that [00:14:00] customer in the form of lower prices. Some of which we keep for ourselves because we're now a more efficient provider and we're entitled to make more money.

The market rewards efficient producers. It rewards them two ways. They make more money and they usually have a more competitive price. Does that make any sense?

Kevin Weitzel: It does. Now, from a personal example, now Bill, I know that you're a ridiculous reader, like you've read, you've read, uh, you could read me under the table, uh, with just a, a month's worth of reading that you do. How do you use AI on a personal level to find the next read, it you will?

Bill Gelbaugh: Well, you know, the thing that's nice about what Al brought it up a little bit earlier with perplexity. I mean, one of the things that everybody should check out is the combination of using Perplexity and using ChatGPT, Claude, you know, the other direct GPTs [00:15:00] is Perplexity is great for doing the research, and then the ChatGPT and Claude is great for doing the book summaries and everything that you want.

So, what I do, I mean, I'll give you a couple of examples. I'm really, I, I usually follow themes. Like right now I'm studying stoicism because I like to go through the different philosophies. So, I'll do as you were talking about, you know, I'll go out and I'll use Perplexity to do the research based on what I'm interested in.

So, that's the nice thing. You can go out and do a fairly zeroed in or detailed search around the fact that this is what I'm interested in and then it will go out and it will suggest the books, it will suggest YouTube videos [00:16:00] that you could watch. It pulls all different kinds of citations from different areas, and then gives you a complete summary of them so that you can go out and check them individually.

Um, but I've used it for, you know, doing, um. research on stoicism, research on OKRs, if your business is, you know, looking to go out and, and develop OKRs. And then, we use it quite a bit here for, um, marketing and, you know, different marketing research that, that we want to do. So, it's, uh, it's, it's really flexible using it. I use it just as much for personal interest. stuff as I do for, you know, aiding the business and its research and development.

Kevin Weitzel: All right. Before we change gears, Al, can I find in this [00:17:00] book called Wisdom, uh, by a famous guy? Oh, Al Trellis. Hey, can I find some stoicism in this guy?

Al Trellis: There's a little stoicism in there. You know, you know what the Spartans said, come home with your shielder on it.

Kevin Weitzel: That's it.

Al Trellis: That's as stoic as it's going to get, baby.

Kevin Weitzel: All right, so, Al, let me come back to you. So, there's a large number of, uh, keys to successful, or to being successful using large language models. Can you name a few of those? Or can you talk us through a bunch, a couple of those?

Al Trellis: This is especially true if you have the paid version. You can do it without the paid version, but whenever you put in a prompt, you always want to tell it who you are, and tell it who you want it to be and then tell it what you want. So, let me give you an example.

A good prompt would say something like this. Well, in fact, let me, at the end, I'll show you a great prompt that we use for editing, which was written by the perplex, but was written by the, um, Claude people. [00:18:00] And it's really, really strong. So, so tell it what you want clearly and descriptively. It loves adjectives.

Like I want, uh, you, you, here you are an expert in punctuation, grammar, vocabulary. You tell it that, so it knows you're interested in those things. But you tell it, I am a builder who builds high priced houses in the southwest. So, now it knows what your focus is about. In the paid version, you can actually put that in as a default.

So, it always knows when it's talking to you. I know that you're a consultant. You got this many clients across North America. You've got 52 years of experience. It knows all these things about me. So, it, it often will say this will be of particular interest to you, Al, because of your knowledge of so and so.

Don't expect a perfect answer the first time. The better the prompt, the closer to a perfect answer you'll get. And then iterate, keep zeroing in to make your answer better and better [00:19:00] and better. And then learn to dig deeper and deeper. Like a lot of times it'll give me like when I did HR system for homebuilders.

It gave me the 10 areas that was going to talk about compensation, retention, etc. I said, I'm very interested in training. Tell me about this one. Give me 10 ways I could train my people better. And then I looked at and then like number five, about safety. Now give me five examples of what I should teach them. Keep digging deeper and deeper and deeper.

And finally, give it examples and ask for examples. I love to do that with it. Like tell me the, this thing, and then give me three examples of how I could use that in my business. If I don't like the three examples, guess what I asked for, Kevin?

Kevin Weitzel: Another example.

Al Trellis: No, I asked for three more.

Cause it doesn't care. It doesn't care if you ask for one or five or 10, it doesn't care.

Kevin Weitzel: Now I get to [00:20:00] being, I get to work in the same office as Bill every once in a while, I get to see the joy in his face, like a kid in a candy store when he comes out to show us this new fangled thing that he's learned about ChatGPT, or some sort of AI. Uh, but Bill, could you give us an example of, or maybe some pointers of stuff that you've learned?

Bill Gelbaugh: Yeah. I mean, I think, you know, one of the important things for this group and what we're talking about and Al really touched on it is for me. I found that there's three layers to really getting what you want out of ChatGPT or any of these is going in as Al was suggesting and taking the time at the highest level to get the paid version. And at that point, you can go in and you under your profile, you can put in a custom profile and you can tell Chat PT, basically exactly who you are, what [00:21:00] persona you are, how you like to be talked to, at what level, and then give it all of the detailed instructions of how you want it to respond.

Like in my case, you know, I want it to respond to somebody who's well read, reasonably intelligent, and likes to have conversations over a beer. So, you know, that's my general prompt for ChatGPT. Then I tell it, you know, the specifics that I want good grammar, and I want it humanized, and a couple other suggestions that we could talk about.

But at that highest level, getting ChatGPT or Claude or any of these that custom profile in there, you will start getting better responses right away. It will start talking to you the way that you want it to respond the way that you want it to talk to you. And then I believe [00:22:00] Al's actually working on one of his own.

But the second level then is to go in and create customized GPTs that will allow you to do specific tasks or responsibilities. And like one of the examples that we used here at the company is, I went and did a writing profile of Kevin's writing. And then I went in and did a writing profile of Jim's writing. And then I went in and did a profile of my writing. And, you know, basically that way, when we want to send out emails or sales or marketing, we can choose any one of those writing styles.

And so, I created a custom GPT for each of those writing styles. So, then when we go and use Perplexity to do the research on a particular market, or [00:23:00] even an individual builder, we can match that with a template done in Kevin's writing style or Jim's writing style and at that point, you know, those day to day tasks become really efficient and they really have an efficacy to them that you never had before. Because you can go, you know, Kevin's writing styles too hard on the nose. We got to go softer. We're going to use Jim's writing style and you can use the same research and combine it into a template and go out.

Al Trellis: But it sounds like it came from two different people.

Bill Gelbaugh: Yeah, it does. It does. And that's one of the most important things, you know, that we learned is it can take on any persona that you want.

 Just for fun. If you want to actually entertain yourself, uh, have Claude, uh, [00:24:00] give you a synopsis or a summary of Genesis from the Bible from the perspective of a circus seventies pimp speaking American jive. It will blow your mind what it comes back with.

Kevin Weitzel: It's ridiculous.

Al Trellis: I'm not going to do that particular one. I'd rather talk about stoicism, but

Kevin Weitzel: Yep. All right. So Al let's, uh, let's go with this. I know that you've talked about, uh, and mentioned the book, the, the, uh, The Score That Takes Care of Itself. Um,

Al Trellis: got to hear the story that goes with it.

Kevin Weitzel: Yeah. Yeah. Go with that story that goes with that and how you needed it and, or how you found it, et cetera.

Al Trellis: I'm helping a builder in Michigan work on a project that belongs to an insurance company, the land does. And so, I'm talking to the president of the insurance company and we're talking about business and what business books do you like to read?

And he says, this book, The Score Takes Care of Itself. What do you think? I said, I'd never read it. I said, but I'll go, I'll send you an email later today about it. So, then after we're done with the conversation, I go to, [00:25:00] I go to, uh, Claude and I say, give me a 500 or Chat, give me a 500 word summary and 20 key points. It does. Perfectly.

I say, give me six ways I can improve myself on point number 10. It blows it and gives me the answer for number 20. And I go, no, you're wrong. I want number 10 commitment to excellence. I tell it the name of it. And now it gives me six ways I can improve myself there. And then I, of the six ways I say, I'm really interested in number three, prioritize continuous learning.

Give me four things I can do to excel at that. And it says set learning goals, leverage diverse learning resources, implement learning schedule, apply what you learn. And it gives me like 60 words for each of these. Okay, then I say on this learning schedule, make me a learning schedule and it does give me a really detailed learning schedule.

But this is the part that kind of blew my mind [00:26:00] unsolicited. It said the following tip tips for success of how to adhere to the learning schedule. I'm thinking this is like fantastic. So, I send this to the guy and I go when you read the book, Did he give you any tips to success on that learning schedule part about how to improve yourself? Because I didn't read the book. I just got the 15 minute summary, but I got this out of it. And he's like, that's fantastic. I'm not going to read any more business books. So, I'm, I'm slowly undermining the publishing industry, I think.

Kevin Weitzel: You're shrinking the research time too, considerably.

Al Trellis: Somebody's got to write the book for Chat to summarize it for you. But you know, like that little book, you held up the Wisdom book. It's a short book. It's only 56 pages, but I asked it like in the book, I have quotes by famous people and what I think of them. And I tested Chat, find me some quotes like the ones I use on the same subject. And then I gave it my writing [00:27:00] styles and I told it write about them the way I would write about them? And I showed them to my wife and there's a little scary of the six. I showed her five of them. She couldn't tell that it wasn't me.

Bill Gelbaugh: Yeah.

Al Trellis: And if I work at it harder, she won't be able to tell for any of them.

Bill Gelbaugh: Well, and, and, you know, to give an example from my own AL, you know, I've used it for, I'm reading Marcus Aurelius's Meditations and, you know, you want a modern translation of those. And so, what I did is I created a persona and I created a persona for Marcus Aurelius. I took a bunch of his writings and, you know, used it to create a GPT that would respond as if it was Marcus Aurelius. Then I gave translations of Marcus Aurelius's meditations and [00:28:00] had it provided in a modern language and vernacular as if the Marcus Aurelius was living today and using our language.

And I got some of the best translations of Marcus Aurelius's meditations that just absolutely blew me away, especially passages that were written by him, translated from the Greek, you know, a hundred years ago by old fellow or something like that. And you're gone, boy, I'm really having a hard time understanding this.

When I would go through and I did about probably 30 of my favorite, um, the translations were incredible because I was having them tailored specifically for the way that I wanted them done. And then it would come back like you did here. Um, and, and, and add additional suggestions, how to ask [00:29:00] for better translation.

Al Trellis: At the end, if we, when we move through at the end, I'm going to show everybody this ChatGPT that we're building one of the early versions. Cause I don't want to show you guys the latest version, but one of the early versions and I trained it. It's trained by my writings. So, a lot of the answers it gives are very similar to the answers I would give if you asked me the same question.

Kevin Weitzel: So Bill, that brings us to a pretty important part. You have to put in the right prompts, like what you would put in would be different than what I would put in, so can you give us some examples of good and bad prompts?

Bill Gelbaugh: Well, you know, that's really where I've spent the most time with, is the prompt engineering area.

And just some brief suggestions for, I mean, I can tell by some of the comments in the chat that are coming up that people are, you know, have actually worked with this and, and, and are offering suggestions to each other. But [00:30:00] as Al was providing, you know, earlier, you know, the most important thing is to provide what persona or what role you want to ChatGPT to take or Claude or any of the others, what role or persona that you want it to take and then tell it what the context is and who the audience is.

And provide an example of the writing if you possibly can, and then tell it the format and the tone that you want. Um, there are GPTs out there that if you have a particular style or whatever, as I'm explaining this, that, that you like, it will go and do an analysis and give you all of this back for yourself. So. As, as Al touched on earlier, if [00:31:00] you provide the persona, the context and example, tell it what format exactly how you want it to respond. And you keep, I keep a prompt library that keeps all of that ready for me just to fill in.

So, the first thing is to create yourself a template for good prompts. So, that you go in and fill in the right information consistently and then using that prompt template every time you can keep tweaking it, you know, till the template gets better, but then your, your prompts right from the get go will be more consistent.

But what I do, a trick that I learned is to take and fill all that information in. And then ask ChatGPT or Claude or Perplexity to write you the perfect [00:32:00] prompt using this information. Never struggle to, you know, sit there and word edit your own prompt. Just get all the details down in a template. Provide that.

And then ask for the AI to actually write its own prompt, and you will get a really, really great response from that. And then, as Al was explaining earlier, the key from the human side then, once you've done that, to write a good prompt. You just don't come in and ask it like you have up here in the bad prompt, give me a summary.

You gotta tell exactly, you know, what persona and how you want it to respond and write that summary out and how many words and what the tone is and all of that stuff. But if you do what I was saying, and have it right, the prompt, [00:33:00] um, you will start getting some of the best responses that you've ever seen.

I mean, it, it will go up 1000 percent over just what you started playing around with and writing the right question. But at the same time, you have to go through, as Al was explaining earlier, a lot of iterations. And if you don't get the exact answer that you want from that prompt, then you keep tweaking it and keep asking it differently and keep tweaking it and saying, no, that's too flowery of language.

I want more concise language, or I want you to expand and provide a more in depth and philosophical analysis and keep notes while you're doing that. And then that way you can go back and keep refining. Your template for your prompts, so that eventually, you [00:34:00] know, they're pretty dead on from the very get go.

Al Trellis: So, I'll show you a couple of big time prompts when we get to the end of the program.

Kevin Weitzel: So, when it comes to the subject, and we're talking about generative AI, Where are people going to go to learn about this, Al? Where, where, where can people head to?

Al Trellis: Well, I took the course on the, I took the course on the right from IBM. I actually didn't take it. I'm 53 percent of the way through it. I haven't, like, I'm too busy to go back and finish it. Because I feel like my prompting skills have gotten to the level where I'm not even sure I need to finish the class. But, these are two examples. One from Vanderbilt. and one from IBM. They're both free through Coursera, which is a great.

Bill Gelbaugh: Yeah, Coursera is really good.

Al Trellis: I've taken four courses, four significant courses from Coursera, one on negotiation, one on model thinking, one on gamification, and one on marketing. And I took them from Yale, University of Pennsylvania Business School, [00:35:00] University of Michigan. I mean, there's some great stuff available for free. It's crazy, but you got to be willing to put the time in. Those classes take like 20, 30 hours of your time to really do them well. So, these are some great sources right here. I also had some articles in the last, if you want the quick version and that last slide, there were two articles from fortune magazine that are really good.

Kevin Weitzel: Bill, what about you? Do you have any websites or books that you'd recommend?

Bill Gelbaugh: I don't have any, you know, specific. I've done what Al's done here. You, if you just want to get that foundation, go to Coursera, find something like the IBM, go through and, you know, they'll, they'll lay out basics, the fundamentals.

But really, I've found the most, as I was explaining earlier with, you know, how to do this well, where you're going to go in and put that custom profile in and then do the standard day to day stuff with the GPTs and then [00:36:00] get really good at prompt engineering. Um, YouTube. I mean, you go to YouTube once you have that foundation and there are people that are really, really good that will walk you through whatever you specifically want to, you know, to be learning about customizing it at the highest level or engineering a prompt.

Kevin Weitzel: Uh, Aaron asked if the slides will be available after the show, and that would be, yes. We'll be sending that out with the video to all attendees and anybody that was registered, just so you know. Hey Bill, let's come back to you. Homebuilders are always looking for AI tools that they can use. Can you give us some examples of, uh, some builder specific homebuilding tools? Oh, we even have a slide.

Bill Gelbaugh: Oh, yeah, yeah. Well, we have, you know, some partners that we've run into over the years with, uh, AtlasRTX and their chatbot [00:37:00] and OpenAI and they're doing the prediction and analysis, um, open house sales projecting and, and all of that. So open, uh, OpenHouse is a great one to check out. And then, uh, Automation Agency is the one that I know the least about. Maybe you could tell them a little bit more about Automation Agency.

Kevin Weitzel: Yeah, automation Agency is a company that has linked AI to CRM integration. Making the ability to reach out to your clients, make that process more efficient. That's really their, their forte. And I've heard nothing but glowing reports from the companies that use them. At least the ones that I've, that I've chatted with that have spoken with them directly. But yeah, that's Automation Agency. So again, for CRM integration, making that more efficient. A lot more efficient.

Bill Gelbaugh: OpenHouse and Atlas are both great.

Kevin Weitzel: Oh yeah.

Bill Gelbaugh: I've really loved our interaction with them.

Kevin Weitzel: All right, so Al, can we do, do me a favor and let's review just the top [00:38:00] AI applications for home builders.

Al Trellis: So, it's fabulous for writing copy. In fact, I think if you're a copywriter, you're out of business. You got to become an editor. Create training materials. That's really good for helping file documents. Uh, good, quick, deep research on almost anything, increasing your operational efficiency on any task that's redundant, uh, repetitious, obtain and analyze market data easily and inexpensively.

We're, we're partners in a company called Real Torch where we get unbelievable data very inexpensively. They're helping me develop the, uh, the app that you mentioned earlier, which I'm going to show you. It's called the Home Builder Advisor. We use it to stimulate creative thought whenever I'm sort of stuck creatively. I'll just ask it some generic questions about what I'm thinking about to help stimulate my thinking. It's really good in any kind of assisting personalized customer experience. You can do a lot to automate the HR function and um, it really helps you optimize your [00:39:00] pricing through better data and better analysis of that data.

Kevin Weitzel: All right. So, Al, uh, let's go ahead and wrap up a little bit. Uh, can you give me a summary of where you are and where we are as an industry with the AI transformation journey?

Al Trellis: So, number one, somebody argued with me the other day. He said, Al, we're in the, we're the second batter up in the first inning, but whichever way you want to look at it, we're early in the game. And just the changes since I started doing this about a year ago till today are unbelievable. I've got 550 hours, I logd my time with AIs, talking to AIs. My wife is like really annoyed with me. So, um, but it's amazing. The difference between when I started and today.

And remember, I really believe this. AI, this is not an incremental thing. This is a transformational thing. The world is really going to change. A lot of the people common phrase is, this is as big as the internet. I think that's a mistake. I think it's bigger than the internet. And one of the reasons it's bigger than the [00:40:00] Internet is because you have the Internet. When you created the Internet, to get everybody to see it, use it, that took a long time. Now you create something on ChatGPT, everybody's got it the next day. So, the cycle time is much quicker.

And since I've started, it went from Chat 3. 5 to 4, to 4, to 4 Omnivore. So, there's two more versions in a year. The small to mid sized builder is under a lot of pressure from bigger builders. If you want to be irrelevant, just keep doing what you're doing, and they'll, they'll take care of it for you. They'll make you irrelevant really fast. So, this is actually a tool that helps the little guy as much as it helps the big guy. But if you don't use it, then, you know, what do you do? There's a famous quote by Mark Twain, There is no difference between a man who can't read and a man who chooses not to read, right?

The tool's here. [00:41:00] Pay 20 bucks a month and learn it. At a minimum, figure out the large language models and just become familiar with them. See what it can do for you. Try different things. And aggressively explore current AI tools and constantly look for new ones. Like I find new ones all the time. That's how I found Perplexity. Someone mentioned it to me, I went there, all of a sudden I love it. So, that's my vision of the future.

Kevin Weitzel: You know, I, Al, I agree with you. I would actually argue that a smaller builder can get more benefit than large nationals because they have more flexibility. They don't have to follow all the corporate rules and regulations they have.

Al Trellis: You talk about flexibility. There's a great, the most famous quote is knowledge is power, right? I think that's an incomplete quote. I tell everybody to listen. That's not a good quote. This is the way the quote should be written. Knowledge is power in the hands of those who know how to and are willing to use it. Otherwise, [00:42:00] it's just information.

Kevin Weitzel: I agree. Bill. Any final thoughts, sir, before we get to the example and question?

Bill Gelbaugh: Yeah, I'm gonna follow up. I'm going to follow up in the same vein that Al was talking about, and I'm going to read this because it just came out, so I haven't memorized it yet, but if you go over and you use Perplexity, one of the things that they do is they publish news on a, you know, a couple times a week about some of the biggest events.

And so just recently they published, uh, OpenAI's five step plan for achieving AGI, which is their term, artificial general intelligence. So, I want to read this for everybody because it'll give you a roadmap to what Al was explaining. Said the first level of OpenAI's AGI roadmap focuses on conversational AI, which represents the current [00:43:00] state of the company's technology.

Where it's moving right now is the second level of OpenAI's AGI roadmap and introduces Reasoners, AI systems capable of solving complex problems with the proficiency Of human experts. And the third level in the road map is what they're calling agents, and they emerge as AI systems capable of operating autonomously for extended periods of time. These advanced models can spend several days acting on a user's behalf, taking on complex tasks, making decisions, and adapting to changing circumstances without constant human oversight.

And the 4th level of OpenAI's [00:44:00] AGI roadmap introduces innovators. AI systems capable of developing groundbreaking ideas and solutions across various fields. And the fifth one is the one that blew me away. The pinnacle of the roadmap, level five, envisions AI organizations that can function as entire business entities processing strategic thinking operational efficiency and adaptability to manage complex systems and achieve organizational goals.

So, um, as Al was saying, as fast as it's gone from 3.0 to 3.5 to 4. 0, you know, 4.0 and 4.0 Omnivore, um, and they're already [00:45:00] starting to slip into stage 2. The foundation is set, like we were talking about in the beginning of the presentation, for them to get these large language models built, the infrastructure in system servers data, you know, get that pre processing done, so they can talk natural language. You know, that took years and years and years. But now that that foundation set, um, I really think this is this is gonna go. This is gonna go fast.

Kevin Weitzel: We're not even talking, we're not even talking prop plane versus jet plane. We're talking about, you know, all the way into interstellar.

Bill Gelbaugh: Yeah, we're gonna go. We're gonna go from driving cars to interstellar, you know, in the next next few years. If you're in leadership or management, this is going to be so disruptive. It's so exciting for the people that can take it under their arms and [00:46:00] learn it well and build their organization, uh, utilizing it. But if you can't, you're going to be left in the dirt and it's going to move fast.

Kevin Weitzel: All right, so Al, did you have that example you can show?

Al Trellis: Yeah, so I'm gonna do two things. First, I'm gonna put, let me share the screen here. This is a prompt, one of my favorite prompts. I'm just, let me bold this so everybody can read it easily. This is a prompt. I want ChatGPT to be an editor. I, this is actually perfect for Claude. It was written for Claude. You are an AI copywriter. With a keen eye for detail and a deep understanding of language, style, and grammar. Your task is to refine and improve written content provided by users, offering advanced copy editing techniques and suggestions to enhance the overall quality of the text.

When a user submits a piece of writing, you will follow these steps. Read through, you see how detailed this is? [00:47:00] And when you're done at the end, you'll give me a fully edited version that takes into account all of your suggestions. Your suggestions should be constructive, insightful, and designed to help the user elevate the quality of their writing.

You like this one, Bill? It's a nice one, isn't it?

Bill Gelbaugh: I absolutely, yeah, that, and that's exactly what we've been talking about. It's a great example to end the discussion because you know that's been tweaked and tweaked and tweaked until you you know you get as close to the result but you have to tell every specific thing that that you want and you did a great job on that prompt

Al Trellis: Well, I stole it. I didn't write it. So, here's here's one here's one that I want you to see this is my ChatGPT account and you can tell it's a paid account because I have my name up here. if I click on my name and I go to, I go to, uh, is it settings or customized?

Bill Gelbaugh: Yeah.

Al Trellis: Right [00:48:00] here. I am a home building consultant. It tells them who, it tells them who I am and I'm going to go back actually after this conversation tonight, I'm going to add another hundred words in here. I'm going to tell it more, much more.

So, okay. So these are where you can, these are other baby GPTs. This is some of the ones I use. You can explore GPTs and see and search all the GPTs that are available. This is the early version of the one that we're building. So if I click on it, up there it will change from chat for Omnivore to the home building advisor.

And I could just tell it, let me move everybody's picture out of the way, I could tell it this, tell me about spec houses. I'm giving you a really simple prompt first just to show you kind of what it, what it knows.

And away it goes.[00:49:00]

In the new version, I got things like, what happens if my spec houses don't move? What should I do? All kinds of stuff. So you can see, it's got a pretty big knowledge base because we loaded. Like 1500 pages of my writing into it and then sometimes at night I train it like I ask it questions and I tell it no dumbass.

That's not quite right. Here's the right answer. Add this. Clarify that. Try again. No, that's better. But you still didn't. This isn't quite true.

Kevin Weitzel: If you have any questions for Al or Bill, their contact information, Jim had that up on the screen in the recording, you'll be able to see that as well. If you have any questions on drafting services, interactive plat maps, interactive floor plans, renderings, virtual tours, do you want to contact me? Those other two guys. They're just going to point you back to me anyway.

Al Trellis: Everybody gets inspired, and I love this, because I'm 77 years old, and I'm the [00:50:00] guy who's talking about AI. And like, my grandchildren look at me like I'm nuts, and they're like, how can you be doing that stuff? And I'm like, because it's really interesting, and it's really thought provoking. And if you really have a thirst for knowledge, this is what it's really all about. It's, it's absolutely amazing. Every day I play with it, I'm more amazed than the day before.

Kevin Weitzel: Al, I'm going to use your quote because I have a lot of respect for both you and Bill, and I've learned a lot from both of you.

I truly do look up to you as, as a beacon of information and knowledge and just forward thinking and I don't like to look at age. I'm not an ageist if you will. Um, you know, everybody thinks you have to be young to understand this. But just like you said, you know, knowledge is power. Knowledge isn't power if you don't utilize it, if you don't leverage it, if you don't implement it. So, thank you for that quote today and for that expansion of that quote that a lot of people like to use. And thank you everybody for your time today um visiting us on this uh webinar. We really appreciate you and the time you took out [00:51:00] of your day to visit with us.

Al Trellis: I appreciate everybody's coming and I'm going to the beach.

Kevin Weitzel: So thank you everybody for your attending today. Have a good day.

Al Trellis: Thanks everybody. Thanks Bill.

Bill Gelbaugh: Thanks Al. Thanks Kevin.

Al Trellis: By the way Bill someone called me a stoic the other day. There's a friend of mine who's, who's just been diagnosed with a problem and I was telling him about a medical thing that I have and, you know, I was basically telling him you can deal with this. It's not that, it only seems more difficult. It always seems more difficult than it is. I mean, my wife thinks I'm a whiner, but I think I'm a stoic. So, I guess I'm somewhere in between.

Bill Gelbaugh: I think you're, I think you're a stoic. It's wonderful. You know, my son, uh, just turned 50 and he's had some struggles and stuff and he's, he's really grown a lot through stoicism. We trade, you know, stuff back and forth, you know, throughout the week, and it's made a big change. So, it's fun, fun to have other [00:52:00] people to talk to about it.

Kevin Weitzel: So Richard, since you asked, uh, how long bill, a little over a year and Al has been, uh, considerably longer because he was actually at the ground level with OpenHouse AI. So he's been involved, uh, considerably longer, but yes.

Al Trellis: Was that the question? How long have I been involved with AI?

Kevin Weitzel: Yes sir.

Al Trellis: Uh, it's about I really got heavily involved about a year and a half ago. I was on a cruise and one of the guest speakers was talking about AI and giving some demonstrations at the time about some early stuff. And I was like, holy crap. I didn't understand this, but I had been working on this stuff for about a year and a half before in a more focused way with open house AI on the home building side of it.

Kevin Weitzel: All right. I just got my note from Jim saying that he's about to shut it down. So have a fantastic rest of your week, everybody, and we'll see you on the flip side. [00:53:00]