Latest / 4kd.ai / Welcome to 4kd.ai: Turning AI Into Real Business Tools

Transcript

Welcome to 4kd.ai: Turning AI Into Real Business Tools · 4kd.ai

• Craig (00:00) Hey, welcome. It's an exciting day. It's the kickoff of Forked, a podcast for better in that. Look, it's a podcast about AI. That's what it's about. I've got David Moses, our good friend, and Jack Bevier here. We're starting a great new podcast, not just for real estate investors, but I'm assuming that, Jack, because we have a lot of listeners from Real Investor Radio that they might be listening, but this is going to be a podcast because we talk so much about AI. Jack BeVier (00:18) Thank you. Craig (00:29) on Real Investor Radio, Jack and David had this great idea. Why don't we start a podcast about AI, specifically what we're all doing with it, what we're all learning. And we have so much to share because we actually had to pull ourselves away from our bots just to record this episode. Guys, welcome to the kickoff episode of Forked. I'm so excited to share everything that we've done with the listeners. We really have no guardrails here. We're going to try to share all the sauce with you guys on how you can make your businesses better, how you can become more efficient, how you can have fun with these tools and really build amazing stuff. In fact, while you sleep. And we're going to share some links with you guys that you can come and visit to see what we're doing, see what I'm doing, and the guys will have their stuff as well. So guys, welcome to the kickoff show. Jack BeVier (00:56) Thank you. Thank Craig (01:19) I'll just say for me, I'm Craig Fuhr, long time real estate investor. Now I'm a loan officer with Dominion Financial Services, working with Jack and Fred. And I've taken a real interest in these tools, Jack. I don't think I've ever felt more creative. All the creative juice is flowing with these tools. And I just can't wait to share everything that I've experienced in the last really year of working with just Chat GPT. all the way up to Open Claw, which we'll discuss today. guys, who wants to go next? That's me. want somebody to introduce themselves. David Moses (01:50) All right. Jack BeVier (01:50) Go ahead Dave. David Moses (01:52) I'm David Moses. Companies behind me, Avenue Group Real Estate. I've been in real estate for 25 years. Maybe 26 years, yeah. It'll be 26 years next month. And yeah, so I'm also very, very excited. Always been kind of a techie, tech forward guy, but never like a computer programmer. And I was always kind of envious of those who could know, program. And now I'm not so envious anymore because, you now I have sympathy and whatever for folks who can program. ⁓ But I think it's wonderful where the world's going is, you know, it really feels like, it feels like what the invention of the internet must have felt like for, you know, geeks in the, the early 90s. Craig (02:27) Yeah, right. Jack BeVier (02:34) Thank you. David Moses (02:40) so I've been in real estate for 25 years and building tools. have a series of management companies that manage homeowners and condo associations to small single-family properties to small multi-family properties. Jack BeVier (02:42) . David Moses (02:54) I do some investing and a skosh of lending, but that's a very small part. And yeah, that's, yeah, I'm here for the tech. Craig (03:02) think what's cool about what the listeners are going to get from this podcast is just three really unique perspectives of guys that are in the trenches, in real estate, very entrepreneurial, all of us are, and super creative. At least I come from a much more creative heart. I think Dave and Jack are very logical and, you know, types. And so I love that perspective because when you start working with these tools, Jack BeVier (03:24) Thank you. Craig (03:28) you know, that none of those are really a barrier to entry. You know, it's really the tools will go as far as your imagination will allow. And so Dave, it's just great to be co-hosting the show with you. And I'm so psyched to share with folks everything that we're learning. Jack, why don't you jump Jack BeVier (03:38) Thank Yes, so was when we were we were chatting about this idea of doing an AI podcast Dave David immediately came to mind is what and because We are fortunate or I'm fortunate to be in the real investor Roundtable mastermind and Dave's been in there for many years So I get to see him and hang out in person three times a year and the past year David Moses (03:51) you Jack BeVier (04:09) and change my favorite. The thing that I'm excited about is what's David working on? What's he been building for the past four months? And every time I go and he blows my mind and I'm like, my God, like I got to catch up. This guy's building stuff himself. He's got it literally like got his portfolio hooked up to his QuickBooks. Like everybody's struggling in like the property management in real estate space to like really put a finger on what the AI use cases are going to be. And I'm like, This guy's got it. This guy's half his organization's running off of this stuff now. And so I've been I've been a fan boy of the stuff that David's been building and then like you said, Craig, though, like my approach is a little bit different. Like I've got people in the organization who are building these tools, but I'm not. I'm not like I just I don't know. I haven't I haven't been able to be, you know, to to pull myself in to be the one who's who's writing stuff. And I've just been trying to create the environment. Craig (05:05) You were the one that inspired an entire company to go out and learn these tools and it's time for you to start using them. Jack BeVier (05:12) I'm trying. Yeah, David Moses (05:14) So yeah, Jack BeVier (05:14) I'm sure. David Moses (05:15) the irony of your statement is not lost on me because literally the... Like I was already tinkering with this stuff, but it was actually a conversation you and I sat next to each other in, I think it was Baltimore. And I'm asking you... And I'm like, look, what I'm running into is I really want to connect all of the different communication points, because that's where management sucks, is the ball gets dropped because communication gets dropped. And that was, and you basically, I still have a picture of the napkin, I probably still have the napkin, which literally drew out. but it could totally work this way. You have a relational database here. You have a vector database here. And then you have like a model chooser here. And then, and you basically drew the whole thing out. I'm like, that could work. And so, so that, like that inspired me to really start. That's what inspired me to start using N8n. That's what inspired me to start really building, you know, agentic. you know, rather than just, hey, we can program this to do that if I hire this guy and tell him what I want. Craig (06:20) So Jack, why don't you bring folks up to speed on who you are, what you do, and what they can expect to hear from you on this. Jack BeVier (06:20) Well, that's great. Yeah, sure. So I'm Jack Bevier. I'm a partner at Dominion Group. We have Dominion Financial Services, which is a business purpose lender who make RTL and DSCR loans to investors all over the country. And then we have a real estate company, Dominion Properties in Baltimore. buy about a hundred houses a year, fix them up and either rent them or sell them. And then we have Dominion Management, which is our property management company that just manages that rental portfolio, which is up to about 900 units. So kind of got a vert, know, have the lending business and also have a vertically real estate, vertically integrated real estate side of things. And I'm looking for use cases, you know, all across each of those companies, you know, just try to figure out where, you know, what the best application of these tools is in a business context. And so I've been trying to, I've been fortunate. We've had a couple of people at Dominion who have really grabbed the ball and run with it. And that has inspired me to try to create an environment that's getting more people. to be inspired to use these tools and just let their brains run. So I'm coming from the problem a little bit more from an organizational perspective, but I'm trying to also keep up personally. I got my open clause set up. I haven't done anything cool with it yet, but I spent a couple hours getting that set up and it's fun. I'm enjoying it also. And I got a bunch of business use cases. that I'm like, I just write them down, know, like I got a list of like 20 little projects that I'm like, we could use it here, we could use it here, we could use it here. So, and in some creative ways that, because I just made it so that some problems that were impractical to solve previously are now totally fine. So, know, going through super large databases and, you know, creating complex automated workflows. So I kind of get off on those ideas, which is why I enjoy. talking to these gentlemen so much and I'm excited to be part of this podcast. Craig (08:19) So if you all are watching right now, you may notice that we're all a bit bleary eyed. I was up and I've been up until three or four o'clock in the morning for at least a month now. So there's a technology folks that is kind of sweeping the world right now, literally the world. And it's it's called open claw and Claude code. Now these are two, two, two separate things. And this is not going to be a how to podcast guys. David Moses (08:24) Thank Craig (08:44) But what we will do is we'll share a lot of ideas and let folks sort of learn as we share. so getting back to OpenClaw and Cloud Code, these two tools are absolutely revolutionary in that I believe that we are now seeing the promise of agentic AI. So in 2025, it was the buzzword, you know, we're all. there will be a million agents and they'll all be off working, doing their thing. And, and I think that promise was largely undelivered until these two tools came out. look, look, if you wanted to get in and get under the hood and tinker, you could absolutely build things that were really cool in 2025. And I did, and I'm sure you guys did too. But I think today what we're seeing with OpenClaw and Claude code, is we're seeing a much easier environment to create these very powerful skills or agents, if you will, that can go off and do work while you're sleeping. And so one of the reasons why I think we're also bleary eyed is that you spend a lot of time sort of setting it all up and giving it ideas to go off and program for you. And so I guess the point of me bringing all of that up was Jack and David, I think we're finally in a time now. where these tools are going mainstream and the skills or agents that we are building are just a lot easier to build and implement. And to everyone who's listening, we'll get way more into it. But what I would, I would encourage anyone who's listening is to just stop doom scrolling at night. This was, this was the mandate from Jack Bevier about a year ago, Dave. We were on a call with 200 people in the company and Jack said, look, If you just spend an hour a day, just an hour a day, not doom scrolling and learning about these tools, you'll be surprised at how far along you'll come in a very short period of time. And I'll be damned if he wasn't right again. And so that's what I would encourage all the listeners to do. Just stop doom scrolling. Get on, just watch one video at night. And to that, I've created a newsletter that we'll be releasing some point very soon. David Moses (10:40) . Craig (10:55) that will be, it'll come out three times a week. Folks can opt into it and it will have all of the trending videos, news, accounts, anything that people should be looking at if they're trying to learn about these tools. And so we'll have that link coming shortly for the listeners. And so you can opt in and check out the free newsletter. It's really cool. Guys, you got to see it. It's so cool. I feel two interfaces to it that would be like mind-blowing. so, yeah, so that's sort of the whole thesis of this thing, we could talk about maybe just some of the things that we've been building and that, you know, sort of the stuff that like the progression of the tools, if you'd like, you know, starting off with a year ago, I was trying to program an N8N, which was a great tool to use. But yeah, I mean, we can kick off like, where did it start for you, David? And then sort of what does it progress to today? of using these tools. David Moses (11:49) So, yeah, so I downloaded Chat GPT, actually it just before, it was like three, right? So was like the app, it was three. So I wasn't like on one or two, whatever. And I went to, was like the next day I went to, you know, met up with everybody in, maybe it was Baltimore again. And I started playing with it and it seemed, Jack BeVier (12:12) Thank David Moses (12:13) kind of unreal. ⁓ You know, you could just ask a question and it would just kind of give you an answer. And I learned about kind of the technology that was behind it. And I said, okay, this is going to change everything. And it's going to change everything very, very, very quickly. And I think it is. I think, I mean, we're already starting. Jack BeVier (12:14) I did. and peace. David Moses (12:39) For bad and for good, I mean, I think there's a lot of people who are scared and I think there's probably a lot of validity to those feelings. But there's also a lot of people who super excited like me who are like, hey, you know, now there are no constraints from a thought perspective. Like, I no longer need an expert in anything to do anything that I want to do. That's really what it is. It's not needing... not needing to rely on heavily on a programmer, an attorney, the worst would be the software support people themselves. It really is just, I can solve a problem in most of the software we use faster than tier whatever tech support can do it. And I just, I wound up just sending emails saying, this is what's wrong, this is what's happening. Craig (13:34) Dave, how have you been surprised at sort of how the tools have evolved, you know, seemingly at times overnight? You know, like I'll give a command to do something and on, let's say on a Monday, and then I try the same thing again on next Monday. And I have no idea that the tools got better overnight. And I one shot something that like blows my mind. with just words, right? So like that's the crazy thing here. None of us are programmers. I've never written a line of code in my life. In fact, my dad, Jack, I think I've told this story before. My dad wanted me to get into computers. He's a pipe fitter at General Motors. And he's like, computers, boy, this was 1985. And here I'm like a singer in a band and I'm, I'm, got big hair and I'm not, I'm not a tech guy. And I was like, yes, dad, that's exactly what I'll do. I'll go to learn how to program computers. It's still green screens back then. Jack BeVier (14:00) I'm not. ⁓ Craig (14:28) I couldn't do it to save my life. I couldn't learn Visual Basic. I couldn't learn C++, Java, none of that crap. I was so lost. But the beauty that Sam Altman says is we are now living in the age of the idea guy. Because programming, the programming part is no longer the hurdle. Like these things program better than the best programmers. And with a little bit of structure that you learn from using the tools, you can get along a lot faster than just by iterating to get it to where you want to go. So, Jack, what do you have to say with regards to sort of how the tools have evolved? And then Dave, I want you to jump in. Jack BeVier (15:06) Yeah, so, the the thing that the barrier to entry, it's intimidating, right? Like it seems intimidating. It's a bunch of new vocabulary. People are talking, you know, a lot of people are talking about this. Not as many people have really dove dove didn't, you know, and to use them and figure out how to like make them productive. And the the thing that strikes me about it is that anything that you don't know how to do, you just ask it if you don't even know if you're like, what? do I prompt this thing? Literally just ask it right like whatever thought comes into your head. Just put it into the prompt and press enter and you this the feeling of frustration is actually like. It just kind of dissipates because wherever you get stuck you just tell it hey here's exactly where I'm stuck and here's the screenshot of here's what I'm seeing on the screen right now and it it steers you towards the right direction and so if you just like keep trying you have a problem and you just keep. Aiming towards the solution and you just keep aiming towards that solution. You'll get there eventually and there's no dead ends and that idea to me is just like I don't know. I'm like infatuated with that idea and so people were like well, you know is there and I and a year ago I was just like hey Leo. Here's this. Here's this newsletter to follow and here's these courses that you know that were being advertised where you could you know they'd walk you through all the basics and stuff and that's like. kind of all well and good, also totally unnecessary. Like you really can just open up the screen and pick whichever LLM you want. It kind of doesn't matter. It doesn't really matter. know, like people start nerding about like, I like this one versus that one. It doesn't matter. Like pick one. And you can just start playing with it and end up down whatever rabbit hole you want to. And so... Just that the pure using it is like the best advice for somebody who who isn't who's just getting started in it. ⁓ But I am. Craig (16:58) I want to give a chat. I was wondering if please go ahead and then I'll jump. Jack BeVier (17:02) Well, I am curious because everyone's got their little favorites and so people listening also, you know, guys are talking about this stuff, but what are you doing exactly? And so I don't think the three of us are even playing with the exact same tech stack. So what you guys, which tool are you guys using or which? David Moses (17:20) Do you guys hear that Craig (17:22) Yeah, I do. David Moses (17:22) feedback thing happening? don't know if that's from Jack or... Jack BeVier (17:26) Maybe it was like so which tools you guys are you guys using right now to to build stuff? Craig (17:32) Dave, go ahead. David Moses (17:34) So I'm using, N8N is, that's my, hey, I want this thing to work in a business case. Everything else is just built around how do I build an N8N and how do I make a decision on what to build and how to build it. But what. Jack BeVier (17:49) And it Craig (17:49) So let's. Jack BeVier (17:50) is an orchestration tool so you can create like business flows like any kind of business logic you can program in with triggers and business logic and you can put an LLM in the loop if you want to. You can hook up APIs there. What else am I missing? You know you use it every day. Craig (18:06) Yes, go ahead. David Moses (18:09) Yeah, no, mean that's essentially what it is. It's really just simple JSON, which is really just, not to be too geeky, it's really just key value pairs. It's just like, here's a field and here's a value. And then it just organizes them and nests them in a certain way. So that's really all N8n is, but it visualizes it for you. So you can actually see it function. You can see it go through the different processes. So it's like, hey, you need to... ask this person for this thing or you need to go grab this document. You can see it grabbing the document and see what it pulls out when it looks at it. And then, you know, that's essentially what N8N is. Craig (18:45) Yeah, I would encourage anyone to just just do a little little bit of a dive on YouTube about N8N. You'll find out everything you need. Dave, are you using N8N inside of their environment or inside of cloud code or cursor? So tell us how tell us how you're programming N8N. David Moses (19:04) So, I don't know what, like today is March 12th and it's 1046 AM. it's just 2026. Just as a caveat, today, that's right. Today, what I'm doing is I have a team of agents inside OpenClaw that have API access to my N8n. Craig (19:12) Right now I'm using. It's not 2027. David Moses (19:33) And they build the workflows based on, so I've given each agent a persona, one's like the architect, one's the builder, one's the validator, one's the tester, and then they go ahead and they build the workflow after we conceptualize and decide, here's what we're gonna use for this, here's what we're gonna use for that, and then it just goes and builds it. It evolved from literally choose this node and connect it this way to After that, was using Claude. I connected Claude via the N8n MCP server. MCP server is basically just a fancy way of saying, you know, this is how two systems talk to each other in the agent world versus, you know, so that's essentially what it was doing. It would use the N8n MCP server, which has all of the, let's call it the ingredients for how to make an N8n workflow. and it would just go ahead and build it. was doing that, that was last week. All right, then the week before that, I was literally just talking about what I wanted into an LLM and I would have literally four LLMs up. I'd have GROK, Gemini, Chat, and Claude, and I was using Sonnet then. And I would just, literally I would say the same thing and copy paste it to all of them. And then it would just give me, here's what I think the workflow should look like. And then I would copy. Craig (20:32) Yeah. David Moses (20:54) each other, know, each one's responses to each other and say, tell me how this is, you know, not going to work. You know, that's how I was building it. That's how I was building it. You know, whatever it was like a month and a half, two months ago, you know, and all the way up until then. was basically what I was doing. It's just asking an LLM build this. It would build the node. would mess it up over time. It would learn. Um, and, uh, and now it's just like, I just say what I want and it Craig (21:00) That's badass. I love that. David Moses (21:23) You know, it one shots damn near all of it. And what I'm changing usually isn't architecture things, it's functionality. I don't want it to do this, I don't want it to do that. I need to add this little tweak in or this gate or this guard. know, like here's what it's doing and fix it. Jack BeVier (21:40) So are you putting that is what are you putting? I gotta ask a couple details. So what interface are you using though? Are you like you have an instance of? You're in discord. Craig (21:40) You know, it's funny. David Moses (21:47) Discord. Craig (21:49) ⁓ very nice. David Moses (21:51) Yeah, I'm using Discord because, so I like Telegram, but the context window gets eaten up very quickly because it's all kind of, you have to continue to add new sessions. If you use Discord, you can have a thread for each thing that you're working on, and it doesn't inundate the context window with a bunch of tokens. ⁓ Jack BeVier (22:10) So you're talking about your open claw here. Okay, so you're using open claw in your business. I'm curious about your perspective on security. So you're using Discord as the way to put voice instructions into, or sorry, thread of instructions rather, and then telling your open claw agent, David Moses (22:23) Yes. Yeah, whether it's voice or yeah, it doesn't matter. You can put anything. Jack BeVier (22:38) what you want to go build. then you've, sounds like you've given that open claw agent access to some of your business processes. What have you done? David Moses (22:48) So the way I've done it in terms of the APIs is I actually don't keep any of the API keys or credentials in OpenClaw. What I'll do is there will be a workflow. And that workflow will basically just be a webhook, an HTTP node, which is just a node that can make the API call to whatever API it is, and then another webhook. and I've got one for each API. So it'll have the credentials built in, so the credentials are still in N8n. But using the webhook, the agent can essentially send an API call to the API, to whatever software it is, and get a response. And I have it hard-coded with a get so that it can't write. And then there's another one that can write, but it just asks me if I'm writing something to a database. Jack BeVier (23:29) Thank David Moses (23:39) through where OpenClaws involved, asks, it'll actually literally send me a code and I'll have to give it a code in order for it to actually make the post or put or anything that writes to the database. So I basically am giving it the ability to read everything and most APIs, all of our APIs, I'm sure most everybody else's APIs block very sensitive data like social security numbers. You can't even get that stuff out of the API in like AppFolio or in... some of the other software that we use, so I'm not worried about it. If somebody were to grab it, I don't think they're gonna grab everybody's sensitive information. And I don't think that I should be the first target for somebody who wants to Social Security numbers, considering I manage low income housing. But it's not exactly the Social Security numbers that you would consider valuable anyway. But anyway, you can't get them. So. Jack BeVier (24:23) So you basically you've by you've hidden the API key inside of the end workflow, but you've given the open claw access to trigger that that that API call and essentially you've created read only access, right? Like that's the net effect is that it doesn't have the API key, but it does have read only access to everything. David Moses (24:35) Mm-hmm. Yes, and it can do API calls, pull the data, sift through it, organize it. I also mirror everything into Mongo. I mean, guess technically it could write to Mongo, like all Mongo is is a mirror of the actual databases. So if it wrote a bunch of crap to Mongo that wasn't right, it wouldn't matter because it would all get rewritten anyway. Jack BeVier (25:04) So walk me through, sorry Craig, walk me through why you are mirroring everything into Mongo. Cause this is something like, this is something I'm trying to figure out right now. Like I've got, I've got Appfolio, I got Buildertrend, got Podio and I got QuickBooks, right? Like, they don't, Buildertrend does not have an API. No, Buildertrend does not have an API. Yeah. So I'm like, what the hell am I going to do? Like I'm trying to figure out what to do, how to figure this out, right? David Moses (25:21) Builder trying to have an API. OK. Continue. The software formerly known as Builder Trends is soon. Craig (25:33) Hahaha Jack BeVier (25:33) I'm saying right, I'm saying like. But so so I'm trying to figure out how to connect these two, because I'm really we're literally still doing like manual entry across multiple databases here, right? Like a tenant moves out and the turnover happens. I need to tell construction about it and they get someone goes and then they type, you know, create new job and then look over the portfolio and they type the stuff that they need to into builder trend. It's like not great. and so I'm trying to figure, but I can't connect those two directly to each other. So I'm like, do I need to pull stuff out and create a database that, that houses all of my business's data? And then these softwares are basically just like application layers that like user interfaces for, for my people, because I'm not going to reinvent property management software. How are you thinking about that? David Moses (26:18) That's, I mean, that is the question and I don't have the answer, but I can tell you the way we're going about it and the reason for Mongo is to make sure that I have all of my data in a way that, know, Mongo is just very flexible with how you ask it things, you know, and if you index things properly, it's very, very quick to read and it's lightning fast to write to. ⁓ So it's a good database, I think, to use. Jack BeVier (26:40) Thank David Moses (26:44) But there are many, many, many others out there that are great. SuperBase is great. Postgres is great. But anyway, so with Mongo, and I actually have Mongo hosting it. I don't self-host Mongo, but you can. And essentially, that way I have the data and I can decide, a lot of what I'll do is I'll ask my agent, Hey, look through this collection in Mongo, right? And tell me how the data is organized, what I can get out of there and what I can't get out of there. And it will use the structure of Mongo, of the data in Mongo to actually decide and help me figure out how the flows should work. But the idea is still the same. It's, have to do this thing and you're trying to find, this is the trigger. Jack BeVier (27:18) This is David Moses (27:34) This is how I know I have to do this thing. This is what I have to do. And this is how I prove that it was done. This is the output. And that's really what all these workflows are. It's just something has to start it. It has to do some work. And then it has to give you an output to show you, yes, I did this. And here's what I did. So that's the way I'm thinking about it. But I don't have a, this is what the world's going to look like in six months or in a year in property management because That's what we're trying to crack our heads open on every day. Jack BeVier (28:08) Alright, so stupid logistical question. How did you get your data from Appfolio into Mongo? How did you get your what other what other databases are using? David Moses (28:16) So, we're using Sync, C-I-N-C. I'm not gonna recommend any of software. I've recommended software to people before and it always bites me in the ass. none of these are recommendations. Nothing. ⁓ So, I will say... Jack BeVier (28:18) or sorry, software is confusing. Craig (28:31) It's like recommending a contractor, Jack. just, hey, use them at your own will, right? Jack BeVier (28:31) Thank Thank David Moses (28:41) So there's two questions, number one is what we're using, app folio and sync, the INC, and then property meld are the main CRMs that we use for property management. And how we get the data in and out is essentially, had all of those APIs ingested into a Mongo collection. And basically all it is is a list of endpoints. Jack BeVier (28:54) the video. David Moses (29:04) Here's the endpoint, here's the data you can get out of that endpoint, here's how you make a call to that endpoint, here are the required fields, the not required fields, and then I also add like here's a sample what an N8N node looks like ⁓ in JSON. So basically just kind of like a library so the LLM can look at it and say, this is how I find this data, this is where I find it. And after making that library, I said, okay, I want you to go into every endpoint. Jack BeVier (29:18) Okay. David Moses (29:30) and I want you to have a collection in Mongo that mirrors that endpoint. it'll really go through and workflows when I originally did this, it was literally a flow for every single endpoint and it looked like a bunch of spaghetti nonsense. Now I've learned and it's basically just, you know, it's basically just one HTTP node and it just cycles through each endpoint and pulls everything out and then updates, you know, uses whatever key to update it in Mongo. and whatever's new goes into Mongo. And some of them go every minute, some of them go every half hour, some of them go every day. Jack BeVier (30:05) man, okay, I got you. All right, so then you've got so then you basically at that point, you've got all of the different systems data into one Mongo database and you core and then your your run and flows on that stuff. The stuff that's in Mongo on the data that's in Mongo. David Moses (30:22) Anything I want to analyze, I go to Mongo. So if I have a workflow that's like, have, meld is the best one I have so far, but it's basically just people upload pictures, they upload videos, they chat in the little chat window, they put in closing notes, they give us invoices. All this stuff happens in property meld. but it's all everywhere and it's impossible to get people to look everywhere or be able to look everywhere. So because it's all going into Mongo, I have agentic flows that basically they just, here's all the stuff in Mongo for this record, analyze it, tell me what looks weird, right? Tell me what should be, you know, tell me does this make sense, right? Somebody spent three and a half hours doing this, that makes no sense, that should only take an hour. Jack BeVier (31:07) minutes. David Moses (31:10) You know, and then somebody's making a phone call or there's a text message that goes out that says, Hey, you didn't, you know, we trust our guys. We know they're not spending three and a half hours, you know, farting around. So it's really, what did you do that you didn't put in here? Cause we need to know. And then the, yeah, no, I did this and I did that. And those things get enhanced. And so by the time the customer sees it or by the time the owner sees it or by the time I see it, I have a full context of why, you know, why this thing was the way it was. you Jack BeVier (31:41) So did you create a workflow that just said like, go look at all the data that's in Mongo and like and me if anything looks weird, like what'd you write? David Moses (31:51) That's, I mean, that's, so first it's getting the data in. And so, because there's a lot of pictures and videos, I have one workflow that all it does is every time a picture or a video is added to a work order, it will look at that picture and it processes it through and says, tell me what this is a picture of, tell me anything that looks strange. And I give it all of the context of the workflow. So I basically say, of the work order. So it says like, what the issue they called in was. and all the chat history between the technician and the manager and the tenant. And here's all of the work orders for the same unit over the past forever. And it basically takes all that context and uses it to analyze one picture. It's really actually not expensive because it doesn't use, it uses a pretty cheap model. It just dumps a bunch of data in and it sifts through it and it says, okay, well, I know, you know, this is not just a picture of a front of a house. Jack BeVier (32:26) Thank Thank David Moses (32:49) picture of the front of a house and the work order said that there was a gutter that was sagging, you know, and now I can see from the picture that the fascia is rotted out, you know, and this is not just going to be somebody going in and screwing in a gutter. I know I've got more work to do. So it analyzes each picture separately, using the same context in every single one. Video, same thing. And I use Gemini, anyone wants to know, I use Gemini to analyze the photos and videos and it gives basically a narrative of each photo. And it's pretty extensive. I don't ask it to, you know, truncate it or make it too concise, because I'm not reading these. It's just, it's just basically just becomes the context for the next part of the flow, which is after all the photos or videos are analyzed and we have a description of each one, then it goes through and it analyzes all of them together. And it gives a kind of like a full, okay, now I know everything about it. Now I know, now I have all the pictures. And then that's where it decides what's weird. Was the work done the way it should be? Did the amount of time that was taken make sense for the work that was done? You know, that kind of thing. Craig (33:53) You got that working? David Moses (33:55) Yeah, that works. That actually, it was a pain in, because I built that all the old way where you basically just tell the, have much you want and it builds the node. I'm sure if I built it today, would look a lot smaller, but it's a beastly workflow. It is. Craig (33:59) God is bad at this. Yeah, I just I just want to level Jack. just want to level set here and just tell folks that are listening that like the promise of these tools now is that anything and I and I do mean anything that can happen on your computer. These tools can do very, very quickly. You just have to give it the right commands. And and so, you know, perfect example of that would be in my case, something. A little more mundane than Dave's last use case there, Jack, is that, you know, it has full access. there's a Mac Mini that sits on my desk here at 32 South Street. I've given my bot, Rocky, autonomy, complete autonomy, Jack, to just go out and figure it out, do whatever you got to do. And so yesterday it called me, they called me during the day and we had a discussion back and forth with one of So again, any function, anything that you can do on your computer, these tools can now do if you just give it the right commands. so, yeah. Jack BeVier (35:04) Thanks. David Moses (35:12) It is, it's really amazing. That is really absolutely... And how did you... What was the impetus for you to have it do that? Jack BeVier (35:14) I'm How do you like? Craig (35:24) So my idea is that I think that one, want to be 10x more efficient at what I do when I sit in my loan officer chair every day. And so I'm trying to develop tools that allow me to do that from email follow up to recording each call and summarizing each call. Jack BeVier (35:34) Thank Craig (35:46) And a lot of that's pretty easy lifting if you know what you're doing. But then sort of how are these tools evolving in such a way that we can present front facing solutions to our clients that enhance the customer experience? you know, how can we make an app that takes five minutes to fill out rather than a half hour to fill out? How can we do follow up with them that feels like it's dynamic, not like high? Jack BeVier (35:48) All right. you later. Craig (36:13) I sent you a quote. It's, John, I sent you a quote for one, two, one, two maple. And here's the quote in a very beautiful fashion. So all of these tools I'm trying like when the hits the fan, Dave, and the market turns as it certainly will, you're going to want to be 10 times more efficient than you are right now so that you can just take on more business. And that's the tools that I think I'm building and, and sort of with always with the customer experience in mind. So Jack BeVier (36:15) Thank Craig (36:40) to the reason why I wanted my bot to call me was it's pretty rough right now. Like when it calls you, it knows what it's talking about, but it's slow because it's reading from a database to get its memory and all that stuff. So it's a little slow, but I can't imagine what it would be like six months from now when I have my bot call up a client and say, hey, I sent you that address yesterday, that quote, and it's in a perfect voice. Jack BeVier (36:55) I think we're not going to be working like a country. And I have my own wealth of private space. I I should expect that after I get to next floor. And so, we're going to be working a group of I'm going be working with a group of people. I think we're going working really hard to get that through. I'm not going be working in that in the next month. So, it's time. I'm to be working on it. Craig (37:08) It's in a great it sounds just like me. It's conversational. And I think we're very, close to having that sort of dynamic experience with ⁓ with a client that frankly they would find just as bad. Look, I think, Jack, that we're living in a time where five years ago, 10 years ago, if you said, hey, Craig, go through the self-checkout line at the giant at the food store. I said, no, I pay my money so that somebody can check my my out and put it in the bag for me now. We're also desensitized to it that we just do it because that's what's available. I think that these tools where people are like, I don't want to get called by some AI, soon you won't even know it's AI. And so that's sort of the envelope that I've been trying to push with the stuff that I've David Moses (37:52) So what do you think the gap is? Like you said, it's almost there. If it's six months off, like what is missing in order to get it there? it, know, because I have my own kind of like skeptical or kind of pessimistic thoughts about that, but I just, I'm curious to know where I'm wrong. Jack BeVier (37:59) Right. Craig (38:12) You know, my thought as I was really going through the clujiness of that two-way call yesterday with my bot, Rocky, was the same. Like, how do I make it so it's faster, so it seems like a duplex conversation between a couple of people, right? Or at least between a human and an AI. And I think that it's a matter of bandwidth, frankly, tokens and bandwidth, really, and processing power. So I'm using a Mac mini on my desk that has 16 gigs of RAM and a 256-gig bar drive and I'm running the world with it. I can't imagine what it would be if I had a couple Mac studios and you were working from that. Would it be faster? Would it be better? I don't know because so much of my memory, of my bot's memory lives on a super base somewhere and you have to pull a lot of that into context as you're doing that sort of duplex conversation. David Moses (39:05) It's sort of like you and I have a conversation and we may be having a conversation about a specific thing. Like let's say I'm a customer and I'm borrowing on a property. You don't know everything about my property off the top of your head. I don't know everything about my property off the of my head, but we still have a conversation that feels natural. And I feel like part of it is the technology just getting to the point where it understands, say, hey, I'm going and getting this information. Give me a sec. Let me look this up. Because that's what I would say. If you asked me a question, I'd know the answer to. And I had a something in front of me that would allow me to find that answer, I would say, hang on a sec, I'm gonna find it. But the AI wants to, it just seems like it always wants to just go get that information instantly and doesn't wanna tell you that it needs to go do that. it's like, that's part of it, that's one side of it. It does. Craig (39:48) tell you what it's doing. Jack BeVier (39:52) Dude, that sounds like so solvable, right? Like, I mean, you're right. You're totally right. Like, let me think about that for a quick second. I say on many like on, you know, 20 times a day, like, yeah, let me let me just go think. And then we all just like allow the human to, you know, use some RAM and then bring it to the forefront of its brain. And then the rest of the conversation, the latency has disappeared because I thought about that for a second. God. Craig (40:17) to you. David Moses (40:17) And I just, I haven't toyed with it that much, that's kind of the frustration. But the other side of it is the context isn't there. Like I haven't built, know, like we're just, now we have phone calls, right? Like as of this week, we have every single phone call that is, you know, a workflow that goes and finds out everything they know about this person. And every, which is easy because they're all known people, you know, generally. Like you guys are dealing with potentially new customers. So it's a different thing. But everything it knows about this person, every work order for their unit, whatever it is, it knows all this stuff. But it goes and finds that information and then summarizes the call with the magic of all of that context. So it actually creates a note that really is awesome. And so we're building that context and then later on it uses that context to, while you're in that phone call, I'm in the phone call with this person. everything I know about everything is going to take forever to pull, right? But if I, second that person calls, we use Dialpad, there's a web hook that hits. And I know who's calling before anyone even answers the phone. It shouldn't take more than a couple of seconds to pull everything I know about that person, right, into the context window of the bot that you're talking to. So really should be able to have near Craig (41:14) Yes. Thanks. Jack BeVier (41:33) Yeah. David Moses (41:36) perfect recall about whatever it is you're asking about. You and I are, you know... Craig (41:40) But what if the question that I would have to that is, but what if you're having a conversation with your bot and it's more global about the business than it is about a single customer? I think in that case for me, I would probably build a report so I could speak specifically about some sort of report that it would pull into context rather than sort of trying to do an instant vector search across my entire database of you know, whatever it is and I'm collecting. Right. And that that might be a way to definitely speed up conversation that you would have with your body is just know the types of conversation you're not going to get on with your body and just and just chit chat. You're probably getting on a phone call with your body because it's an emergency. Your bots run into some sort of issue that it can't handle itself or whatever reason. Or you're looking for that sort of instant data on a part of your business that's critical. And I think that that will be an interesting way to program a bot to sort of have those reports ready to go in context as you sort of commanded it to bring that up into context. David Moses (42:48) Well, how would you do with a human, right? Like with a human, you wouldn't just say, hey, come into my office, we're gonna sit down and we're gonna talk about blah, blah, blah, blah, right? Like they're not prepared, they have nothing for you. But if you said, hey, I'm gonna schedule a meeting for Thursday at 11 a.m. and here's what I want you to be prepared to talk about, then maybe you're giving the agent a fighting chance to go out and get the context it needs. to actually have a conversation with you that's intelligent. But we think of it as like a magic box that we just talk to and we want the answers instantly because isn't that how it's trained us so far? Because we ask about things that are stupidly detailed like about space travel or about black holes or about, and it knows everything about everything because it's all. built into it, but what it doesn't know is it doesn't know our businesses. doesn't know that John is 30 days late and alone. It just doesn't know that because it's not part of its training. you kind of have to treat it more like a person than you treat it like the chat bot that we've all used for however many years. Craig (43:53) What the thing that I really ⁓ love that I'm seeing that I'm diving a lot more into now is there, are just a lot of great minds out there that are playing with these tools and they're making their products open source on GitHub, on open claw hub. And while you have to be careful with some of those things to downloading them and just using them, if you just do a search, do it, do a deep search. Jack BeVier (43:55) Thank you. . David Moses (44:17) You should say that like twice. Be very careful before you start downloading these things and using them. Craig (44:24) sort of elementary use cases that folks can start playing with tools in a second. But I think one of the things that I was doing last night was, hey, man, I'm not an expert in this stuff, but go do some deep research on people that are using these tools for memory. How are they loading the context? What's getting them to the finish line? And it comes up with this great 10 page report. Jack BeVier (44:31) Thank you. Craig (44:45) that now you can load into context and say, OK, use this report to develop me the greatest memory system for my for my bot ever. And then just go from there. Right. Like you're just iterating off of that at that point. But it's a good start if you if you just don't know what the better minds that are out there using these tools to the fullest are doing like right. I love being on the cutting edge of the stuff. Jack BeVier (44:51) never. from the other side. ⁓ Thank Craig (45:06) Yeah, Jack, anything? David Moses (45:06) Yeah. Jack BeVier (45:08) My head's freaking spinning guys. like I'm like I have so much work to do Craig (45:10) I love There's nothing better Dave when Jack sits back in his chair and he kind of gives that look like that puts his head back a little bit and you just know like something big's coming something big's coming. So guys I have this I so two things I want to share with listeners. If you if you're just thinking like hey man I want to get started when I dip my foot in the pool. think Google right now guys and you can jump in. think Google is really putting out like a lot of great tools that sort of live. all in different places. I don't know why they can't get everything under the hood of like all the great stuff that they're building. And a lot of it, believe it or not, is free. You can bang on these things pretty good before you hit the paywall and they want you to put up a credit card. And so one of the things that I would ask folks to do is look into Google AI Studio. Google AI Studio, if you just sign up for it, you know, when you go into Google guys and you hit just regular Google search, I feel lucky. Jack BeVier (46:01) Yeah, I'll speed you up. Craig (46:08) And then you hit that and it brings up whatever it can find for you in a regular Google search. Google AI Studio does the same thing. You can say, feel lucky. And it'll bring up like a use case that you can sit then and just hit go, watch it do it, watch it make your program for you. And then you can iterate with it. You can say, I like this. I don't like that. And this is how you start playing with the tools in a very, very palpable way where like you can design something literally in a half hour. Jack BeVier (46:29) I'll take this. Okay, go ahead. Craig (46:37) Google AI studio, make an account, go in there and start playing with it. And Jack, I had dinner with a guy, a young kid about a week and a half ago. And he works for a very large self storage company. And the kid never played with finance major, got straight A's in school, never played with AI in his life. And so we sat for about two hours having some beers. And I said, look, go home and play with Google AI studio. Jack BeVier (46:52) Thank you. We'll back for about two hours. Craig (47:02) So basically he spends his entire day looking on Loopnet, Crexie, Google for development opportunities for this company. So he's just basically looking at little plots of land all day long to see if it works. He's developed two tools that are going to save him hours and hours of time, literally in one week of just using these agentic tools. And so that's the power of just jumping in and trying it out. Jack BeVier (47:05) Thank you. Thank Craig (47:29) One of the great things that I can recommend for anybody listening right now who wants to jump in is Google AI Studio. Jack BeVier (47:35) The, I was having this conversation yesterday with a guy at a bank actually, and we were talking about the underwriting. Craig (47:41) just waiting in line or he was the president. David Moses (47:43) Hahaha ⁓ Jack BeVier (47:44) Nice, the tech guy. He's the he's the tech guy and and ⁓ David Moses (47:48) Of course that's the guy Jack is talking to. Craig (47:50) I don't need the guy with keys to the vault. I just need the guy that runs your computers. Jack BeVier (47:55) Yeah, yeah. Well, they there, there's mortgage division and they're trying to figure out how to grow their mortgage division and they want to use AI and there's like, you know, some tools out there that, you know, they call themselves mortgage AI and we've demoed a lot of them. Most of what we've seen is really just been like OCR. It's just been like, you know, pulling, pulling the data off of the documents and then sending it for, you know, not actual AI analysis, but rather just like hard coded underwriting, like coded. underwriting guidelines. And, and we're talking about what we've built, which is a purely AI driven underwriting system, and the way we had to architect it to get over the, the hallucination issue, which you just can't have, you know, you just can't have it. So, but we think that's like totally overcomeable. And anyway, my point was that he's trying to they're trying to figure out strategically from the bank's perspective, like, hey, how do we go tackle this AI thing? you know, do we build it ourselves or, um, or is the, you know, the some product going to come onto the market and hide this like very strongly held opinion that like, like no one is going to come in with an AI product. No one from Silicon Valley is going to show up with something that is built for Podio and Buildertrend and Abfolio and QuickBooks and by, you know, and, understands all the workflows that I have in Baltimore. And unless someone comes up with that, right, like a mirror, you know, a, some somehow I have the most generic business ever that they can that they create a product that works for that generic ass business. I think it ain't gonna happen. So like but you can but but but you know but with a little bit of learnings we're hey we're gonna have to roll our sleeves up a little bit. You can pull all the data into a centralized place like you were talking about David and then start running flows based off of you as the technical guy right like the as the not not the tech side you as the technical expert in terms of like Craig (49:27) Yeah. Jack BeVier (49:46) I know how to run an HOA management company. I know how to flip houses. I know how to make, you know, hard money loans in this particular niche that I think is the sweet spot from a risk adjusted return perspective. Nobody's going to build your like literally this is the opportunity to build your secret sauce into the workflows and no one else knows that your sauce, you know, no one else knows the recipe, only you and but you can then infinitely scale your secret sauce, right? And I'm like, I just, these tools are gonna get built from the inside. They're gonna get built from the practitioners. Silicon Valley is not gonna have a bolt on for us. Like, I just feel like it's a new tool that we, a new set of tools that we all just, as entrepreneurs, need to roll up our sleeves. Like, learn how to add value in society, super hard, right? And then translate that. that formula, that secret sauce for making money into your tools and you'll be able to just run. So I'm just like infatuated with that idea. And I feel very strongly that that's the case, that Silicon Valley is not gonna come save us. You're not gonna be able to just wait and bolt it on later and pay for it and bolt it on later. I just think everyone who's waiting right now is wrong. Like they're just doing it wrong. Craig (51:03) Yeah, not only that, Jack, but I think the more you wait, the more you get run over by the tools. You have to be learning sort of from the ground up right now to be able to take advantage of the arbitrage. It's going to happen between those who understand the tools and those who don't. And just just just getting your hands a little dirty with it now, I think is going to pay off in spades six months, 12 months, 18 months from now when all of your competitors are behind the eight ball because they know they have to learn it and they're going to be behind the eight And some would say, but Craig, you're wrong because the tools in 18 months from now will be so much easier to learn. And so those people will just jump on board. And I disagree. think you have to get your, you really have to start getting your feet. David Moses (51:41) Yeah, most of the last year and a half for all of us has been, you know, we don't know what we don't know. And yeah, there are things that we are taking for granted today because we're just saying, hey, I want to build this and it's building it and it's building it reliably where we had to fight through that, you know, a week ago, a month ago, you know. Craig (52:04) Dave, like to say that Jack BeVier (52:06) Thank Craig (52:06) OpenClaude, six months from now, is gonna look like MS-DOS. know, like, we think it's like the second coming, and six months, six, eight months from now, it'll look completely different. David Moses (52:10) Hahaha. All they really did was take a bunch of technology that already existed and they just took the guardrails off. Right? Jack BeVier (52:23) Yep, so I'm to go. I'm going to go back to my to my IT team and say hey, I want to. I want to hook up open claw. While we're going through a SOC 2 compliance audit and see if I can, if I can get open claw into a SOC 2 compliant, make it through a SOC 2 audit will have something right like that will be. David Moses (52:34) Ha ha ha ha ha ha ha ha ha ha ha ha ha Craig (52:43) you Jack BeVier (52:44) That'll be the formula to like, this is how you get these agents into your system is passing a SOC 2 audit. So I'm literally going through that right now. That'll be my weekly recurring update is how have I married OpenClaw with SOC 2 compliance. David Moses (52:55) my god, that sounds like fun. you Craig (53:01) I have this great idea for what I'd love to do each week. I would love if each one of us came with like a use case that we implemented that we could show off to the listeners. so the one that I did. So my bot. So the whole idea of having this this this system of virtual assistants is how do I get them to work 24 7. How do I get these damn things to do when I'm sleeping? And I think I've, you know, I've got it. It's, it's, you know, I'm up late at night making it do that. But last night I reprogrammed the entire thing. There's three files that live inside of open call that are really important. It's agents MD that kind of like tells your agent, your main agent, like how it should act. And then there's identity kind of like, you know, what's it like? And then, and then it has a soul. And you can look at about those three files, but I, after a month, Jack of using my bot Rocky to what I thought was its fullest. re I rejiggered the entire thing last night, literally rewrote all of those files and started over again and really, and I said, and I want to be surprised when I wake up in the morning, I want you to do something that's going to surprise me. And so this is the part of the idea. I want to come to the podcast every week with some mind blowing thing that our bot did. that's going to, that we can show the users that they'll get inspired by. So mine last night, Jack, I woke up this morning and mine created a magazine where it told the story of exactly what we did last night. And I showed you guys prior to the podcast and I think you'll find it was pretty impressive. The things got its own persona. It has a look, looks like this old dude with a beard and it put itself in a magazine, like a literally an art, an article where it told its story of Jack BeVier (54:39) your breath at the same time. here. Craig (54:50) Everything that we created last night and I woke up to that this morning and I about my pants because it was so beautiful. Like I could sell it. And so that's going to be the thing that I show off to the listeners. Again, they'll be like an opt in, just, you know, throwing your email and you get access to all that cool stuff that we're working on. So I got the newsletter and I got the magazine. I think are going to blow people's minds that as to what you can do with these things. Jack BeVier (55:05) We must not put the last one. David Moses (55:12) So. Okay. I want to take that one more step. What, what, what do you want to build in the next week where you can come back and next week you can say, know what? I, I like, this is, this is what, this is kind of something I, I wanted to do. And by, by next week, I'm to build it. Craig (55:16) Let's go. Jack BeVier (55:27) Thank you. Craig (55:32) So I'm a little hesitant right now to show off the things that I'm actually building that make me more competitive and more efficient. But I'll just tell you what we did last week. Jack BeVier (55:45) Come on, man, that's the good stuff. That's why they dial in. David Moses (55:47) That's yeah, that's that's the Craig (55:48) I'll tell you what we did last night. I had lunch with Fred Lewis about six or eight months ago and Fred was talking about a new platform that Jack and he were launching called Dominion Lender Finance. It's very exciting. We're lending to lenders. So basically it's a headache free line of credit, David, for anybody who's balance sheeting their loans. If you've got a million dollars in the loans, you get set up with us and we'll send you back $750,000 for those loans. Jack, know I'm sort of giving the broad strokes there. Jack BeVier (56:00) Mm. Craig (56:17) I think it's an amazing use case. think that there's so many small to mid-size lenders who have cashflow issues. so Jack and I used a database that's available to us called Elementics, and we pooled the top 100 to 1,000 lenders in the United States. So in terms of size, anybody from 100 to 1,000. And then we sort of augmented all of the stuff that we got from that database, which is Jack BeVier (56:40) So, I'm going to get back to it. Craig (56:44) with some additional information from a skid tracing platform. So now I've got this great CSV file Dave of all of these, you Jack BeVier (56:49) So thank Craig (56:53) everything that you need to know about these lenders, the loans that they do, the people who they lend to, their websites, where they can be found on LinkedIn and all of that. And I created a skill last night that basically is a five-faced approach of more as much Intel as we can get on all 171. I think it was like 500. I apologize. 571 of these companies and pull it into super base. Jack BeVier (56:58) I'm sorry. Craig (57:21) So that ultimately when I do outreach to these folks, I know literally at my fingertips all about their businesses, where they are, who they lend to. And I can have communication with them that is far more substantive than if I didn't have all of that. And so right now it's like phase four, pulling all of that Intel in. And then the second part, which I'm really excited about is I love posting social media. Jack BeVier (57:23) Obviously. Craig (57:49) And now imagine if I've got this immersive database of information, how easy it would be to target these folks with content, find out who's commenting on their sites and their LinkedIn pages, and then just create. I have an entire app that just creates content on the fly. It goes out, finds stuff that people are, that's trending in our industry. It then sort of re, you know, remixes that content for me and then posts it automatically to So that's just a couple of things that I've been working on over the past month. David Moses (58:22) Do marketing use cases always make me giddy? I'm just a terrible marketer. It's just not. That's pretty cool. Craig (58:27) So one of the things that I always go back to when I started working with Dominion back in 2015, Jack had this crazy idea to blow up the lending company and go from the volume that they were then to something much larger. And Jack was able to find this really great list of all rehabbers in the United States. The property addresses, how much they bought them for. Jack BeVier (58:43) Thank you. . Craig (58:53) kind of surmised at how much they would have rehabbed and then how much they sold the properties for. And we were able to create direct mail. Remember, Jack, that was just so, it was so dynamic. It was like we were stalking them without actually like without seeing creepy, but like we developed those letters and they were phenomenally successful. And I'm thinking like, well, what's the electronic version of that now? What's the AI version of that now? Right. Jack BeVier (58:58) Thank I'm glad. David Moses (59:07) I remember I used to get it. Jack BeVier (59:08) Yeah. Yep, by the way, that was like 10. was literally just 10 years ago, right? Like 10 years ago, a SQL query and a mail merge was like we invented fire, you know? David Moses (59:22) Hahaha Ha Craig (59:27) And I mean, the letters, they we got such an insane Jack, I would get calls from people like I left Dominion and like two years later, like who how the hell like how did you know this stuff? Because letters were beautiful and it's all this dynamic data in them. And yeah, I think that I think that AI is going to make that you know, 100 times more. Jack, what are you bringing to the table? David Moses (59:46) Yeah. Jack BeVier (59:47) I'm gonna go. on my sock to compliance. I know I'm gonna go. I'm gonna go and figure out how to get open claw into a sock to compliant context. That's that's like inventing fire right now. Like everyone's all freaked out about you. You know, if you say open claw in the context of business, everyone's like, that's that's a terrible idea. Like the like, you know, corporate will definitely not let you do that. Right, David. The issue is your corporate, right? You're like, well, it. It's my database. If I if I blow it up, it's on me. And so you're Craig (1:00:11) Yeah. Jack BeVier (1:00:13) letting you're doing it right you're just making it happen which is freaking amazing right like that's that's going to be a huge differentiator everybody else is just blaming compliance and blaming the risk department and we're blaming their regulators if you're a bank right and they're not going to get allowed to do it and we're all just going to freaking walk them right like i'm not asking right like you know when i'm going to these folks i'm just like hey i'm i want to use this tool you know that's what i'm going with them right was like hey i want to use this tool but i want to do it in a sock you know, sock to compliant way. They're like, oh, no, you can't do that. like, oh, I'm sorry. You thought that I was asking like, I'm not like, I'm going to do that. Yeah. I'm going to do it. Let's figure out how to do it as safely as possible. It wasn't, it wasn't, I didn't, I'm not asking permission. I'm asking for collaboration. But, but I, just feel very strongly that like, no, it's just not going to work. David Moses (1:00:48) I'm going to do it. Jack BeVier (1:01:09) Like that's not that's not what I'm looking for here. Like that's not that's going to get us put out of business by everybody else who says, you know, yes, but you know, yes, but fine. But no is not. So anyway, that's what I'm going to go. I'm literally going to go work on that. I'll report back. Craig (1:01:24) we will expect a report back. David, what are you working on? So when we come on to the next episode, you'll be blowing minds with your use cases. David Moses (1:01:25) Awesome. So on Tuesday, we had this plan. No, no, that's fine. So on Tuesday, and we run on EOS, which is like a business operating system. Anyway, so on Tuesday at our leadership L10 meeting, somebody said, hey, I can't get the... Jack BeVier (1:01:35) Thank you. Craig (1:01:40) Jack, this is the greatest podcast idea. I'm sorry, David. I am so loving this. Keep going, I apologize. Jack BeVier (1:01:43) Okay. David Moses (1:02:02) I can't get connected to the API for mile IQ. And we were going to use mile IQ to have everybody log their miles and get reimbursed for their mileage. but apparently they say they have an API, but they really don't. So, so I said, okay, what's the way around this? And I was thinking about this last night. no, it two nights ago. Sorry. The day's all kind of, it could kind of melt together. but, ⁓ so. Jack BeVier (1:02:11) Thank Thank you. Thank David Moses (1:02:28) There's a free software called TrackCar, T-R-A-C-C-A-R. And you can set it to just essentially ping in whatever intervals you want a GPS location and send it to a server if you serve it yourself. So I was like, okay, let's not do this for mileage because that's just way too simple. I actually want to get rid of checking in and checking out from work orders. I just want this to track them throughout their day. I want to take Jack BeVier (1:02:31) Thank All the time. David Moses (1:02:55) their GPS locations for a 24 hour period. I want the AI to figure out this is how much time they spent at this job, this is how much time they spent at that job, this is when they got to work, this is when they left work. And I think I can get that done in a week. I think I can build that in a week. ⁓ Jack BeVier (1:03:01) Thank you. Yeah, just like I'm replacing a software like I'm paying I'm paying whatever you know paying three grand a year for this I don't anymore because I just Created it I replicated all the essential features myself and if I've got a but if I've got an upgrade idea I'm it's it's instant upgrade. I'm just gonna do it myself I don't I'm not gonna put a ticket in and find out that it's on the roadmap But yeah, we're gonna get that feature in seven months when I'm like that's kind of a very problematic time frame for me to deal with this for the next seven months David Moses (1:03:19) Yeah. I'm just going to do it. Jack BeVier (1:03:40) Nope, I'm gonna just write it in. I'm gonna write that feature into here and it's gonna be part of my software. Love it. Craig (1:03:45) Let me see if I can break that down to like an easier use case. all of our phones have location services. And I've said anything that you can do on a computer, you can now do with these tools. So I gave mine full access to my location. And when I get to work, it knows I'm at work. And it knows that I have a daily flow that I need to follow at work. Part of that is I got to get down on the floor and do some pushups. throughout the day. so Rocky, in his very succinct way, will tell me, hey, it's time to get down on the floor and get those pushups in. And then it tracks how many I do it. It tracks my streak. How many how many I'm over 3000 in less than 30 days, guys. So it's cool. And so it's the location thing that's so cool about that, right? It knows where I am. And it's and it puts that in context. It knows that like eyes at home, he's not going to push ups now. David Moses (1:04:31) Nice, nice, nice. Jack BeVier (1:04:32) Cardamom Craig (1:04:44) drinking some bourbon. And so it hits me during the day or when I'm at work for certain things as well. It also calls me in the morning and wakes me up with a message like, hey, get the hell out of bed and get to the gym message. So and it knows because of location where my gym is and it knows if I went. it creates that, Craig's got one day at the gym. That's a streak of one. If I don't show up tomorrow, the streak is over. And all of that's in a really cool mission control that I've built. So. Jack BeVier (1:05:14) And the thing the other thing I love about because Craig was telling me about this a week or so ago, and you've got it trained on like different motivational, you know, personalities, you know, you want Tony Robbins, David Goggins, whoever you want, like, whatever gets you going, right? You say, hey, be be this way. Craig (1:05:29) So David, I David, I used to have a or I still have a very good friend who was very high up at Blackstone. She was basically like the whisper between Larry Fink and like all of his all of his underlings. And so she's highly trained in like performance coaching. And. ⁓ Yes, exactly, Jack. Very good. Yeah, exactly. And so. Jack BeVier (1:05:47) She's Wendy on Billions. David Moses (1:05:50) nice, okay. I know exactly what talking about. Craig (1:05:53) Because I know her, I know all of the certifications that she's spent years of her life getting. And I basically sat down one night and I was like, hey, if you were going to train yourself in this certification and this certification and really be an expert, I gave it the websites and all this documentation. was like, program yourself to know me better than I know myself. And there is a tool that these people use called Inneagram. David Moses (1:06:21) Yeah. Craig (1:06:22) Dude, when I tell you I've had sessions, had there, I'm in a five phase session with this coach. Now that it, when I tell you it's talking to me, like as if it's known me all my life, like my childhood and why I've, why I've built a certain way. It made me cry. I mean, it was that like the coaching was, it made me cry. Now let's say that it's got that great memory of you at this, this, this skill. has all of that memory and that, and that, and that training. but you tune the voice, you tune the voice that comes back at you as something different. Like I want it to be nice to me, so I'm going to tune a nice voice. I know I want it to be David Goggins, knows all about me and all my little idiosyncrasies, but it's tuned to Goggins to keep me motivated. And so that is something that I'm working on right now that I will tell you is absolutely mind blowing at the results that I'm getting off of it. And I've only had like four sessions with it. It records everything. Go ahead, Jack. Jack BeVier (1:07:14) So yeah, so Dave, like I'm a branch off of that idea, which I like love is like, and so I need to build one of these for every single one of my employees, right? Like when they come into their, you know, they're, they're going to come into work and they used to, they used to, you know, open up three different browsers because they had three different databases that they needed access, right? I need to build them a UI that aggregates the data from each of those places and has and frankly is designed for their workflow so that it's very easy for them to go through and do their work in a beautiful place. And then I need and then I need a personal coach on the top top right. I want Wendy from Billions like keeping them going and see and you know seeing that they're like a little down like you know or and you know keep keep them up and make and make it a place where they're like they open up that computer and they're like I like it here right. It's Craig (1:07:45) Yes. Yeah. Jack BeVier (1:08:07) It's a it's a Zen feng shui. I like my coach and I enjoy doing my work. I'm like, if I can create that whole baby like that'll be that'll be fire also like. Craig (1:08:15) Dave. I'm going to write that down. I've developed David a dashboard because I believe that over the over the coming months and years, if you think that people are distracted now, wait until they have all of this at their fingertips. If you think that it's hard to stay focused now. And so I believe that like a work environment, we should never have to open Outlook. We should never have to open Salesforce. I should never have to open 17 different browsers. Everything can be a widget. a dashboard of widgets that can be dragged and dropped and each little widget has its own sort of like function, right? Like it's its own little app, but it all lives on this really slick dashboard that I've created. And so that's something that, and then the second part is you develop the widget creator, the idea lab. And so, so instead of having to teach everybody, all 200 people in the company, how to learn how to use all of these tools, you give them a widget creator. that allows them to create these little things that they put on their dashboard that make their lives easier. They can just drag and drop them anywhere they want on this grid, which I call the playground. yeah. Jack, one last use case and we'll stop. Kyle's dying right now. I didn't know it was going to be a five hour episode here. So, Jack, I have to tell them, I was at a conference. I won't say what conference it was, Jack, but we were at a very large conference a few weeks ago, Dave, with... David Moses (1:09:33) Ha ha ha. Jack BeVier (1:09:35) Thanks for watching. Thank you very much. Craig (1:09:44) 70,000 attendees. And part of the conference is like any conference is that you get a lanyard and the lanyard has your name and company name and that it has a QR code that if you have the very expensive scanner that the conference sells, you can scan the QR code and it goes into a little database that you can then when you get back to your office. Now you've got all the names of the people you talk to and some notes, their phone numbers, emails, etc. The scanner app is $500 a person. Jack BeVier (1:09:56) Okay. it can be commanded from a device. ⁓ Thank Craig (1:10:15) for the three days that you're there. have four people there. So I went to public school. That's 2000 bucks. I'm pretty sure about that. And so I was like, well, the hell with that. I don't want to do that. I've got a lanyard. Let me see if I can scan the QR code right now and see what comes up. Well, it was encrypted. The only thing that would come up was the name and the company name. What I couldn't get was the phone and email. Dude, I developed the most beautiful, insane It was the most incredible app. Jack BeVier (1:10:17) Thank you. ⁓ Here. Craig (1:10:44) that worked perfectly, looked beautiful, gave it to all four people. all took our voice notes, came back with, I think, 190 leads from, and we didn't spend a dime on their stupid $500 scanner app. It was the best thing. I called it OSnap because, OSnap, I just took your data. OSnap. Yeah, it was awesome. Jack BeVier (1:10:53) Thank you. David Moses (1:11:03) That's awesome. That's really, that's, that's, that's pretty cool. I actually have a conference I'm going to tomorrow. ⁓ that, I'm going to wear one of those plod recording things. I don't know if they're going to do that, but if, you know, maybe, maybe they will have the, the QR, but what I've been doing is just like, you know, I'll have it record my whole day and then it will, it, automatically goes into Mongo and then it'll, you know, just kind of like reads the transcript of the whole day and figures out, you talk to these people. Jack BeVier (1:11:10) Yeah. Craig (1:11:11) me to send it over? Jack BeVier (1:11:19) Thank you. Thank David Moses (1:11:32) people. ⁓ Craig (1:11:34) So what do you do? Just keep it like, how do you, how do you have it record everything? Like just it's an app that just keeps, keeps the mic on. David Moses (1:11:39) It just, it's a little device on my wrist and I push, I don't have it on right now, but it's called Plaud, P-L-A-U-D. You can also get one that looks kind of like a credit card and you just push the button, turn it on and it will, on a full battery, it'll go all day. I don't know how long it'll actually go. Maybe it'll go longer, you know, but I've had five hours, six hour recordings. And it's, you know, it's a gigantic amount of data, but you know, it's a pretty... Jack BeVier (1:11:47) Thank ⁓ Craig (1:11:48) Yep, yep, Yeah, I've seen it. Jack BeVier (1:11:55) Okay. ⁓ Thank you. David Moses (1:12:08) cheap model to take the transcription and actually it's part of your plaudron. You can just pick any model and it'll just do it. You can actually pick Gemini 3 Pro. It'll actually go and summarize it based on Gemini 3 Pro and you can give it the prompt you want and it'll do the summary or you can just take the raw text and do whatever you want with it. Craig (1:12:13) Yeah, they're about 100. Jack BeVier (1:12:14) Okay, stop. Craig (1:12:31) Jack, any last thoughts, Jack BeVier (1:12:33) No, man, I'm really enjoy the conversation the time always flies and I think it's gonna yeah I like the idea of having the running narrative here of like stuff we're working on that'll be fun Craig (1:12:43) Yeah, be on. David Moses (1:12:43) Yeah. So next week we have a marketing automated marketing automation, SOC 2 compliance and automated automated time tracking. Jack BeVier (1:12:52) Mm-hmm. Let's go. All right, boys. Craig (1:12:54) you go. All right, folks. Well, that was the kickoff episode of Fort. Pretty soon you'll be able to find us at fort.ai. It's the number 4kd.ai. We're super excited. Hope you guys enjoyed this fairly lengthy episode. Don't know if they'll all be this long, but guys, Mr. David Moses, Jack Bevier, such a pleasure to hang with you guys for an hour and 20 minutes today. Can't wait for the next one. Hope you guys enjoyed it. We'll see you on the next one. David Moses (1:13:23) Thanks, Craig. Thanks, Jack. Craig (1:13:25) I