Latest / Real Investor Radio Podcast / 4kd.ai Ep. 7 | The AI Leverage Revolution

Transcript

4kd.ai Ep. 7 | The AI Leverage Revolution · Real Investor Radio Podcast

• Craig (00:01) Hey, welcome back to Forked AI, the podcast for those who want to know more about all things artificial intelligence. Joined again today by the great David Moses. Good to see you, sir. Unfortunately, Mr. Bevere will not be with us again today. He's traveling. so it's just you and I at the helm. Let's let's get rolling, man. I've got so much I want to talk to you about. But right before we were right before we pressed record, David Moses (00:14) Good to see you. Craig (00:30) You and I were just going back and forth on like, there just seems to be, if you allow yourself to get sort of taken by it all, there seems to be this overwhelming rate of change and tools that we're seeing that are coming to the market. And you and I were talking about, there was a club code conference about a week and a half ago where Anthropic announced quite a bit. And you specifically mentioned the overwhelming aspect of everything that's coming at us here. Talk about that for a second and then we'll talk about with Anthropic Released. That was really cool that people could start using today in their Outlook box to try to create more organization and outlook. So I'll talk about that in a second. what are you saying, David Moses (01:14) Yeah, so we were talking about this before we started, I really, feel like, you know, we're kind of like a kid in a candy store with these tools, like really, like I can't believe I can code, I can't believe I can do things that, you know, it feels godlike in terms of what we can create. But when I look at the foundation model companies and the things they're coming out with and how quickly they're coming out with them, it seems like... these developers, these rockstar developers at the foundation model companies and other places, those people armed with these same tools are coming out with really amazing things really frequently and it just makes your head spin. How do you keep up? How do you know, that would work for my use case? It seems like every time I try to build something, Craig (01:57) Yeah. David Moses (02:04) someone tells me or I read about something that like, that already does it. But do I want to use it? Do I want to pay for that? Or do I just want to recreate it myself? Like why would I, you know, that and it's just that phenomenon is becoming more and more difficult to like those lines are becoming blurrier and blurrier. Craig (02:09) Yeah. Right. tell you the most recent interesting example that speaks directly to that. So on a podcast episode a few ago, we talked about Claude design. Claude design is basically a figma killer and it allows you to create your user interfaces much more efficiently. Effectively, you can give it various skills so that it'll make it look like what you want it to look like. And it was it's actually a really great product that's brand new by Anthropic. Well, someone has already copied that product and now it's a full open source repo on GitHub. But the crazy thing about it, Dave, is that it goes so far beyond what the Anthropic Claude design app does. So like here you've got this groundbreaking app that a major, you know, LLM provider comes out with, Claude design. And someone goes, that's fantastic. I'm just going to make it way better. And they do exactly that. And it works just like the original, but what you can do with it is extended because they've added better skills instead of five skills. There's 119 of them now that can, that can develop your UI. And I just think that that, that speed of change that we're seeing right now, You're really starting to notice it now. You're really starting to notice like, oh, Rev 1 is amazing. Wait a minute, Rev 3 is like a day later. And so yeah, I'm going to do a little show and tell later on something that I've built that I'm frankly blown away by the fact that I was able to do it in a couple of hours. So what are you excited about? What are you working on? What do you got going on? David Moses (04:11) So I'm still working on really boring backend accounting stuff. And I know it sounds boring, I really do think when Claude has this new thing called Anthropic for Small, or Claude for Small Business, which is really, it looks to me to be like a suite of connectors. I haven't kind of played with it too much yet, but their intro video, Craig (04:29) yes, yes. David Moses (04:37) talks about the QuickBooks connector and talks about how, okay, you just wanna say, okay, what's my cash position for payroll, next payroll period? And it literally runs through all the QuickBooks data that it needs to to build a cashflow forecast and make a recommendation to a small business owner on whether their cash position is, they'll have to, stress about payroll or not. And I thought that that, you know, I looked at it and I'm like, yeah, here's the problem with that. Most small business owners, they don't really keep their books, like, up to the date or minute, right? So if you're operating on data that's a month old, two weeks old, whatever it is, Craig (05:21) Sure, sure. David Moses (05:27) If they're entering and reconciling their transactions once a month, which is reasonable for a small business owner, they're not going to be able to use a tool like that. But if they can implement a tool that looks at the transactions on a daily basis and suggests how to record them, uses context from phone calls and emails to... Craig (05:32) Mm. Yes. David Moses (05:51) enrich that and really understand what that transaction is and how it should be booked and whether it's recurring and what's gonna happen next week, the week after, the month after that. Now I think you can, know, now those books are gonna be stellar and up to the minute and with a tremendous amount of good data. So I'm doing a lot of boring back end processing of documents. Craig (05:56) Right. So, so let me, let's play that out for a second. So you've got this connector that is made by Anthropic and I would assume with some cooperation from QuickBooks, right? And it, and you, and you basically install that and what you're saying is, is that, hey, it's a great, it's a great tool and it'll only get better. But if your use case is if you're a small business guy and you're really not keeping your books up to date, then it's probably not that useful. So why not take the time now to build something that what I find is like with this particular tool. OK, so let's say you like you set up your QuickBooks and now you're getting really pristine accounting. What you're saying is, hey man, when I connect that with my phone API, with my CRM API, when I connect that with all of those things, now it becomes really powerful. David Moses (07:10) Yeah, think that's the, kind of that's where it's headed. Someone will eventually have some sort of product. Someone will productize this, but I don't know how much real value there is in kind of solving that because there's so much. you have to do from a business perspective to make sure that your people are buying into whatever processes, right? It's like how many people at XYZ company, let's say a mortgage company, You have loan officers that are handling transactions with their customers. Are they? texting them and calling them from their personal phone? Well, if they are, the system is completely blind to everything that goes on in that interaction. So that becomes where you first need buy-in. Everybody needs to be contributing to the data that the AI will ultimately use to make these really incredible insights and decisions. Craig (07:56) That's correct. Yes. David Moses (08:15) and we really do have to, so you have to get buy-in. So it's like whatever somebody productizes later on down the road, that's only maybe a third of the battle. It really doesn't matter. You could literally have every single phone call, email, and text message, if it goes through a system that has an API, you could just put all that stuff into Notion or, I mean, a series of MD files or a series of text files. it doesn't really, AI doesn't care where it gets the data from. As long as it has some reasonable understanding of what this transaction was and when it occurred and then can read the actual hard data from the interaction, it'll figure it out. And if it can't figure it out now, we're days, weeks, months away, not years away from it being able to figure this out. So buy-in is really, really super important. You just have to have the data going in. But going back to the... Craig (08:44) Mm-hmm. David Moses (09:09) QuickBooks example, your financial data is kind of, I mean, how often does a small business owner look at a P &L and say, what is that? Why did we spend money on it? Why is it that price? Who's using it? Why are we paying for it every month? Those questions are asked every day in every small business in America. And, Craig (09:31) Yeah. David Moses (09:32) if it had the additional context from all of the communication that happens, those questions would be answerable, those questions would be answered, and then real money would be saved, and real money would be kind of reallocated to better uses if people had a really good understanding of what that transaction actually means without having to... Make a phone call, send an email, send a text. Hey, what is this? Why is this? Why are we paying for this? Why is this on this credit card? Why is it coming out of that bank account? Whatever it is. Craig (10:08) Agreed. So getting back to what I was teasing in terms of the announcements. So today you can. And I believe this was as this announcement was there as recent as about Thursday of last week. So if you have a twenty dollar anthropic account you can connect now Claude to any of your Microsoft apps suite. So Outlook. Teams, PowerPoint, Excel. And now you can have Claude running natively inside of those apps. what's the use case, Craig? Well, you can download today Claude for Outlook. And let's say you're just not a person who loves email. Well, now you've got a Claude agent that runs inside of your email box. And it's like a little sidecar. It's like a little side panel pops out and you can now speak to your inbox through Claude. So, hey Claude, triage the last three hours of my inbox and let me know if there's anything in there that appears that I need to respond to quickly. Otherwise, flag the rest for later today. And it'll do that for you in an instant. I'm sure you could think of a million use cases of where you could unleash. Claude in your Outlook inbox. But it's pretty exciting, man. I mean, it works. You know, there was Copilot already. And frankly, and I will give this to Microsoft, Copilot is actually a little bit more useful these days to do similar tasks. But the full power of Claude, especially if it understands, if it has an understanding of what it is that you're really doing, like you kind of pre-program it, you trained it a little bit. It becomes a very powerful little tool inside of your Outlook inbox. The question though, though I had though Dave is, so we've got API access, right? Through Microsoft PIM and Entra, and you can have like your own app that runs inside of your inbox that's outside of what I'm speaking of now with the cloud in your inbox. The question is, how do you get the one that runs natively in the inbox? to talk to the API outside so you can really do some interesting stuff. You can't do, with the CloudFir Outlook, can't schedule tasks, you can't have like a routine that runs every day, hey, check my stuff at eight o'clock in the morning so that when I get here at 8.30, you you've got some report for me that is going to be my email triage. You can't do that. David Moses (12:33) So how long do you think it'll be until it has that? Days? Weeks? Yeah. No, mean like, it might already be the case. Craig (12:36) man, the end of the week. It's like, I could be right this second, you know? Yeah, yeah, yeah. And I'll tell you, man, for me, we talked about it on the last podcast with Jack, and he, you know, he doesn't seem to have a problem staying organized with all that he does in Outlook. I just I just find email to be sort of the bane of my existence. And I find that there's a lot of email that I sent. you can do you can also do drafts with the Outlook, the Claude for Outlook. You can can easily have it write drafts for you and then review and send those drafts. But that to me is the most coolest part. How can I get it to look at what the email says, write up a dynamic draft that includes the borrower or whoever it is that was sending me the email. And then just, I take 15 seconds to review and send. I don't think we're too far away from that really working on sort of a David Moses (13:09) Very cool. Craig (13:31) on table stakes basis. David Moses (13:33) No, it's kind of like what I built for, so our phone call system, whenever somebody calls in, it takes the transcript and it takes the summary, AI summary, and it first does entity recognition. It basically says, this is the customer or tenant or whoever is calling, vendor who's calling. And then it looks through to figure out, can it go a step further? Craig (13:54) Yeah. David Moses (13:59) you know, this is the invoice or work order this vendor is calling about, this is the property this tenant is calling about, know, figures out entity enrichment is like, that's the most critical part, is figuring out, getting a 360 view of not just who the person is, but what are they talking about? And what records do we have in the CRM from other phone calls, from emails that relate to the topic that they're talking about, And then... Craig (14:16) Yeah. And is that just in their record or in all records? So I've got a thousand potential clients. Is it looking at sort of like similar calls from all calls or just the customer account call, like their account call, previous history of calls? Just curious. David Moses (14:39) ⁓ It's them and then the interactions where they relate to others. So like if it's a vendor and they're talking about a work order, it'll pull in the interactions between them and that tenant, right? ⁓ But it's interesting to, because it certainly could do some sentiment analysis and say, you know, Craig (14:53) interesting. David Moses (15:02) yet again, we're having a call about this topic. This is a pain point for the company and we need to figure this out. That's kind what you're talking about is that if 10 people complain about the same thing, my system's not going to catch that because it's really just looking at that person's situation. If they complain 10 times about it, I'm going to see it 10 times, but if 10 different people complained about the same thing, it won't show up. But that's brilliant. But what it does is it will create a recap. Craig (15:04) Right. Sure. David Moses (15:29) an internal recap that goes into our CRM of the phone call, what was talked about, all the entities related. And then it'll create a recap that the user can click a button and it'll create a draft in their email box to send that recap of the phone call to that other person. So here's a recap of our phone call. discussed this. I mentioned I'd get back to you on this date. I mentioned I need to get this to you or you need to get this to me on this date, whatever it is, right? So it kind of gives them a recap, something they have in writing that kind of memorializes what we talked about. Craig (16:08) let you know you heard them. great. And it's just another little customer touch if that's a customer service touch, right? Like it's a great little thing. David Moses (16:14) Yeah, exactly. And then it also, for us, it removes the friction, right? There's a lot of friction that happens when a phone call happens and somebody doesn't understand or misinterprets or misses something, right? Those recaps become, here's what I got out of the conversation. And it says at the end, if there's something I missed or if there's something I got wrong, please respond. And that way we clear it up at the time the phone call happens instead of, you know, in an extreme situation, you know, going into a courtroom and saying, you know, the tenant made a promise to pay such and such a date. We gave them an extra grace beyond that of this and, you know, and now we're here. and the tenants, they're saying, no, no, no, no, they told me this, this, this, and this. And, know, and then it's like, well, your honor, as a matter of fact. Craig (17:02) Yeah. David Moses (17:03) We had this conversation and we sent them a recap. We asked them to respond if the recap was wrong and they didn't respond. But more often than not, what you're really doing is just telling the other person, hey, I've got my shit together. Craig (17:17) and I heard you, I logged it, and I let you know that I heard you by logging that. Yeah, like, I don't know how you come across as anything less than having your shit together at that point. David Moses (17:30) Yeah, I think people appreciate that and they value that. For us, it's community management and property management, and I think the people on the other end of the phone, I think they do place a value on having competence, especially any tenant or community board member or vendor who's dealt with a management company that doesn't know what the hell they're doing or doesn't have their shit together and how... quickly and terribly that can go off the rails. Those people are the ones who appreciate it most. And that's what we're trying to build, It's just trying to build, use these tools to put us in another league, a league that's beyond maybe our competitors and maybe a league that we couldn't have played in before these tools, right? A league that, whether it be on the individual level, right, like somebody at a big company who just, maybe they don't have the experience that the guy in the cubicle next to them has or the guy in the office across the hall has, but they can still appear that they do because where they're inexperienced or where they, know, that person and that person's knowledge and experience so far combined with a tool like Claude or Chad or... you know, name your LLM, combined with all the context and all of the, yeah, connectivity, yeah. Now you become a superstar very quickly. You know, very, very quickly. Craig (18:54) connective, the connectors, right, that we have now, yeah. Dave, I've been thinking, I love talking with you and Jack because you're sitting at the helm, right? Like you both own companies, successful companies, and I love your take, your more global take on sort of what your vision is for what AI can do for your company. What I find that I think more about in my time as I program these tools is how does it affect my job, right? And what am I building that's going to exponentiate my memory, my follow-up skills, my every process that is computer related that is going to enhance me. And Dave, I'm like, here's my question. I think there is a, there is a X, some sort of exponential factor to implementing those tools for an LO. What is it? Am I 10Xing my ability to take on more volume? Am I 2Xing? What will it be? I just know that it's going to be a lot more with these tools. And so if you believe in the abundance that's going to happen as a result of these tools, even down to the real estate investor level, as a guy who originates loans for investors, I know how hard it can be to be a volume person. not really, it should not be a volume job. It should be like, I've got really, I got a great book of customers that I service really well. And the question is how many more will I be able to service really well because of these tools? And I think that's, I think it's at least a two to three X factor, but way more. David Moses (20:35) I think you're right. think from the standpoint of an LO in a lending organization, it may not be servicing a volume of customers as much as it might become servicing a smaller or the same size group of customers, but helping them get their volume increased. Craig (20:54) Oh Dave, 100 % sir, 100%. David Moses (20:57) I'm still having the same conversations with the same people who I like working with, I'm compatible with, and I'm just gonna help those people 2X or 5X their businesses, and then by extension my business, know, 2Xs and 5Xs. Craig (21:09) love that. I love that. And I think to the whole business aspect of it, think the more that if you're like, let's just stay with the yellow role. I think the more that a company empowers those people now, not 12 months from now, but now to be, hey, you're the subject matter expert in this thing. I'm gonna give you the tools to be the greatest subject matter expert ever, right? And not because you'll be able to take on way more volume, you'll be able to go much deeper with the great clients that you have in providing real value for them. Right? Yeah. David Moses (21:49) Yeah, we're already seeing it. We're already seeing it. I gave this tool last, I think I said this last week, that I was going to give or that I had just given Claude and connected it to our community management software. I created, actually had, it was, I used... Craig (22:02) Yeah. Is this the one where you give a report to your management companies and that report is like a real bane of your existence? But like you just like, I'll just give it to them and they can create their own. Is that what we're talking about? David Moses (22:16) Well, was kind of the impetus for why I created the connector. But I went into OpenClaw and I went into my agent, I call him Forge, who builds shit for me. Because Claude wasn't going to build this. It needed much more root level access to do these things. So it created a Claude connector, a custom connector for Sync, which is our community management software. And it basically created a tool for every single thing in Sync that... Craig (22:20) Yeah. David Moses (22:41) anybody would even want to do. It went through all the endpoints, which I had already had Claude run through and create an endpoints library of every endpoint, what it does, what it'll find there, what the schema looks like. And that took five minutes. mean, that sounds really complicated, but it's really not. And so I created this connector and I gave it to Bernadette, our lead, who is our senior community manager. She runs community management, essentially. Craig (22:43) Mm-hmm. Yes. Alright. David Moses (23:07) And she, and I said, you know, just ask it a question. And she's like, okay, how many delinquencies do I have at this community? And just spit out an answer for her. And she was like, and it gave her like the details, how many were this many days, this many days late. And she was like, wow, like that was very quick. So she got, it just so happens that she had this email, right, from a treasurer asking how they were going to, they had just foreclosed on a unit. Craig (23:23) Ha! David Moses (23:33) the community foreclosed on a unit for, I don't remember if the person passed away or whatever it was, nobody was paying the dues, they had to foreclose on the unit. And they needed to know how, an accounting needed to book the foreclosure, which obviously you can imagine doesn't happen very often, it's not like a journal entry we do every day, right? And she said, well, I need to ask the accountant how we book this. And I go, just go ahead and ask Claude. So Claude went through. Craig (23:51) Right, right. David Moses (24:01) looked through the ledger history of the co-owner and looked at the foreclosure documentation. And literally this was all, like it just grabbed it all that was already attached in sync. Read it all and then put an email together, without even asking to put the email together, put an email together for the accountant suggesting exactly how the journal entry should look and why. and giving him the background of what happened. And she sent that email and CC'd the treasurer of the community. And they were like, wow, how did that happen? did that happen so quickly? And so quickly, like the treasurer wasn't even expecting it, but they were blown away by it. And then since then, they've had like in anticipation of board meetings, she's gone through and, you know, hey, take the meeting minutes from the last board meeting. and go through all of the phone calls between co-owners and board members of that community since those minutes. Tell me what on those minutes is still an issue. Like something we talked about that people are still talking about. And it identified some things that, you know, she could give an update on. And those things, it's like giving them the tools. I would never have thought to... Like that's... I'm not a community manager. Craig (25:01) Yeah. David Moses (25:18) every day of my life. I'm a community manager when I have to my community manager hat on, when someone comes into my office and says, this is broken, we need to fix it, or this is not working, or this person's upset. And I get involved usually as a firefighter or someone to clean up shit. But in reality, that's not where the magic is. When you get to that point, you've already failed. Something's already gone wrong. People are already upset. Craig (25:19) Sure, sure, right. Right. Right. You gotta deal with shit. David Moses (25:46) Right. And you can't, and you have nothing to distract them with because the fires have to be put out. Right. So now it's like, you know, you know, you've got these tools, you give them, give the community managers the tools that they need and they're plugging away in there and getting data and it's, and Claude's putting it together in this like beautiful presented way. and they, know, and they can just double check it, make sure it's right. And then boom, it fires out an email or, or, or a custom report. Craig (25:49) Right? David Moses (26:14) of whatever it is the treasurer or the president wanted to see. But you gotta give them the tools. You have to pass those tools to your team because they're the ones who are really gonna figure out, well two things will happen, what we've talked about in the past, which is you'll see where buy-in is, right? People who are buying into the tools and people who are sabotaging them. People who are ready for what's coming and people who are ready to run for the hills or hiding, burying their head in the sand and protect. Craig (26:29) Yep. David Moses (26:39) hoping that it's not really gonna happen. And then the other thing that it does is it gives them the tools to really explore. And the people who are curious are gonna be the people who continue to have jobs and the, you know, because we talked about the doomsday scenarios, but as a business owner, and I'm sure Jack would, you know, I'm sure Jack would echo this, but as a business owner who's responsible for a group of, Craig (26:41) Yeah. David Moses (27:07) you know, more than a dozen people, you for us it's 30. What I care most about isn't my customers, right? Because I can't service my customers directly, right? I don't care most about my vendors. I care most about my employees. I care most, like, they're who I care the most about. The ones who buy in and the ones who are ready to kind of push, move the ship forward, those are the people I care about. That's my customer. My customer is the employee. Because if they're not happy, my customers aren't going to be happy. doesn't matter how many bells and whistles and how many things I do and how cheap I do it. If the people who take care of them are happy, those relationships are going to remain strong. Craig (27:46) Agreed. Yeah, I mean, it's a culture thing, obviously, as well. So you want to you want to do a little show and tell? David Moses (27:54) Yeah, yeah. Craig (27:55) Let me see if I can make this happen without breaking anything. Kyle, just stick with me here. All right. So here's the idea, Dave. The idea is the end result idea is I have a a LinkedIn content generator, right? Like I just want to create LinkedIn content very quickly with little to no intervention from me. The question is, OK, so what is it creating content about? Well, there's many times throughout the day that I'm on my doom scrolling on my phone on Instagram and I'll come across a new story that I think is pertinent for either my lender finance clients that I have here or people who want to learn about AI. so I developed first the database, the SuperBase database that would essentially pull in all the possible data that it could from a scrape from an appify scraper of Instagram. You're with me so far. I can't see you so you'll have to forgive me. So this morning just to give folks an idea of what they're looking at here. I said now the app is done and I'm about ready to show it you in a second but I was like hey I'm getting ready to do a podcast. Can you just take a couple screenshots and make like like two marketing slides for me. David Moses (29:05) It's okay. Craig (29:19) use like the iPhone as a way to show off the UI. And in a matter of seconds and one prompt, it made these really, these cool two slides for me. And yes, you know, they need some work, but as a one shot, you know, not bad, not a bad start to show off sort of the UI and what this is all about. do you understand the concept so far of what I'm trying to do? David Moses (29:42) Yeah, I get it. Craig (29:45) So the UI, so the app that I built, the InstaScrape app, honestly, was more of a vanity thing. Like everything's being saved to SuperBase automatically through the app, but I wanted to develop like some sort of cool UI where I could see my library of scrapes on Instagram. And so let me show you what that looks like on my iPhone. Let's see. All right, so let me turn this off and connect. That's just a simulator that runs on my my Mac mini that shows off exactly what's on my iPhone. Now, I'm currently in Instagram. This is on my phone. And if I want to capture this, this particular news story, because it pertains obviously to, you know, people who invest in real estate, I just hit the copy. And now I don't even need to go into InstaScrape app, Dave. I can just share it directly using the connector in the InstaScrape. It automatically copies the link. I can hit save at this point, or I can have notes that I can add to it now that will enrich its basic direction for when it then goes over to LinkedIn to try to create a piece of content off of it. So I can say why I saved it, give it like my angle, tell it to research these various things. And then I can, you know, I can say this is market data, it's housing, I can tag it and folder it. And then I can hit save. Well, I've already done that for you, but I can, at this point I would hit save. Now I can go over, if I want to see that it did it correctly, I can just open up InstaScrape and I can see that in the library, it's saved that it's given me, it's saving all of the carousel so that I can maybe even use the words on the carousel as part of the content that I'm going to synthesize. Still working on the upload part of it, Dave, but you can see very cool here. so now the second part is that I'm working on that I should have done by the end of the week is the LinkedIn part. Now that you've got something to make a LinkedIn content off of, how does the LinkedIn content generator make that? So I'm developing, you know, I've got an audience MD that knows how to tune it to my audience. I've got a copywriter skill. I've got a research skill that will go out and you know, enrich this post. And then I've got a carousel skill that will then design the carousel. And so hopefully by the end of the week, that part will be done and I can pretty much generate LinkedIn content on the fly. So that's my show and tell. And it took me, by the way, by the way, this app, Dave, I'm sorry, has probably taken me less than three full hours so far to build this app out. David Moses (32:12) That's very- No, it's good. That's mind blowing. That's really cool. ⁓ Is it understanding those, the pictures? Is it analyzing them when it creates the content for you? Is it understanding what those pictures are? Or do you? Craig (32:24) Not cool. That's a really, yeah. So that's part of definitely of these. So when I get over to LinkedIn, I know what you're saying. Is it actually going to understand what's on the pictures? Maybe the content that's on the pictures or if it's a video that I, that I capture, will it capture like the inline, the transcription that happens on the video? Don't know yet. I'll have to find out. We're learning together on this one. So yeah. David Moses (32:59) Yeah, that's it. It's really awesome. Craig (33:03) Thanks, man. Am I still sharing or do I do I need to do you guys still see it or no? David Moses (33:08) No, at least not on my end, I don't see it. So you wanna, I can do a show and tell if you want on field manager portal. I also, really wanna do one on the invoice dashboard. Craig (33:10) Okay, very good. Thanks. Awesome. David Moses (33:22) because it's so cool, but I think I need to get some dummy data in there first because it's, you know, it's, I don't think the communities would really care if they, you know, if I publicly share. Yeah, yeah, I'm not gonna. You shared our landscaping invoice from February of 2020. Anyway. Craig (33:27) Yeah. Well, if you're if it's even a pang of anxiety, you should probably go with your gut instincts on it. Yeah. We had the best deal in town until you shared that. Yeah. So while you're bringing that up, I'll just say that in terms of the why on that LinkedIn content generator, so you can now hook up to LinkedIn by their sales navigator or some third party app, Dave. And you can say, this is my exact audience. These are the people that I am literally content generating for. David Moses (33:49) Yeah, exactly. Craig (34:11) And it will then go out and survey sort of what those people are talking about so that then you can inform your content creator skill on exactly what your audience is talking about. So and then and then the final thing that I would say about it is the time savings of, I want to be someone who is who talks on LinkedIn and hopefully generates business off of it. What what's the time factor on that if I want to do two or three posts a day to really get to the top of the algorithm? That's hours. And I think what I've developed here is going to be a 45 minutes a day of just sort of shaping the content, getting first draft ready, and then being able to copy and paste it into a LinkedIn scheduler. So I'm really excited about that. It's going to save a lot of time. David Moses (34:56) Yeah, I definitely think that's awesome. I did this marketing workshop, which we never actually talked about, but I did this really awesome marketing workshop. Scott Effringham, think is the way you pronounce his last name. But I can get the info. What he has is a series of prompts, and he gives most of it away for free on his website, but if you want to sign up with him, it's not cheap. Craig (35:03) Yeah. David Moses (35:19) But he talks about lead magnets, basically putting on your website. He has a series of prompts that you can just walk through with Claude, because he's a marketing expert, has been doing this for Fortune 500 companies for decades. So he understands exactly what he wants. He created a series of prompts to put into Claude to kind of do it yourself, to create these kind of lead magnets, which are essentially the things that show up in your email. Craig (35:31) Mm-hmm. David Moses (35:44) or on websites that are going to incentivize you to trade your email for some meaningful content. So for us, yeah, so for us it was like we created this, we haven't launched it yet, but we created a scorecard that communities, know, treasure board members essentially, Craig (35:54) Yeah, sure. Carrot content. David Moses (36:11) can take this 18 question survey and it really kind of responds, like it goes through the major pain points of community management from a board member's perspective and lets them essentially rate their current management company and it gives them a score and it explains where the pain points are and it gives them kind of like an idea of what's available in the market. do you have a portal where you know what's going on with a work order within in real time, and questions like that that'll actually walk them through, and it gives them a score. And in order to get that report, they have to put their email. And we don't sell their, this is just for us, but I think it's useful, the marketing stuff that he came up with for small businesses, it's absolutely brilliant. But that's what essentially you're doing. You're trying to raise awareness. You're trying to get posts out there that drive people to... drive people to ultimately work with you. Craig (37:05) Sure. And genuinely, and genuinely trying to use AI in a way that is not necessarily the easiest in creating content, sort of capturing your voice, making sure that like, hey man, like I hope that I get business off of it, but I'm creating LinkedIn content really as a, I want to provide significant value in the content. I don't just want to put AI slop out there, which is really frankly, ⁓ David Moses (37:34) what a lot of it is. Craig (37:35) A lot of it is that. so yeah, that's the real, that's the fly in the ointment. That's the magic of really working, drilling down into the tools and making sure that you're getting great output on stuff like this. It's not easy to get artistic output that really is meaningful. So I'm still a little dubious as to whether or not I'll be able to get this thing to be tuned right. So we'll find out. David Moses (37:56) You'll get it. if you're focused on it, you'll get it. But I really, in a world where the content on the internet and in our email boxes and our text messages and voicemails or phone calls if you don't have a spam blocker, it is diluting real content so rapidly right now that I think you might see the whole thing come full circle. Craig (37:59) Yeah. Yeah. Yeah. David Moses (38:21) where the people who actually use the tools to create things that are really cutting edge and they take the same amount of time that they used to take, like you're talking about hours a day, Well, if you can do really advanced, incredible content, spend that same few hours a day, but just have the content be that much more spectacular, then I think that differentiates people from, I mean, what are the old adages in a? Craig (38:39) Yeah. David Moses (38:49) In the land of the blind, the one-eyed man is king, right? All right, so here's what we got. This is a field manager portal, and basically what this does is I everybody install, I talked about this previously, but it works now. We installed this thing called Trackar, T-R-A-C-C-A-R. We serve it, it's freeware, there's no cost to it. Craig (38:51) Yes, yes, very good. Yes, exactly. All right, show us what you got. Show us what you got. I'm excited to see it. David Moses (39:15) It doesn't collect our data. They do have paid versions and things like that, but they also have a free one. You just download, you serve it yourself. And it just goes on the phones and they will, you know, they're sharing their location. It's a web socket, so they're sharing location all the time. Whenever they turn it off, it's off. Whenever they turn it on, it's on. And they basically, you know, we ingest all of these GPS data points. We compare them. to all of our open work orders and to the work orders they log into. And we come up with a way of allocating their time and their mileage. So I'll give Joe as an example. You know, it will, this is like the allocation. Craig (39:55) And just so folks that are listening get it. this app that you have built is essentially for your technicians in the field that are out there making repairs and maintenance guys, correct? And this little free app that you've found called TrackCar allows David Moses (40:07) That's exactly right. Craig (40:16) you to basically track their location to provide the report that you're about to show us now or to provide the functionality that you're about to show us now. David Moses (40:25) Correct, so I can actually show what the workflow looks like. because we we use N8N. Craig (40:32) While you're doing this, when did you start working on this? Like how long have you been working on this to get into this? David Moses (40:36) This has actually been a pain in my ass. I've probably spent 25 hours on this. This has been a pain in my ass because the biggest problem is that there are hundreds of thousands of GPS points that get logged every week and really had to pare it down and just stop, stop the madness, so to speak. Craig (40:53) Yeah, what's the token fee on that? Like were you, were you blown through token? David Moses (40:55) There's no... Ingesting them really is no... It's not a real large amount of data, but we programmatically limit it. So it's like, you know, there's a deterministic workflow. Deterministic basically, for those who don't know, it's just a non... It's not an agent who's coming up with the answer. It's a... It is simply code. Yeah, it's just code that handles the data and... Craig (41:03) Makes sense. It's a process. Right. David Moses (41:24) pre-processes it for AI. So we can basically say, we know that anything more than a mile away from this location or half a mile away from the location that they say they're at is irrelevant data, so don't bring it in. And then it kind of filters that down and distills it down and gives the AI some choices to make. ⁓ I'll get to the workflow in a minute. ⁓ Craig (41:43) Okay, keep, Yeah, yeah. David Moses (41:47) So brings us to this page, right? This is Joe. He's in there as Joe. He does have a last name, I promise. But he's in there as Joe. But this winds up being, know, top it just gives me some data, you know, number of hours. This is kind of like the cost to the property. don't think that really means anything to anybody but me. Craig (41:49) ⁓ here we go. Show is good. David Moses (42:07) reimbursement for mileage if he's in his personal vehicle, but if he's driving a company vehicle, it just changes that to show that that's what the company saves by not having to pay, by owning the vehicle. You know how many miles he drove. And then this is the allocation. So each one of these is a work order. Craig (42:18) Very cool. David Moses (42:24) All right, so I'm gonna go, yep. So now you can see me tabbing through these things. So each one of these, if I click on it, it's gonna take me to the actual, that was helpful. Craig (42:25) There we go. David Moses (42:35) I use LastPass. This is just a free plug to them to say, everybody should use a password. One password is great. ⁓ Craig (42:42) I use one password, been using it for probably 15 years. That's great CLI too. David Moses (42:48) Yeah. All right. So let me go here. you click that, it will actually bring the work order up. And I can see this is how much time has been allocated. And then it shows me this PM log. So they logged in, right? And they said they checked into the workflow at 9, 10 AM. That's when they actually clicked to check in. And this is when they clicked to check out. Craig (42:53) Yep. Mm-hmm. David Moses (43:10) you know, and this is what he did, right? But my allocation actually said he spent an hour and 29 minutes there, which either means he checked in late, right, or he checked out early. Whatever it was, it says he spent more time there than he, that, you know, it allocated him more time than what he allocated himself. So he wanted to, he actually made more money there. Craig (43:32) Mm-hmm. David Moses (43:34) You know, these are pretty damn close. This is a minute off, this is two minutes off, is 16 minutes off, this is four minutes off. But this one, caught and gave him credit for time that he... Craig (43:39) Yeah. David Moses (43:49) would otherwise have not gotten. And the way that works is down here at the travel log, it'll show, all right, he stopped here. He was here for a certain amount of time. And then that's where it gets allocated to. This allocates it on the cost side for us. So we know exactly at a granular level every minute he spends and whether it gets charged to the company or whether it gets charged to a specific job. Craig (44:00) Right? It's awesome. David Moses (44:14) Most of this is really for our own internal costing. We don't bill based on it because most of our bills are kind of quoted work. But we know what our costs are. And this will show, like if he stops at Home Depot, it knows, well, hey, I went to Home Depot. But then after I went to Home Depot, I wound up going to this, know, wait a second. So this will actually show up on the next day. So at the end of his day, he went to Home Depot and this actually got allocated to this gutter thing. ⁓ no, I'm sorry, no, he went to this, he from Home Depot to here. So it assumed that he went to Home Depot to get materials for this job. So it automatically figured that out and it has, we went through and over the last 18 months, every Craig (44:39) Okay. Okay. David Moses (45:02) vendor, every supplier that we went to, physically went to a location to pick something up. And then we loaded in not only that supplier, but all the locations that supplier has. So every Lowe's, every Home Depot, every Sherwin-Williams paint store, every ABC supply shop, whatever it was, there was like 750 different locations. So anytime they go to any known supplier that we actually buy stuff at, it will recognize that that's where they're at. And then it will allocate it. either to the next job or it will allocate it to the previous job or the job after that. Like say they went to Home Depot and they wound up going to two jobs after that, both that had materials. The system will automatically allocate the time they spent at Home Depot and split it between those jobs. Craig (45:50) That's wild. That's very cool. David Moses (45:52) Yeah, so this, really helps our cost accounting. really helps us get very granular on exactly what we're, know, kind of what we're doing and how we're doing. And then each day they click submit approval and that goes to our accounting person and our accounting person clicks another button and all those hours go right into the payroll system so those guys can get paid for it. Craig (46:18) Sure. That is wild. What do your field managers, and you don't have to go through it right this second, but it would be cool to see, like, what do they see when they're in the field? Do they have like a phone interface or like an app that runs on their phone that kind of tells them where their next stop is or how does that work? I'm putting you on the spot, sorry. More editing for Kyle. David Moses (46:37) Let me see if I can show you this. Let me see. Craig (46:45) and while you're bringing that up. I assume that you built what we just saw first and then sort of the reporting mechanism to the guys who are in the field, correct? Or did you build the app that the guys see in the field and then build the cool backend that we just saw? David Moses (46:59) No, we built the back end so that they could, so that their hours got allocated. They were missing hours. So here, I'll show you this. Craig (47:08) Yeah. David Moses (47:10) So this is mission control. And this shows, I'll just show the map. Craig (47:12) this is amazing! What's the ticker at the top? What's the ticker that was just rolling by like that? David Moses (47:20) ⁓ this is my brother being my brother, right? Like that's every, probably every open work order and when it was closed and where someone's at. He had a field day with us. He clearly had fun doing it. But he built this so it opens up like this and then, now this shows, what this shows is every, Craig (47:30) I'm such a sucker for the eye candy. I'm like the biggest sucker ever. David Moses (47:48) So like every single open work order is bright red, right? And a lot of these are, know, people have called stuff in and we've told them we're not getting to it because it's not important for now. We've noted it. So we have to, now we have to start cleaning this stuff up because now you have like this open work order, right? So we gotta close those work orders now or in some other way designate them that they're on the back burner. Craig (48:00) Yeah. sort of filter it out, right? David Moses (48:15) progress. You know, so like each one you can get directions to. Craig (48:18) So this not only shows where the work order itself, but where exactly the property is located. And I think you also have an overlay that shows you where all your vendors are at the same time, right? David Moses (48:28) Yeah, all these little ones are vendors. There's Ace Hardware, Advanced Auto Parts, literally anywhere they've shopped in the last 18 months. It'll show every one of their locations. That didn't take... Craig (48:37) That's crazy. And is the green, is the green, what are the green check marks, the ones that have already been closed? David Moses (48:43) Green are where they've, that's where they've been today. right, so that's, they just got done, these are the things they've completed today. And then there is, let's see what other colors there are. this is where they are now, so that's where Joe is. And it says he's moving at six miles an hour. This is. Craig (48:46) that's so cool. Now, let me ask you this. Does it route them in a way that the route stays way more efficient, based on the, like you follow me. David Moses (49:07) Not yet. Not yet. mean, it's literally the next thing on this roadmap is going to be, it's a rock for my sister who runs maintenance for the end of this quarter, which is to have smart routing so it knows whenever a work order comes in and it's close to another person, it will identify, hey, this person is five minutes away from here. Let them at least go set eyes on it. Take a look and see what's going on. So all these bank looking vault things, those are our communities that we manage. And it shows all the open work orders at those communities. So you may have things where the community has, or a property owner has rejected a work order and said, no, I'm not ready to do that. But if I've got a guy five minutes away and I can call them and say, we're going to wave the first hour, or we're going to... Craig (49:35) Right? Yeah. David Moses (50:03) you know, because our guy is five minutes away. So I can, you know, cut the first hour and a half for you if you'll approve it so I can go take care of it. And if I'm right, you know, so then it creates efficiency and handle, yeah, we essentially trade off a little bit of revenue in exchange for efficiency and they're getting a better deal in exchange for paying it now, right? And we... Craig (50:09) Sure. So, so on your tickets, I would assume that like there's critical tickets and sort of non critical tickets and, you know, certain, you know, maybe some maintenance tickets that have to happen each month. Would you then sort of have a status for the tickets? So like my guy's five minutes away, this one's critical. Let's get him over there. Or my guy's five minutes away, but we know that this one is not so critical. So maybe we won't route them in that direction right the second. based on sort of the status of the ticket, the urgency of the ticket, if you will. David Moses (50:58) Yeah, so we will, so we do maintain the urgency at, you know, kind of like the internal level. I don't know why it's not in here. Like maybe what we could do is take it a step further and at least show right here what the priority level is. It would make a lot of sense. because the guys can see this, right? So they can see, not only can they see where they're at relative to all the open work orders kind of in their proximity, they can also see where each other are at. So they can... Craig (51:12) Right. David Moses (51:27) communicate to each other and say, hey, because the way we now create an incentive program that incentivizes them to complete melds, like complete work orders, like gives them a per work order incentive, and then it also gives them a share of any additional, like some share skin in the game for additional work that they find and take care of. Craig (51:29) I need help over here or, yeah. David Moses (51:52) Right? So we... They have this inspection sheet that they do when they go to a property. Hey, there's all these other things that may need to get done. And if those things then get done in the future, then they'll get a little piece of that. But they share in those commissions. So because they share in those commissions, they're incentivized to cooperate with one another to have the most things done. Craig (52:11) Sure, sure. David Moses (52:20) to us is great. it's obviously there's the gatekeep on the other side because we own most of these properties. So, you know, we care, do we really need to do this? Like, is this something that really needs to get done? But nine times out of 10, if they're calling it out, it's something that if they do it now, it's this much money. And if we wait and they do it when the tenant calls, it's that much money. And so those are things that, you know, if they can take an extra Craig (52:44) Right. David Moses (52:48) half hour at a place and it's going to cost someone nothing for them to just stay there and address three more things. Just do it. Craig (52:54) And do you, do you, so does each of your guys have like an iPad or are running this off of their phones? how does that work? David Moses (53:01) They do it off their phones. We have found that the phones, it's something they always have. It's something that they, like they're big enough now where they can read them. not, they'll get a one that's big enough where they can read it. And then they're, yeah, I mean they're logging in, they're logging their time, they're taking pictures and uploading them. Every picture that they upload tells. Craig (53:08) Yep. David Moses (53:26) tells a story for us. That's why I asked you about those pictures earlier. We have a workflow that literally takes every single picture that's uploaded by a tenant, the property manager, or the technician or vendor, and it runs each picture that's attached to a work order through Gemini. Because Gemini does a really good job with reviewing photos. And Gemini can also review video. Craig (53:29) Yeah, yeah, of course. It does. David Moses (53:51) So they'll take those videos and photos and kind of analyze them with the context of everything that we know about the work order, all the communication back and forth, the questions that were asked, the original ticket, the urgency level, everything. It takes into account everything when it reviews one picture, which is something that no human could ever do. And then after it's done reviewing each picture and giving some kind of synopsis on that picture, We don't see those individual synopses because they're useless to us. Then we have another workflow that takes all the synopses and meshes it all together and that's where the magic happens. With the context of the whole work order and the very detailed explanation of what each picture shows, which in the picture is like, if the picture says they a water heater, it's gonna say, it looks like it's missing a drip pan or it looks like it's, you know, it looks like there's corrosion on the whatever, right? So like. Craig (54:33) Sure. David Moses (54:43) it'll call that stuff out in each individual picture. And then when you put it in the context of the whole thing, it says, well, that was a picture of, that was a before picture. So obviously it doesn't have a drip pan, it was an old water heater. I ever see the words corrosion, but now there's this other picture over here that says it's a brand new water heater and there's no corrosion. Craig (54:59) Dave, how, like, don't mean to belabor the point here, but how good is that? Like, how good is that translation of those sort of very discreet, minute details in between two pictures? I've got a water heater that appears to, I said water because I'm from Maryland. I've got a water heater that is perfectly installed and then, because that's the after. David Moses (55:18) you Craig (55:23) And then I've got one that literally doesn't have a drip pan. mean, how much iteration and training did you have to go through to like make sure that what it sees is really what it says? Really? Wow. David Moses (55:32) None. ⁓ I just to wait for the models to get good enough to do that. Like, Chat 4.0 couldn't do that. But Gemini 3.1 does. But the magic happens because of the context you feed it. So if the work order explains that the water heater was bad, you're just telling it what happened. And so the picture in the context of a picture of an old busted water heater where the Craig (55:45) Yeah. yes, yes. David Moses (55:59) the AI already knows that the water heater was replaced, right? It's going to be able to call out, hey, I don't see a picture of a new water heater here anywhere. Was it really replaced? What happened here? Right? Or it'll say, I see where it was old. I see where it was messed up. I see where it was new. It might be the drip pan is in neither place. And it's like, hey, we forgot. The technician forgot to put in the drip pan when he replaced it. Right? And now it called that out because Craig (56:03) Right. David Moses (56:28) It's not only looking for what's wrong with this picture. Once it analyzes all the pictures as a whole, now it has enough information to tell us to really reliably call out the red flags. Hey. Craig (56:39) I'll you a crazy that tracks directly with this is my house is about 11 years old now. We're getting ready to sell it. And in the basement, have my hot water heater. The hot water heater has a little condensate or not a condensation drip, but a pressure release drip that goes directly into the sump pump. Notice that my sump pump was running like every three minutes, whatever reason I walked downstairs and I feel here this constant drip going into the sump pump from the hot water heater. So I basically explained the problem. I wrote my own work order, right? Like, hey, here's the problem. And then I took pictures of my hot water tank and it said, this could only be one of two things. You've got an expansion tank issue. Your bladder has, is gone bad in the expansion tank, or you've got a pressure relief valve issue. I can see exactly what the hot water heater is. And by the way, if you call a guy, this is probably what it's going to cost to fix if it's one of these two items. And it did that within seconds. I was able to call my plumber. I told him what the issue was. sends a guy over and he was like, yeah, it was the expansion tank. Like mind blowing that you're doing that on scale. Like literally like massive scale. David Moses (57:44) and it gives a But that's all doing is we're doing the same thing you're doing, right? We're just taking the pictures, making sure, hey, tenant, did you actually upload the pictures we would need to make that assessment? If you're not, upload the pictures, right? And then when we send it to the technician, we ask them, you know, here's what we see from the pictures, here's what we think it could be. Do you have enough information or do you need more information, right? So it's all it is, is just taking the logical kind of, you know, how we go about our business here. using a tool that is, if you don't use it, it's, I mean, the ROI on using ChatGBT or Claude to pre-assess a work order in real estate is astronomical. If you do it manually by, like literally take the picture and do exactly what you just did, there's a tremendous ROI on that and a ton of peace of mind that goes with it. Craig (58:32) Ugh. David Moses (58:45) that you know this guy isn't replacing your water heater when all it was was a relief valve, right? Or expansion, right? So, and then, or God forbid, it's an expansion tank, you replace the water heater and it's like, it's doing the exact same thing. So, you know, it's like you got the peace of mind to know you identified the right thing, you fixed the right thing, and you also have an understanding of what it's likely to cost you. A lot of times it's just, hey, Craig (58:51) late. Yeah, that was $2,000 more than it should have been. Yeah, right. David Moses (59:13) This is something that we have a technician that could do that. I'm not calling an HVAC guy. I'm not calling a plumber. Because a relief valve on a water heater is something that a handyman can do. Right? Not the expansion tank, but like some of these things are not difficult fixes. you can send, you know, so the idea that like you have all around, I mean, all around the country you have Craig (59:25) That's right. David Moses (59:40) tenants who are calling in work orders, you have owners who are homeowners and property owners who are not doing this. They're not just take a picture, put it into ChatGPT, and try to understand for your own sake what's going on here. And once you've done that, so yeah, we do it at scale, but we're not doing anything complicated. We're just doing what everyone should be doing. Like if you own a home, Craig (1:00:04) And can you do this? Can you do this because you've got a workflow that is not water heater specific? Can you do this for essentially anything in a house that that a tenant might report as an issue? I've got a I've got a leaky faucet. I've got, you know, name it right like. David Moses (1:00:21) There is no shortage of data that chat already has on just about every component, no matter how obscure in a house to troubleshoot it. Now, where maybe it doesn't, know, it probably, like we've thought about connecting to these massive data sources that have asset by asset. Craig (1:00:32) It's true. It's true. David Moses (1:00:46) And when I say asset, mean like a specific water heater, you know, model, right? A specific, you know, AC condenser manufacturer. And it really, it understands at a much more global level what the potential issues are in that particular asset. So it's like, you know, hey, you know, this furnace has a, an issue with burning through Craig (1:01:13) Yeah. David Moses (1:01:13) whatever part, right? And it's because of this other thing, right? And so you can attach to national databases that'll actually give you some information. like, hey, you may as well just replace this in this machine because it's just gonna keep breaking other thing that's super cheap to replace, but you're just gonna keep replacing it over and over again if you don't replace this other thing. So it's like, yeah, you can get better data, but you're way, way, way... Craig (1:01:25) Right. No, initial. David Moses (1:01:38) like 90, 95 % of the way there by simply taking a picture and going through five minutes with ChatGPT to figure out what it probably is and go from there. Craig (1:01:50) Last and we'll end the show. So it's all about adoption. We said it earlier in the show, right? You develop these amazing tools and you, you believe you have a vision for and you believe that it's going to have a tremendous ROI for not only your company, but for the people who work for you. What, what have you found? This is an amazing tool. I mean, it really just is so stunning. what have you found? like amongst the technicians. And here's the reason asked. There's a guy at Dominion, had him on the show, Steve White, and he developed the RTL underwriting platform for Dominion. The people that use that in the department now would absolutely have a massive mutiny if he said, hey, I'm turning this thing off, we're not going to use it anymore because it's changed their jobs that much. What have you found? What is the feedback that you've gotten from the technicians? David Moses (1:02:27) Okay. It's the same when it doesn't work. not so Craig (1:02:44) when they have to use it. David Moses (1:02:46) the technicians, it's really just a thank you. They didn't know that they were missing out. mean, they didn't know that they were missing out. Very rarely, we didn't have any technicians that were checking into things before they got there, but they would always forget five, 10 minutes, 12 minutes to check in or sometimes a half hour. It takes care of that, makes sure that they get Craig (1:03:00) Yeah. David Moses (1:03:08) they get taken care of and we got a thank you for that. But when the AI processing went down, my sister nearly blew a gasket. And it was really simple. was like, deprecated a ChatGPT API key and it didn't update that workflow to the new key. so it kept, like, Gemini would review all the photos and then review all the videos. But then when it came to processing all of the Craig (1:03:21) Yeah. David Moses (1:03:35) processed photos, it was saying API key is dead and delivering no information. And she nearly lost her mind because she had like, I don't understand what's going This tool that like, for everybody, like the general sentiment of AI, if you ask 10 people, seven or eight of them will give you a negative reaction to AI and two or three will give you a positive reaction to AI. but you give them the tool, right? And then you ask them, can you take the tool away? Does it really mean anything or matter to you? Like 99 out of 100 will tell you they can't live without it. So. Craig (1:04:16) Yeah, ⁓ I, I grapple at night like what we're building here. It's like, because if it's if they're that good now, what was how good will they be six months from now, 12 months from now to of what, you know, just the next rev of what you've built already versus, know, the other things that I know you're going to build and how addictive, how addicted we are going to be to tokens, man. I mean, it's just like, yeah. I know that what to your point, what you've built here does not necessarily use a tremendous amount of tokens to do what we're looking at right this second, because it's an app that just runs. It's the reporting that you get off of it that I think that we would be very addicted to without it once we get even further down the path with these tools. And if you don't think that the companies that those three companies know that, then I've got a data center to sell you. David Moses (1:05:07) Yeah, exactly. And this as dumb and as expensive as these tools will ever be. And every time we meet, that is a true statement every single time. This is as dumb as they'll ever be, and this is as expensive as they will, it will only get significantly cheaper. Would Grok come out, I think they changed their API costs. Craig (1:05:17) Right. David Moses (1:05:33) for 4.3 to like $2 per million input and like $6 per million out. They're like a third of the cost of, and this is just gonna keep getting cheaper and cheaper and cheaper and cheaper as the data centers get built out, as the compute comes online, as the chips come offline. It's just gonna get cheaper and cheaper and smarter and smarter and faster and faster. Craig (1:05:37) For a million? Right. Yeah, we live in an amazing time where what you see in your head is what you can actually absolutely build on the screen. And we've taken away the hurdles to programming. It's just, you know, if you're just speaking in English and you've got an idea, you can build something if you're just a little bit So Dave, I got to wrap it. You and I have a call. Say again. David Moses (1:06:11) And ⁓ if you don't speak English, can speak about 150 languages. Craig (1:06:18) Right? Yeah. So we're going to wrap it here, folks. That's fork.ai. Dave, you and I have a call right after this. Give me five minutes and I'll give you a And yeah, we'd love your thoughts on this episode and anything that you might like to see on upcoming episodes. Feel free to leave us a comment, subscribe, and we'll see you on the next fork.ai. Hope you enjoyed this one. See you.