Latest / Remodelers On The Rise / DataX and the AI Workforce Every Remodeler Needs

Transcript

DataX and the AI Workforce Every Remodeler Needs · Remodelers On The Rise

• Kyle Hunt: Thanks for tuning into the Remodelers on the Rise show. Whether you're listening or watching, I appreciate you being here. If this was helpful, make sure you're subscribed so you don't miss the next one. We're putting out new episodes every single week focused on helping you build a better remodeling business with real stories, practical ideas, and things you can actually take and use. If you're on YouTube, hit that like button and turn on notifications so you know when new episodes drop. Welcome to the Remodelers on the Rise podcast. Peter, what number podcast recording is this on the Remodelers on the Rise podcast for you? Peter Ranney: I don't know. We didn't talk about that before he came on. I'm not sure maybe three or four. Kyle Hunt: ⁓ are you somebody that doesn't like to be put on the spot? I'm sorry. Kyle: Before we jump in, a quick shout out to our friends at BuilderFunnel. AI is changing how homeowners find remodelers, and BuilderFunnel is already helping remodelers win big. Their clients are becoming top recommendations in ChatGPT and other AI platforms, generating new referral leads directly from AI search and seeing results like a 400% increase in AI driven referrals for a Charlotte remodeler. Kyle Hunt: If you're listening on a podcast app, a five-star review goes a long way and helps more remodelers find the show. We've got great links below or in the show notes where you can connect with us, check out our remodelers community and learn more about our coaching and resources. Appreciate you very much. See you on the next episode. Peter Ranney: I'm assuming three or four. I'm not sure Bailey can fact-check us Kyle Hunt: I bet you four. I bet you four. I'll have my people check into that. Probably number four. ⁓ Elliot, this is your maiden voyage. How are you feeling? Feeling good? Peter Ranney: Okay. Elliott Wittstruck: Yeah, feel great, man. I've got my sea legs ready. Kyle Hunt: ⁓ he now he's quick on the feet. I said maiden voyage. He's using he's using nautical examples and whatnot. right. Elliott Wittstruck: You Peter Ranney: It was his, his dad was in the boat industry, so he's already got those references locked down. Kyle: But it's not just leads, they've generated multiple millions for their clients from AI already. If you want your remodeling company to be one that homeowners find first, whether they're searching Google or asking AI, head over to builderfunnel.com slash R O T R. Builderfunnel.com slash R-O-T-R. Kyle Hunt: ⁓ interesting, interesting. So I've got Peter Rainey, I got a couple titles. Rainey Blair-Wiedman is the remodeling side. Sunshine on a Rainey Day is the nonprofit. Do look up Sunshine on a Rainey, R-A-N-N-E-Y Day ⁓ in Roswell, Georgia. And for the last ⁓ one year, five months, he has been a remodellers on the rise peer group coach. He runs our Ignite peer group. And then I have Elliot Whitstruck. Built with love. I like your logo. I want your website. I'm like, ah, this is just so nice down in Nashville, Tennessee. And then both of you are also, what's your title with DataX? Kyle: One proof that AI visibility is driving real business. Builder Funnel has helped their clients generate qualified leads and projects from ChatGPT and Gemini, expand into new markets while attracting higher-end projects, and helped a Jacksonville remodeler become the number one recommendation across AI platforms. Most agencies are still talking about AI. BuilderFunnel is already delivering measurable results with it to the tune of. Elliott Wittstruck: founders, technically lead a product for me and lead of marketing or lead of sales for Peter. Kyle Hunt: founders. All right, there we go. So we're to talk some data X, which is their AI custom AI agent. Did I say, did I even say that right? Elliott Wittstruck: Yeah. Peter Ranney: Yeah, doing great. Kyle Hunt: Okay, good. Those AI things, it can be tricky, right? ⁓ So we're talk about DataX, and we're also gonna talk about this topic that nobody's talking about these days, which is AI. In particular, how do we use it? Not just like, let me give you these pie in the sky ideas. Let me give you this 50,000 view ideas. No, no, no. Let's get into the practicality of it. Let's get into the messiness of it. Let's get into the clunkiness of it. Let's also get into the time savings and the efficiency and the productivity that can come from it. Kyle: ten million and counting. Visit builderfunnel.com slash R O T R to see how they're helping remodelers like you stay ahead of the competition. Builderfunnel.com slash R O T R. Kyle Hunt: And ⁓ maybe this title of this is just getting real with AI and digging into that. So if we can start by, give us a little background on ⁓ kind of how DataX came to be. What kind of problems were you trying to solve? How did that get started? Peter Ranney: Yeah, DataX started last January and really LA and I as business owners, construction business owners, we're trying to solve our own problems. And I had much less technical experience background. LA has much more technical experience background as a coder and all those, that nerdy stuff. I'm like a blue collar nerd so to speak, but we're trying to solve business, like problems around businesses. And that started with, well, I would like to have like Zillow information on my leads, or I would love to have like a cover photo from Google on my job. So I don't have to like populate a new photo. So it's just like some simple things and quo and dial pad. Those became relevant to as him texting and calling these, my clients. I love that information just to automatically go into my record, my customer record. And so we were just kind of creating these things on our own. And to solve our own problems, we're telling our other contractor friends, buddies, things like that. They're like, well, I would like to have that. And really just kind of spawn this idea of data X and let's, let's make a service available for other contractors at. They probably are facing the same issues and want the same solutions for their company. And of course, fast forward through that, we have a bunch of integrations, probably 20 plus integrations. AI has officially hit the scene. It's been here for a while. It's been here in theory since like the fifties, but it's here, here now. And so we've transitioned. Yeah, it's all used. Yeah. Like the 1950s. Yeah. You know. Kyle Hunt: Since the 1950s? Elliott Wittstruck: Yeah. Peter Ranney: But it was all theoretical research. How could we research things faster, more efficient, more consistent? And so the thought process was there, but we didn't have the compute power, all the nerdy. So we didn't have all that stuff to make it happen. And now we do. And it's now at our fingertips to use. So ultimately the toolbox, we still do that. A lot of our users love the toolbox integrations, but AI has definitely taken like the forefront, the stage, like taking all the spotlight and in a good way. Right. And that's where we've kind of developed these agents. started out with NCP, which is also super powerful for more, I'd say maybe some more advanced users or users trying to do like bigger things. But the agents is like a place where we really settled in and these, this idea of like an autonomous agent that can be set up specific to your business and do things. Now, with that being said, we see a lot of users trying to set up these, these kind of like monolith things. It's like, this is going to be my CEO. It's going to make decisions for me and read my emails and make, you know, do all the stuff. Start small, like start literally small, like grammar checker, right? A grammar checker agent to just make sure you're having a decent proper sentence structure, misspelled words, things like that. I think the opportunity now for most business owners comes down to ⁓ reporting and having as a small business, we've never had good reporting structures. Like people were porting information back to us. That's like the best use case that we're seeing now is to have your source of truth job trade, whatever that could be like. Kyle Hunt: Hmm. Gotcha. Peter Ranney: report back to you, get emails in your inbox or whatever that is. So that's, think that's simplest use case now. Kyle Hunt: Peter, Peter, Peter, slow down little bit. We got like seven, seven ideas going there. ⁓ Elliot, what would you kind of, what would you kind of say related some of what he was sharing? Elliott Wittstruck: I would back up. You asked how we got started. ⁓ I'm built with love and this at the same time. So my first thought is like, how do I manage both of those? How do I manage both of those companies? And so like finding the solutions to do that. So it's like, okay, can I automate or can I get like Peter said, the Zillow data, I was doing that manually and built with love. How can I automate some of this stuff in Built With Love so I have more time to do data X? How can I automate some data X stuff to make sure I'm focusing on Built With Love and back and forth? So we're looking at like receipt processing, lead intake. Like you said, the grammar checker is pretty big. How do I do material procurement? How do I do support tickets in data X? How do I do something there to help with that, alleviate that? So it's always just thinking about what do we need ourselves and kind of building that. Kyle Hunt: Mm-hmm. Gotcha. Yeah. And I think I just want to kind of get this out on the, on the radar of, you know, I asked you what problems were you trying to solve with data X. And I think my, my kind of hypothesis, the thing I've been thrown out to my clients lately is, Hey, instead of looking at everything and anything that, that people can do with AI also comparing yourself to somebody who has literally invested a hundred plus hours in understanding AI. And then you going, well, shoot. Elliott Wittstruck: direction. Kyle Hunt: They look at, look at what they're able to do. Yeah, but they've done the work. This is not a snap of the finger. I don't want to pour like too much cold water on it, but I think it's good to pour a little cold water on it of going, this isn't like super dirt simple. just click and it's done. Like you've got to wrestle with this. You have to understand it. You have to invest time. You have to bring in experts if needed on it. But I think that idea of stepping back from AI and as a remodeler, I'm saying what problem do you want to solve with AI and get in like Peter and Elliot, like you're both saying. Let's get a little bit more micro with it. Let's find a specific use case and let's set that up and let's get that going. It's similar to, you know, when somebody five years ago was getting started with a program like job trend and people were like, ⁓ I'm so overwhelmed with stuff. Well, what'd we tell them? Hey, a piece of the time, just start with this part. Start with this part. Do you agree that when it comes to AI, maybe going a little bit more micro might be ⁓ a less annoying slash frustrating way of approaching it? Or am I being a wuss? Peter Ranney: No, it's, it solving a problem in your business doesn't exist as a problem in itself, whether it's trying to implement AI or whatever, any kind of process within your business. as an encouragement to most people, there's a chart that's out there on the internet and it's kind of like the usage of AI and like 84 % of the world has never even touched AI. And so you're talking about like 0.3 % of the world has paid for AI. So Kyle Hunt: Mm-hmm. Peter Ranney: Even us talking about it, or people listening and trying to get to that, you're so far, you're like way ahead of the curve. so like taking, know, start, start small. it's, it's that, that's going to allow you and don't, don't feel like you're missing out. Like FOMO is real in this, right? You're seeing there's a, person set the other thing and it does all stuff and I want that mobile. has, he has different problems in your business. So that's why he set that up. Kyle Hunt: Yes. Good, good. Elliot, tell me AI agents, AI agents. What is an AI agent? How do you kind of explain that in plain language? And then where I kind of want to start moving towards is, let's look at some specific examples. I might do a little screen sharing, might get into the weeds of things, but AI agents, tell me about them. Elliott Wittstruck: Sure, let me ask you a question, Kyle. Do you use, when you use AI... Kyle Hunt: Ew, whoa, whoa, time out, Elliot, time out. Elliot, you've never been a guest on my podcast. I'm the boss. I'm the one that asks questions. I don't like to be put in the hot seat. That's why I built a business where all I do is talk. Okay, permission granted. What do you got? Elliott Wittstruck: Okay, it's an easy question. When you use AI, how do you get it? How do you interact with AI? Is it typing or talking to it? Kyle Hunt: ⁓ Primarily typing is how I use it. Elliott Wittstruck: Okay, and then you ask it to perform tasks and questions, I'm sure, Or answer questions and research. Okay, does it answer those questions or do the research without you typing or talking to it? Kyle Hunt: Yes, that is true. ⁓ not very well known. Or at all, I suppose. Elliott Wittstruck: Yeah. So the idea of an AI agent is having it do what you're asking to do automatically without you asking to do it. So, and how you get that is like a prompt or a set of instructions you give it. And then it is triggered maybe on by emails or by a text message or by a web hook, which is like a trigger from other web, from any piece of software sends a web hook, which triggers it. Kyle Hunt: Hmm. Elliott Wittstruck: Or you set it on a schedule every day at five o'clock, do a project manager analysis of my active jobs. Those are agents because they take your prompt or instructions and perform the task automatically. And so it's essentially just performing the prompt that you would normally just type to it every day at five o'clock. You might go type, hey, let's do an analysis. Instead of you typing to it, you're already starting the grill up because you're trying to get dinner early, right? So you can have more playtime ⁓ with the family and kids. Kyle Hunt: Hmm. No, we're not starting the grill up early because of that. We're starting it up because as a 43 year old man, I see very clearly how we are going to be eating in the four o'clock hour. Five o'clock on the dot rolls around and I'm ready to go. Like it's dinner time. What do mean we're eating at 615? What is this crap? Elliott Wittstruck: It's Yeah. Yeah. Well then you don't have time to type to your agent or type to your AI. So the idea of an agent is for it to be scheduled and it runs that prompt and analysis for you at the time you want it to. Kyle Hunt: Bingo. Cool. So, and if you were to describe data X to me and marry it with AI agent, describe that a little bit more. Elliott Wittstruck: Okay. So data X is, it has a bunch of different components. Peter talked about like our automation integrations for like Zillow data, Google street view, Google satellite view, a Quo and ring central integration. That's like an automation that's, that's built inside of data X that you can connect to your job thread. And then the other part is like what we just talked about, like typing to an AI that is also inside of data X and it's connected to all that other data we just talked about. And then the agent part is it can do those prompts that you're typing to it automatically on these different triggers, whether it be on a schedule, from an email that you send it, or from a webhook from any other software. And it can encompass all that data that we just talked about. it's kind of, DataX was built on this platform and it just kind of grows and encompasses all of your data and it kind of connects it, runs instructions, runs reports and analysis based on how you want it to. Kyle Hunt: Good. I'm just thinking of the people listening to this and going, some people are starting right from the very beginning. Some of them have some different things going again. Can't encourage you enough. Like just start with something small, start with something simple. So for example, if people were like, all right, I already got, I got my job trend all set up. I got that going. I sign up with data X, ⁓ and get that moving, walk me through, walk me through a few of the steps and then show me one of the AI agents. Ooh, we're already looking at me go AI agent. understand what that means better than before, along with my listeners. This is going great, but walk us through a few steps and then maybe show me the receipt feature or show me that Zillow feature and let's get practical. Elliott Wittstruck: Yeah, we'll show you something easier. We'll start with Peter likes to show this one off. It is the receipt agent. I'm not a receipt agent. It is the grammar agent, which is when your project. Kyle Hunt: Grammar Agent. Peter, why do you like that one so much? Do you feel like sometimes you might mistype things? Peter Ranney: of course, you know, construction field guys, got big fingers or callous, you know, they're not sensitive typers. They're typing fast. Kyle Hunt: ⁓ Elliott Wittstruck: They're typing fast and maybe their idea doesn't always get there. Peter Ranney: Or if they're dictating, right? It's like the dictation is always perfect. Kyle Hunt: Yes. It's good. Elliott Wittstruck: And so what a grammar checker agent would do is you could set it up maybe every time your project managers or field guys submit a daily log. It goes through their daily log and just updates the grammar to be more professional or takes out the misspellings. Same thing with comments. That would be a good way to use a grammar checker. And it's not intrusive at all. And so that can just set up and run in the background. Like you said, Kyle, to like experiment with some AI integration to make sure. Kyle Hunt: Hmm. Elliott Wittstruck: you know, you're getting your toes wet. Kyle Hunt: In that example, that would be a data X AI agent. The grammar one is coming in. Okay. Yes. So if you're listening to this on the audio version, glad you're listening to this. If you go to remodellersontherise.com and click on podcast, you'll see this episode probably near the top, if you're listening to it. And then you can click on the video version if you want to see, ⁓ see the screen. We'll put the video version. Elliott Wittstruck: Mm-hmm. You want me to you? Kyle Hunt: in the show notes as well, so it's even easier. Elliott Wittstruck: So I've got my grammar agent up here, but how you would get to it is if you're on data X, you would click on this agents tab here and maybe you don't have any agents yet. So you want to add a new agent from the library. We've got all these pre-built agents for you. I would encourage you to read through the instructions and update them, but the grammar checker, when a new common or daily logs created inside job trade, you will receive webhook data. Your job is to look at the common and daily log and fix the grammar to be more professional. while still capturing the essence and emotion of the content. Update the comment and data log with your edits. When finished, give me a summary of what the previous content was and the new content you shared with. So you would install that. Kyle Hunt: So that last part is where do you view that? You view that in data X before you approve it. Elliott Wittstruck: No, I let it auto approve. I let this just do it. And it just shows if I go back through my grammar checker logs, I can see the logs, the historical logs of everything it has done. So that's where I would see that. So you would install it. Kyle Hunt: Okay, okay, okay. And even, and apologize, going back one more step, sign up with DataX, some linking to go, I need DataX to communicate with job treads, just some getting started steps that's pretty straightforward. And then once you have that connection going, bam, you can start messing with the agents. Okay, thank you. Elliott Wittstruck: Mm-hmm. Yeah, exactly. ⁓ So we can go through here. Here's one that came in. So we got the webhook. And the webhook looks crazy, but the AI knows how to read this. And so this is asking, ⁓ this is just a message, an internal comment logged from Cyril asking what the paint colors were for. And then it updated to, could you please confirm the paint colors for the SH back offices? rather than just paint colors for the SHBAC offices? Could you please confirm that it's so important. It's just four little words and it might not seem big to some people, but to me, that's a huge deal because we're always about providing context. We need to be able to take everybody's message inside of the company. ⁓ Every message needs to stand alone in context. So we don't get confused and the clients will get confused. Kyle Hunt: Yeah. And just professionalism, right? Professionalism in the Hollywood community. So ⁓ again, this one, I clicked, I was out in the field, I typed in, colors for SAH back offices, question mark. And when I click submit there, the agent ran right then and there, updated it. By the time the client sees it, it's probably the updated version. Okay. Elliott Wittstruck: Mm-hmm. Correct. Yep. Yep. Yep. And it probably, it probably took two seconds for it to do that. And so the client would not have seen it that quick. I don't think. Yeah. Kyle Hunt: Yeah, yeah, good. that one's a simple one. That one's a simple one. Hey, every daily log that we're creating, every message to clients, it's automatically grabbing that. That was fun. Move on to the next one. Peter, what's one of your favorite ones? Elliott Wittstruck: Okay. Peter Ranney: the receipt processor. Well that is our most popular. Kyle Hunt: What do love about that one? Peter Ranney: Well, it's the biggest part is job costing, right? So like a lot of contractors, we suffer from poor job costing. Well, now you get an invoice or receipt in your inbox or maybe you're at Home Depot or Lowe's or Menards. Take a picture, email it to the agent. It runs to the instructions that you've set up and then it finds your job, finds your budget, finds your cost item, uploads the set attachment and job costs it for you. you can mark it as paid. You can mark it as final if you care to, but there's It's a clean, just simple way to like have this usually tedious process of a pile of receipts. Where do these go? A lot of clicks through job tread. Kyle Hunt: Now, and Elliott, the trigger for this one is sending it to a specific email. Elliott Wittstruck: Yeah, so each agent in data can get its own email. So this one for me, bwllbuiltwithlove is my prefix. And then I can set this agent to receipts. So bwll.receipts is the email address that this agent gets. So I can copy this right here, go open my ⁓ Google and just start forwarding email receipts to it. And it's going to follow a set of instructions. Peter mentioned it just goes through some steps. Kyle Hunt: Pause there again. I'm pausing just because I want to get super practical. that's my email for that and I have my carpenters taking receipt pictures, my project manager, the other step would be to, if we're using Gmail as an example, is kind of add that as a contact. So when I type in receipts, it automatically pulls bwl.receipt.agent. But now everybody's got that. All I need to do when I see that receipt is click forward. start typing in the word receipt, there's the email. And when I click send, that's the trigger for this AI agent to go through the receipt processor process. Elliott Wittstruck: Yeah, exactly. And so here's our dumpster bill from the other day. And so it came in via email with an attachment here. You can see it was $28. A light trailer that day, I guess. And so it starts by reading the PDF and finding the vendor and everything. And it finds the job that it was for, finds the total, finds the date. Kyle Hunt: 28 bucks? ⁓ man. Elliott Wittstruck: and it goes through and it creates the bill. And so now in job tread, we could open up that vendor bill. We would see it costed to the job, to the cost item for dumpster on that job. And we would see it already marked as paid. It sounds like an easy task, but we can just, we can just for the emails and no one has to do it. And it just runs in the background while you like keep your brain on other more important things. Right. And then you just come back and check it or check in job tread. You don't even have to go back to data to check it. Just check it in job tray because you'll see the vendor bill with the attachment and the job costing there. Kyle Hunt: Next one, please. This is too much fun to only start at two and be done at two. Or some other ones. Peter Ranney: How about the project manager report? Yeah, love this one. Elliott Wittstruck: Okay Yeah. So this, yeah, this was Peter's idea. Like every day at a certain time when all the work is done, let's look at all active jobs. We're going to review each job and we're going to look at budget and performance. We're going to use a financial snapshot and we're going to look at the schedule as well. So we're going to look at schedule, what was budgeted on the schedule or where it was versus what the baseline versus what it is actuals and we're going to look at the budget versus actual costing and then we're going to read the daily logs and comments for the day and see if there's any red flags, green flags, yellow flags to report. It's going to put all. Kyle Hunt: Hmm. And Peter, this was one of the first things you were mentioning when you kind of were going through the list of, um, yeah, we can do this, this of going, all right, we've got all of this information in, in this case, job tread. How can we just pull a report out of it where I, as the owner or even as the project manager can get a little bit of a pulse check on where we're at schedule, job costing wise, budget versus actual, uh, et cetera, and have the AI agent called PM daily overview. courtesy of having this agent set up in data X have this run for us. And you said daily, this could be do this on Tuesdays and Thursdays, do this every day, whatever the case might be. Yeah, click, click done. Elliott Wittstruck: Yeah, you can set your days, frequency weekly, every 15 minutes, every hour. You set your time. So I have this running at five o'clock central. And if we look at ⁓ something like this, it's gonna, so there's the instructions just automatically populate this agent to start at five o'clock and it's gonna review everything. You can see these are all different tools. It's working, the AI is working really hard. It's working some more, it's working some more. Kyle Hunt: Mm-hmm. Elliott Wittstruck: and it sends an email. So it gives me an overview here. It doesn't give me the actual email in here, but if I open my email, I remember this one being several paragraphs, each job having maybe two paragraphs of like deep analysis. Here's your red flags. Here's what to do. Here's how to get back on task. Here's what you're doing well. Here's who you need a compliment from a good job today based on XYZ. And it's pretty cool. Kyle Hunt: Good. And it's, one of those things where it reminds us also just big picture technology use overall. A lot of the power of AI is also going to be corresponding with how well are we just utilizing our software and technology? If we're not doing daily logs, that's going to be a problem. If we create a schedule and we don't update the schedule because something changed, well, the report that we just ran isn't going to be as strong. If we haven't fully integrated all of our. our AR and AP in here, got our cost code. So it's a reminder that make sure your core software is being used regularly and consistently. And then when we have these AI agents that can pull that data and simplify that data for us, now we've got a really nice combo. Is that fair? Okay. Elliott Wittstruck: Yeah. And I think Kyle, going back to another thought you had of about how do people get started? These agents might seem big and daunting. Even using it straight out of the library might seem like a huge task. What's really cool is like we built these agents to update themselves. So if you just open a blank agent that a little bit. Yeah. So this was the first prompt I gave my Kyle Hunt: ⁓ is this the whole robots are taking over going to take over the world thing? crap. Elliott Wittstruck: I gave this agent, this PM Daily Overview agent, I said, write instructions to look at active jobs and look at each budget versus actual, then highlight red flags, breach red flag, offer a solution, write these instructions yourself. It wrote the instructions we showed you or that this agent uses. And anytime we need to update something, because it's just need to tweak it, you can just ask it to update itself because it knows what it needs to get the results you're asking for it. Kyle Hunt: Hmm. Hmm. So we're, so we're focused on the AI agents. Now, when I look in here in inside data X where it says new chat, that almost looks kind of Claude or chat GPT-esque. Tell us, tell us about that. Elliott Wittstruck: Yep, that's what it is. So one thing that data... Kyle Hunt: ⁓ Peter also I heard to tell me not too long ago, know a chat GPT guy chat GPT is the horse and buggy of AI. I have never been insulted and cut to the bone so hard. And so Peter Ranney: There's nothing wrong with the horse and buggy if you like going slow and, you know, enjoying the scenery. Kyle Hunt: Hmm. Hmm. So for those people that are using chat GPT, psst, get over to Claude. Is that the punchline of that? Peter Ranney: I mean, I think it's a race and at one point someone like Claude is currently ahead. There's indications OpenAI has put a lot of money and effort into their ⁓ competitiveness with Claude. So six months from now they could be ahead. But Claude, think overall from feedback is the best. Elliott Wittstruck: Yeah. But both of those frontier, we call them their labs, like their AI labs, and we call them frontier labs or frontier models. They use every single ⁓ thing that you type to it as training for the next model. so Kyle, you asked like what, like this looks like ChetGBT. This is a little different. Peter, if you want to explain what DX1 is and maybe why someone would be interested in this versus ChetGBT. Peter Ranney: When it first came out, this AI stuff, users were asking us like, what's hap- what happens to my data? Right. And this idea that Elliot's describing is when you type your stuff in, goes into some thing that we don't know what that is. So we, we host our own AI model. So your data is secure, like within our system. It never gets released. Like the other AI is like, it can't be trained on in that sense. So there's like a security factor because you're, you're asking it for, you know, your business information. Kyle Hunt: Mm. Hmm. Oof. That adds a whole other level of work that is involved on the data exercise, I would imagine. Elliott Wittstruck: So, so. Yeah, it is. But it's, ⁓ we like to be nerdy about this kind of stuff and try to stay forefront. this is a huge, a lot of people are concerned about security. So if you don't, if you don't want to deal with that, or if you don't want to expose all of your employees and your company to that, you like on a single data X account, they all get access to this. ⁓ This, this DX one AI model that the data X is hosting. Kyle Hunt: Some of us do. Elliott Wittstruck: It's a safe place where they can only talk to your job tread and it has basic general knowledge. So you don't have to worry about them doing like ⁓ side work using your AI subscriptions or security issues or leaks and data leaks like that. Kyle Hunt: Yeah. What are, what are some of the prompts that you'd like to put in here? Some of the questions I see a couple of them down there of like, show me my active jobs. Boom. Pulling that out. But what are some other things you guys enjoy putting in there or users do? Elliott Wittstruck: Peter, do you have some favorite prompts? I've got some big nerdy ones. Peter Ranney: Yeah. Some big nerdy ones. I like to use a chat for like simple, quick things. I'm trying to understand on like specific jobs. So, yeah. Am I over budget on this Smith bathroom job or like simple, quick stuff like that. Elliott Wittstruck: I like to ask it, ⁓ so there's this new theory that I'm kind of trying to do with my AI. Instead of me telling it what to do, I'm trying to get it to tell me what to do. So I try to give it all the context and all the data about a situation. And I say, okay, tell me what to do, or how do I get the most ROI? how do I get, these are the goals help me get to, help me get to those goals based on this context. And so that's kind of where I'm trying to go to. So a good one is like, hey, you're an expert in... you're an expert in residential construction in middle Tennessee. And I've got a client that, ⁓ refuses to pay until, ⁓ this certain milestone, but it was outlined in the contract at this milestone. You know, what's the language I need to help, you know, get there or how do we, how do we get to resolving this? Things like that. Kyle Hunt: Okay, got it. We were in St. Louis, not too long ago last month, and one of our peer group clients, Jacob with Home Run Design and Remodel out in Washington, he said, all right, we kind of set aside 45 minutes, and he took his technologies and took DataX and took JobTread and took Plot and did that, and said, let's create an actual estimate. out of this, that and the other, and then taking it and schedule it. So we're talking about some very practical things. Now, if I'm a remodeler and what Jacob did, and I just kind of like your guys' comments and your experiences with this, he had plod, do I have a plod? I have my plod, it's over there, P-L-A-U-D, plod, which is more of an audio recording, and he recorded an hour and a half on-site meeting with his client where they're talking about everything. He had a prompt that he put through, ⁓ the recording through to organize it by room. So now he's taken audio transcript and organized it by room into actual transcript writing. He then said, all right, here's my prompt in Claude, in Claude with a C, you all with me still? Start to create kind of a scope of work, da da, and then he took that results from Claude, put it through data X and that transcription and Elliott Wittstruck: and that's prescription and that's... Kyle Hunt: DataX and JobTread communicated back and forth, by the being it spit back out an estimate for us. So that's a little bit more advanced. I don't think I butchered that. I think I did a pretty good job of explaining that. Talk to me about some of this more advanced stuff like that. What are you guys seeing? What are you guys working on? What are you guys realizing for time savings, et cetera? Elliott Wittstruck: That is a good one. call it a ⁓ lead in Peter calls it lead intake or I call it, I call it cold lead intake. And it does the exact same thing from maybe an email that we get. have a web form on our website, the lands in our inbox. We scan it real quick, but then it gets forwarded to this agent that does ⁓ lead intake. And it does pretty much everything you explained except the recording. Cause we haven't met them yet. So it gives us a great baseline to then call them. Peter Ranney: Oui. Elliott Wittstruck: set up a meeting and present to them a lot of data pretty quick. Peter Ranney: I've seen some use cases where contractors, they go to the house, they identify the scope, you know, in their business model, they identify the scope quickly. And essentially whether it's a dictation or they've written down notes, they basically sent it to the agent, like on the way back to the office and the instructions they have are something about, you know, building my budget in this format, my schedule, and come back to me with any comments or concerns or red flags you might, you know, I might've missed. And then they have things in there and their instructions saying, I'm you, usually miss on dumpsters and I don't count for enough port of potties and all that kind of stuff. So he gets back to his office and he has his job tread, the schedule budget and these notes kind of already like ready for him to go. this that's, you want to talk about producing estimates quickly and you know, that's, that's like top notch. Kyle Hunt: Mm. I feel it. Go ahead, Elliot. Elliott Wittstruck: Yeah, I was just gonna say, Kyle, your example, the next step that we have that I use in Built With Love is after the estimate, it creates the contract with all the pay phasing. puts the pay phasing in the schedule. It builds out the rough scope of work in the schedule with all the pay phasing attached as dependencies. And then it'll build all the draft invoices for all the pay phases and it'll build like the draft contract to go with that. Kyle Hunt: Mm. Elliott Wittstruck: I did see another good example this week. you might have to help me. Was it Anshul? Was it Michael Anshul that had, when they do an estimate, everything we've talked about except the next step is that they build, his AI builds a landing page. So now it's like his website.com slash Smith bathroom that they can open up and present to the client. And they can see it all there with check boxes to change, to change numbers to get, they took pictures and it like, Kyle Hunt: Hmm Hmm. Elliott Wittstruck: He runs it through Google's Nano Banana to read the AI render, these images that get uploaded on this thing. He says it takes about five minutes so he can do it while he's in the house. And so before he even leaves, he's like, hey, let's go through your landing page presentation. And it's like this beautiful thing with six photos of their space that he got when he showed up that have AI rendered to be what they were looking for. Peter Ranney: Yeah. Kyle Hunt: Hmm. Hmm. Peter Ranney: And it is the one example, we had a call to action. You could click and say, great, sign me up for your project development agreement. Kyle Hunt: Right, the possibilities are already vast right now. I mean, it feels like over the last 12 months is when things have started clicking a little bit easier and stronger and better and seems just more realistic. And we can imagine that 12 months from now, 24 months from now, obviously it's gonna hopefully get more and more easy there. Again, I think the getting in the weeds and getting very practical is important because the examples that we just gave, some people are like, That would be so cool. That'd be amazing to go to. And it's like time out, just walk before you run. You haven't even signed up for data X yet. You haven't even gotten some of these other basic things going. Like get that going. Even the process. What I've liked about this recording is if you're kind of new to AI and I said, all right, define an AI agent for me. People are like, ⁓ I know that better now. There's, there's so much to learn. There's so much to dig into. The other thing I would mention to anybody listening to this, maybe present company excluded, cause you guys are obviously eyeball deep in this. there's just flat out regular work that some of you remodelers need to be doing. And you need to give yourself some guardrails. Maybe you guys need to hear this too. On how many hours per week you are allowed to devote to AI. What I like to say a lot of times is you basically, as far as on the business time, a lot of us might have 10 % of our time be on the business time. Peter Ranney: Thank Kyle Hunt: The other 90 % is in the business. If I'm a business owner, I'm doing sales, I'm doing some marketing, I'm doing some content. Yes, I'm developing some scope of work. Yes, I'm reviewing some designs. Yes, I'm thinking through production meetings, et cetera. Most all of that work will continue to go on. And what I'm seeing a lot of people do is devote 40 % of their time to researching and digging into AI. And all of a sudden now they're behind on other things. At the end of the day, the client experience is still number one. You need to deliver a remarkable client experience for your customers. but give yourself a little bit of a break. You do not need to have AI all integrated into your Mali business in the next 40 days. Think about it over the next 40 months. In the month of May, I am going to get this part of AI researched and figured out. In the month of June, I'm going to get this. Take some pressure off. It's not a sprint. It's a little bit of a longer runway. And frankly, two months from now to probably be easier to implement some of this stuff. And four months from now it'll be easier and Ellie, it'll be up until two in the morning, figuring out that next agent. And Peter will be like, Hey, I got this idea. How do we get this done? So I just want to emphasize that chill out, everybody chill out. It's going to be amazing. It's going to continue to be amazing. There's already practical things that you can be doing in your business, but it's okay. If you don't have it all done, focus still on your team, focus still on your clients, focus still on improving processes. And keep integrating AI as you go. And you look up six months from now, you're like, man, we've got this, this, and this done. But I just feel like some people need to hear that. And also, again, what problem do you want to solve with AI? Start with something small like that. And don't be overwhelmed. Sorry, I had to get on my soapbox there, Elliott Wittstruck: It's rude and you don't think systems are really needed. No, 100%, I think that's a lot of what we were experiencing at the conference last week together when we were there. Yeah, and it's like a lot of people were talking way up in the clouds. A lot of people were at the bottom, like kind of asking, hey, help us get up there. And there was like this big gap in between and both sides were frustrated, you know, on that same part. Peter Ranney: Yep. Kyle Hunt: You guys were at the FAST Mm-hmm. And that's what we've been talking about here. Like, let's get into the ground level of this. ⁓ so you sign up there. ⁓ so you click that. ⁓ so this how it works. This is where a lot of people need to be living right now is in seeing the actual how do we do it and then going and working on mimicking that. Hmm. Well done. Tell me about, go ahead, please. Peter Ranney: It's like, it's like anything else, like the levels of, you know, you're, you're like in our industry, a craftsmanship, right? You don't just show up day one, you're a master crafts master carpenter. There's, there's levels you have to pass through. kind of have this like seven levels thing that we've kind of view AI through. And most people like 11, level seven is like, you've got all your stuff connected and it's all doing, it's all autonomous things. And it's like, that's where people want to be. We need to be at level one and one, two, and three, which is learning how to use as like a, uh, know, drafting your emails. Kyle Hunt: Mm-hmm. Peter Ranney: reading stuff, reviewing stuff, like a research tool, a thinking partner first, right? You can just show up like, ⁓ I can do this master carpentry stuff, even though I'm an apprentice. It doesn't work like that. know, work up. Let's learn. Kyle Hunt: Yes. And then what's with the tokens thing? How does the tokens thing work? Define that and describe that please. Elliott Wittstruck: hey Kyle, can we share our screen again? I think that would be. Kyle Hunt: You can't, can't. me your best final six minutes. The best stuff that you want to share. Elliott Wittstruck: Sure. Peter Ranney: Okay. Elliott Wittstruck: Let's go through the seven levels Peter was talking about real quick, because we do talk about the tokens and context window and kind of what that is in there. So the first idea is level one, you're at a writing assistant level, autocorrect, grammar correction, drafting emails, drafting content. Level two, you're asking AI to ask you questions, and that's where I really like to live when I'm trying to get ideas out there. And I think a lot of people can jump up to level two AI, AI level two usage pretty easily by just changing how you prompt the AI. Level three is a research tool asking it to compare different materials or techniques, go find the hard to find details. had a customer that wanted a very specific drop ceiling style, uploaded that screenshot to AI, said go research and it found it from this like really random website. ⁓ Kyle Hunt: Hmm. I like this levels of AI adoption. It's kind of, it just relaxes the whole conversation a little bit, doesn't it? And people listening to stop trying to be at level six, which you haven't even seen on the screen yet. You're still at level two and that's okay. That's okay. You are a yes. It's okay. Elliott Wittstruck: Yeah. Peter Ranney: Totally okay. Totally okay. Elliott Wittstruck: The bottom of level three says you start asking it to write prompts for you. So you're saying, hey, I really want to build an estimate. Can you help me write a prompt so we can get to that estimate so you can build estimates first time? So you might go through ⁓ Claude or Chachibit, asking it to do all this stuff until you get the final result. You're like, that's a beautiful estimate, final result, whatever we just worked on. Give me a prompt so we can get that first time next time. Kyle Hunt: Hmm. Elliott Wittstruck: It'll give you the prompt it needs. ⁓ four is a power user. That's where it starts getting really fun. You're using projects, you're using context, which we'll talk about in a second. You're building artifacts and micro apps. you don't like a scheduling app that you pay money for. You ask Claude to build your own, and that's when you start building these micro apps inside of your company. ⁓ Kyle Hunt: Nice. Mm. Elliott Wittstruck: And we start connecting your business data. Maybe you start connecting job trade, you start connecting Todoist or monday.com or these other things to your AI. But context is, starts to get important here. And what that is, imagine like a piece of paper. Kyle Hunt: What do you know Peter and I it makes sense why we're wearing hats is that your actual hair? What do you got a hat on for? To Peter if Peter and I had that kind of hair we wouldn't be wearing hats Elliott Wittstruck: Yeah, that's my actual hair. ⁓ I didn't do it this morning. You know, it just takes a lot of work. It takes seven minutes to do it every morning and just skipped it. Peter Ranney: And AI can't do it yet for him. So, you know, put a hat on. Elliott Wittstruck: Kyle, you asked about context. You want to think of that as like a piece of paper. You have the front and back of the piece of paper. You can change the font size to get more words on the piece of paper, but the piece of paper is still the piece of paper. And different AI models have figured out how to get the font smaller, how to get the font on the front and the back of the paper. But once you give it all the context about your situation and it fills up the paper, when you give it new info, it has to erase the stuff that you gave it first. So that's where people... start hearing about like your context window, you lose the AI forgets what you told it maybe an hour ago because it was at the end of the context and it had to remove it to add the new context what you're giving it now. There's this thing called compaction, which when your paper gets 75 % full, the AI will go and compact the context down to be, let's say 15 % of context from 75%. Kyle Hunt: Mm. Elliott Wittstruck: but it's going to lose the details, but it understands the macro of what you're telling it. Right. And so compaction is one way to help use your context window. You'll see Claude, they just announced the 1 million token context window, which is like the biggest context window that we have right now. So you can give it a bunch of information. You can give it like four or five massive novel books and it would remember every detail of it. So that's what you want to. Talk about context but context relates to tokens tokens think about is like three or four characters ⁓ That make the smallest amount of characters that make sense in a word So like bird might be two tokens bi rd or bir D is separate it depends on the models all the models cut up cut up tokens differently and read tokens differently or if you see like ⁓ Characters right here. Ch a might be a token RA might be a token, CT, ERS. So you've got four or five tokens in that word characters there. That's the smallest chunk that makes sense to the AI. And so when you talk about context window, it's talking about how many tokens. So some models have 128,000 token context windows, some have 256,000 context windows, and some have a million token context window. Getting back to the levels. Level five, you're the builder. You start creating solutions on purpose. Kyle Hunt: Got some deep nerdy waters there for a second. I kind of liked it. I kind of liked it. Yeah, it was fun. Elliott Wittstruck: Yeah, that's good. You start automating tasks, you start connecting your business data seriously. So we talked about connecting job tread and maybe your to-do's, then you start putting it in an agent form where it's starting to run and analyze that data automatically. The architect, you immediately see solutions to all your problems using AI. So if you're a, remodeling space and you say, ⁓ I need to capture video. Kyle Hunt: Hmm. Elliott Wittstruck: and I need to get it into my computer and do a 3D model. You start thinking about how can I do that with AI or with software versus the old way of thinking where like, I just have to do this the hard way. I have to hire somebody. You start trying to figure out how to do it yourself or how to bring on different software and different AI to incorporate multiple softwares into a single workflow to get you the results you're looking for. That's the architect. Cyborg is you start automating tasks without human triggers. That's what we talked about, the AI agents doing stuff without human interaction and it just does stuff. So that's when you really start stepping into the cyborg era. Automations trigger other automations, ⁓ automated analysis triggered automated solutions. So the next level to that project manager analysis agent that we saw earlier. The next level, we haven't gotten there yet, but as if that project manager analysis at five o'clock every night, if it has a red flag, how cool would it be if it woke up another agent, said, hey, I see some red flags for financial stuff. Hey, financial agent, can you go dig into that deeper? Hey, schedule agent, can you go look into that deeper? So that's the next level. It doesn't really exist yet, but where the agents can start waking up other agents and they're helping and talking to each other. And it sounds kind of crazy. Kyle Hunt: Hmm. Mm-hmm. Elliott Wittstruck: But then there's a bonus level utopias where humans no longer work and we get to focus all our time on leisure and creative activities. I don't know if we'll ever get there. Maybe, maybe soon we'll see. But like I said, cyborg is very hard right now with the tools that exist. ⁓ So we're still a ways away from that, but we can dream, right? Kyle Hunt: We can. We can, folks. We can also get very practical and just think about the very specific, simple ways that we can utilize AI that have nothing to do with cyborgs and utopia. Some might even argue, is that utopia? I see that the definition is next utopia. Hmm. We'll have all kinds of time to be all philosophical because the cyborg will be doing the work. Who knows? I do like the fact that we're in the remodeling world. And while there are probably some robots that will be able to do some of that painting on that convert, da, da, da, da, all things considered, it's a really good industry to be in during the, uh, I was, I don't know the era, the era of AI. It's a really, it's a really good space to be. And from being a coach for 18 years, it's super interesting how I can look back and go through. had a number of years where it was like, wow, just the cutting edge people are starting to use technology. And you're still what? You're handwriting what? You're still, what's your spreadsheet? Wait, you're doing, you're doing that. And it's just been interesting over really since COVID and the year after, man, the, the implementation of software and tools to increase productivity efficiency has just been rapid. And now we're on this new kind of frontier of, right, we can use some of these AI things to simplify things for us. And I think that's the key. Make sure you're working on this to simplify things, not to overcomplicate. And you do have to wrestle. It's not snap finger and it's done. You got to dig into it. You got to work through some of those levels to get there. ⁓ If people want to know more about Data X, where do they go, gentlemen? Peter Ranney: DataX.TO, DataX, so D-A-T-A-X.TO. Kyle Hunt: y.to. Peter Ranney: dot com wasn't available. Elliott Wittstruck: connecting all your data to ⁓ all that too I was gonna I was gonna say you're connecting all your data Kyle Hunt: Peter's a straight shooter. He's like, that's what was available, dude. That's why. When I saw it earlier, I was like, I wonder why that's .to, but it's all good. DataX.to. Go ahead. Peter Ranney: Well, it's as, yeah, as I was saying, we're data X's we're connecting together. first two words, letters of together is T O. Kyle Hunt: Bingo, bingo. Some would say that those two characters are a token, just one token, but it depends on what model you're using. Sometimes that might be two tokens. How'd I do, Elliot? I was listening, I was listening. And then Peter, final word. If they heard nothing else, but this one thing, maybe a reminder or maybe kind of a good takeaway, what would be kind of your big takeaway for our listeners? Peter Ranney: right? Ha ha ha ha. Elliott Wittstruck: Yeah, that's great. Perfect. Peter Ranney: Overall, know, FOMO is real and don't be chasing that. Don't try to solve problems that don't exist in your business. Kyle Hunt: I like it. Bam, final word, even though I said final. I I guess I get the final word because I was wrapping it up. Huh? That's true. That's true. Gentlemen, thank you so much. Thanks for getting into the weeds of AI and also for teaching on a variety of things. I think the people listening to this are leaving it not overwhelmed other than that sideboard thing, but a little bit more empowered on how to do this. Check out datax.to and we'll see you guys soon. Keep up the good work. Peter Ranney: phrase. I had a lot of tokens left. had a lot of tokens left. I had to use them. Thanks. Good to see you. Elliott Wittstruck: Thanks.