This podcast episode is sponsored by ChiefAIOfficer.com, offering training and certification through the International Association of Chief AI Officers. Interested in a new career or leveling up your value in the marketplace? ChiefAIOfficer.com can help. Welcome to Using AI at Work. This is going to be a fantastic episode. I had the opportunity to sit with Joshua Pellister, founder of Real-Time AI, and also the producer of the World AI Summit, which I had the good fortune of being a speaker at in June of 2024 in Austin. Joshua happened to be in Austin visiting and working on a project, and I was able to get him for about an hour, and we sat down and had a fantastic conversation really about where he sees AI going, some of the cool things that he's doing, and the future of work. So with that, let's get started. Welcome to Using AI at Work. I'm your host, Chris Daigle. Each week we'll be learning how today's business owners, entrepreneurs, and ambitious professionals are getting more done with smart use of tomorrow's tech. Let's get started. So that's an amazing opportunity here. So anyway, um, normally I bring on guests and we have a very specific agenda that I'm looking to extract some certain information from them. But with Joshua, because he's working with so many different individuals, so many different industries, and just has such a broad understanding of how generative AI is being used in businesses, I figure um we're gonna start here, Joshua. What's got you excited about generative AI?
SPEAKER_00Yeah, so um lately, let's see. I mean, every day there's something new coming out. It's really neat. And you can see these projects sort of building over time, uh, getting momentum. The two things that I think are probably the most exciting for me are the open source community, is unbelievable. Uh, you can go on Hugging Face, you can go on GitHub, and you can find amazing uh entirely built programs, another early stage sort of nascent programs, but you can customize them, you can upload them yourself and use them. I think this is going to seriously disrupt the program for sale business and industry, right?
SPEAKER_01Tell me more about that. Like, first off, most of the audience that we have tends to be non-technical. Yeah. Do you consider yourself technical?
SPEAKER_00No, I'm not technical at all.
SPEAKER_01So you're able to do a lot of this uh early stage AI-powered SaaS, essentially, uh-huh and not be a technical individual.
SPEAKER_00That's right. Yeah. And this is the biggest shift I've ever seen in business opportunity. Uh, we kind of knew that going into this whole thing that it was going to be like that, because it doesn't take very much for you to see. Uh, you know, you see a few different things that are happening out there and it blows your mind, and you use it for your own business or something, it changes everything, right? So it's not very difficult to imagine this is going to be the biggest thing that the world has ever seen business-wise. Um, however, um, it's it takes time for it all to roll out to the right and to be developed in the right you know order. And um as that starts to happen, and it's only been like what a year and a half or something, like two years maybe max, that this whole thing started, the generative AI started happening. And you'll get now like a uh a text-to-app, you know, text-to-program, um, a fully, fully functional, uh deployed website or um or program that is custom exactly how you want it to be, for like $15 a month. You can build as many as you want, you know, costs you pretty much nothing. I got that on Replit. So if if uh I'm not associated with this company, but I'd love them.
SPEAKER_02Yeah.
SPEAKER_00Replit is uh, you know, you deploy your own websites and you put your own, and not just like websites, it's not really for websites, it's for uh apps and programs, right? Um and you put your code up there, and in the beginning I was like, oh, holy shit, like I can actually, you know, go into Chat GPT, explain a program that I want, it'll tell me all the files I need, and then it will write the freaking code for me, and it and it I can copy and paste it over and put it, and it just tells me what to do. So in like maybe three, four hours, I have a basic but working version of a program. And I was like, whoa, that blew my mind, right? That was totally crazy. But then Replit came out with this agent, and the agent has access to all of the functions and API functions of Replit. So that means that they're gonna see a lot of this coming up, right? Agent uh interaction for programs instead of user interface interactions, where you just talk to the thing and you and you give it feedback and it starts updating things. So now I can go in and go, hey, you know, I want a uh a shopping cart system, right? A custom shopping cart system where you use these APIs or suggest some APIs for me, and it will tell me what to do. It'll give me a link I can click on to go find an API key for that thing. Some of them I I buy and some of them I don't have to. It'll be way cheaper than anything else I've ever seen, and it's custom. So now I can make it do exactly what I want it to do. It's bespoke software. And all I have to do is jump into Replit, tell it what I want, it'll launch the agent, and the agent starts writing the files, naming them, and putting them aside. And there you can see it pop into a folder, boom, boom, boom, boom. And then it's like, okay, now it's time to run it. Boom, it runs it, and you can see a user interface, and it's mind-blowing. I've never seen that before.
SPEAKER_01What was the first application like where you had success? You're like, I want to try this. I you you use the tools, and you end up like, wow, this is working.
SPEAKER_00Uh for Replit?
SPEAKER_01Or just for any of these things you're talking about now?
SPEAKER_00Yeah, Coda, Coda was the first one to blow my mind, in my opinion, right? So we did this thing. Um I'm not sure if I told you this. I think we probably told you, but uh but people don't really know that we did this. When when AI first started coming out, generative AI, my business partner Peter and I, uh, we uh we didn't have the company yet, we were just kind of into it and and just so nerdy about this whole thing. It was so cool, right? Um but the thing the first thing that really blew us away is that we're like, you know what, we're gonna do we're gonna use A AI to help us launch a business in a day, right? And like, cool, from from idea, from scratch. So we're like, all right, well two, we're two dudes. We actually had four guys, but two of the guys didn't have to do anything because we didn't need their help, honestly. But we just had me and Peter really doing all the work. So we're like, okay, we started in the morning, woke up, like, cool, this is our this is our sprint day. We went on to Chat GPT. Our friend, what he did is he went on chat GPT and said, I want to make a business that I can sell to millennials. Um, here's the amount of money I want to make. What should I do? And it was like, oh, one of the suggestions was you should uh make an app that sends a text message every day with something positive, right? And they'll pay, you know, uh whatever an amount of money, like small, eight dollars a month or something, to receive daily positive affirmation texts, right? And these are these are not everybody, they're not like you know, custom texts. And so um we're like, cool, we got a name for it, like we're gonna go for it. So that's the idea we took. But the only rule How long did that take?
SPEAKER_01Minutes?
SPEAKER_00Yeah, that was that took that took ten minutes for him to do that. The only rule was that we were not using AI as an assistant, we were the assistants for AI. And that's how we looked at it. Everybody, me and and Peter both were just like, okay, it's gonna tell us what to do. We're gonna be the workhorse for AI. We're gonna follow its plan, not our plan. That was a really big important distinction because we wanted to prove something here, right? So we sat down that day to do it, and um, and six hours later we were done. Fully functioning, all the marketing written, um, the website was up, the program worked, all the APIs were connected, we had a checkout page, everything was linked together, and we had our first test, and it worked, right? So we signed up, we got a text, and every day after that we would get these amazing texts. All the texts were written with AI, everything with AI. And then it was $230 to do it. Wow. Then, um, and it took us six hours, right? Then uh two days later, somebody um posted it in this in a like a WhatsApp group and said, guys, check this out. These guys built this for it took them six hours using AI. And we started getting signups. People were like ling ling ling, they're buying it. Nice. Then a week later, we got an offer to buy the company for fifty thousand dollars. And it was just a joke, you know. Yeah, yeah, yeah. I mean, it was a great product, honestly. I love it. I wake up in the morning, look at it, smile. Yeah, yeah. A couple reasons why. One was that it was a beautiful thing to see. Sure. The other was that it reminded me of the kind of world we live in now. Yeah. Every day when I read it, I'd go, wow, this took us six hours and it's a fully functioning business. You know, impacting lives. Yeah, it's crazy. Yeah. So I um I was blown away by that. That was the first thing. The second thing that blew me away was I was doing coaching at the time. It was like coaching, consulting, and then and then uh I was kind of getting back into the world of coaching, like kind of groups, you know. And I did that forever, and and this was this was you know, me picking it back up again. And I decided to make a an uh program. You know, it's like not I say program, that's really that's like too much. General what it is. Yeah, it wasn't really that. It was like a um spreadsheet with the mind, you know, with an A on a mind. And um, I used Coda. Coda was is amazing. It's my number one tool. It's like a spreadsheet and a and a um and a word doc together.
SPEAKER_01Ballpark, what are you paying per month for coda?
SPEAKER_0015 bucks, something like that. Okay, you know? Uh maybe 30. I can't remember if I pay 30, but between not more than nothing. Yeah. I make so much money from Coda. Yeah. I mean, it's it's gotta be the greatest investment that I have is that $15 a month or $30 a month, I turned into uh over $100,000 a month.
SPEAKER_01What is Coda intended like what what's its stated intention for its use?
SPEAKER_00That's a great question. I think that what they were doing originally was just having a place for you to dump all your ideas.
SPEAKER_01Okay.
SPEAKER_00Right, and organize everything, right? Kind of like Notion.
SPEAKER_01Kind of like Notion, okay.
SPEAKER_00And then, and I think it was actually a spin-off of Notion or something at the beginning. But then they did something really wild, which I think is if you're gonna if you're gonna um uh you know get a new app right now, an AI app, and you're like, which one's gonna be around in six months? There's a there's a way you can tell, right? One of the ways you can tell is is uh does it let you make API integrations? Or do you have to wait for them to make integrations? If you have to wait for the company to make integrations, it's not gonna be around. No matter how great it is. At some point, someone's going to get annoyed. They want to they want to put these apps into workflows. They don't want to you know, they don't want to have to go between this page and this page and this page anymore. No one wants to do that anymore. It's and it's coming up now. You can see it. It's like, well, this is really cool, but I don't want to go to your user interface. I want an API and I want to be able to put it in my machine and go, hey, whenever you're done writing this thing, then go there and make a video for me, and then go here and post it on Twitter. You can't do Twitter, but yeah, Facebook, right? And then I just jump in, push a button, and it's bling and it does the whole thing. That's what I want. I don't want to sit there and you know, I don't want to work the same amount every day and get twice as much done. I want to work a 10% of the amount every day and get 10 times the amount done. Yeah. It's not acceptable anymore to just like shave off some hours. So what I did is I was like, you know, I I love coaching and and training people so much. It's like my favorite thing ever. And uh, but I I need to charge a lot in order to do it, and I don't have time to follow up. Yeah. So they're kind of floating around in between calls, like, and I was doing you know, twice a month or once a month calls to people, and they'd be totally different place when I talk to them the next time. Yeah. So here's what I did I said, all right, during our call, I'm going to uh record it. As soon as it's done recording, I'm gonna I synced my Zoom recordings and transcripts with coda, which means that now whenever something happens on Zoom, the transcript pops up in coda on a page that I made, right? And I was like, cool, that was easy. I just clicked the couple buttons, put my password in, done. And it and it pulled them all in. Super simple. Then once the transcript was there, I said, hey, if something shows up in this in this box, yeah, then I want you to uh I want you to push this button. And the button would take the transcript and send it to AI. And the prompt that I had, it was the same prompt every time, right? Because you can customize the prompt and hard code it in. It was, hey, this this person is being coached. Uh inside of here is are all these implied missions, like these implied goals they have. So I want you to extract their number one goal from this conversation. So it did, boom. And their goal is when one of the guys was, I wanted to learn how to be a barber, right? He just he just it's what he wanted to do. Like I was asking him, hey, what if you didn't have to work, you know, and do anything else, and you had all the money in the world, what would you want to do if you know for fun and like what talent would you want to develop? He was like, surprise me. He's like, I wouldn't want to be a barber. Great, you know, I'll probably never need you.
SPEAKER_03Yeah, yeah, yeah.
SPEAKER_00But um, but he should do that. He was scared to do it. So I said, Cool. Next thing that happened is it took that that goal and I said, Here's the transcript, here's their goal. Now I want you to make 15 different micro steps that get them one major step closer to their goal, right? Not all the way there, right? Yeah. So it made 15 micro steps, and then I said, Cool, now turn all 15 of these into text messages, short ones, and say, Here's your daily mission. And then I said it did it, and I just made a system. Every day it would change the number, like every day it would count the number of days that they had joined, right? So every day it's like, you know, nine, ten days, eleven days. And and it would read the number, it would look at the list, it would find the number of the text message, and then it would, with Twilio, which is integrated right in the coda, it would just send them a text message in the morning. And then they got this thing that said, Hey, you here's your ultimate goal, right? Very valuable. So valuable. So I'm I wasn't selling uh coaching forever, right? So I came back into this and I wanted to have a trial group. Uh-huh. So I lowered my price a whole bunch, right? Just the way that I can do it. So I was like, okay, I'll for a thousand dollars a month, I'll coach you. I'll have one meeting with you, and sometimes I had two because I was testing that out. And then um, I'm gonna use AI to follow up with you. They're gonna send you text messages, but it's gonna be your mission. Yeah, it's not gonna be some random text message like, hey, buddy, have a don't have a good day, have a great day.
SPEAKER_03Yeah, yeah, yeah.
SPEAKER_00And like it's gonna be something that's gonna help you get what you want. And this blew my mind. First off, guys were like, great, they didn't care at all. They didn't care that it wasn't me texting them, they just wanted something to say. The accountability or the structure, yeah. To keep them on their track and everything. And once a month they got to talk to me, right? So it was effective.
SPEAKER_03Yeah.
SPEAKER_00But it but what really blew me away was how much progress these guys made. They were unbelievable. They were the best students I've ever had. They in between calls, the next time I'd call them, they'd be like, oh my gosh, you're not gonna believe what happened. I did what you instructed, the missions that I got. I was in the middle of doing my mission for yesterday, and someone asked me if I wanted to be an apprentice to them as a barber. I had to not go to school, I got to skip the entire thing, like, whoa, he was blown away, and he's now apprenticing to be a barber. And and I I could I didn't have to do that. And if I just had that one meeting and told him, you know, you want you want to be a barber, right? He would he would be like, cool. And then he'd wait a month and he'd be like, I didn't do anything, you know, I didn't go. Yeah, but this kept him on it. So um that was a thousand dollars a month. They were stoked to pay it. I probably could have charged a lot more than that. Yeah, yeah. But anybody can do this now. If you're if you're a coach, all you have to do is get a coda account, figure out a couple of things in coda, and then set your set up your own system and and do that. You have your meeting, it dumps in, you push a button if you want to go review it, review it. I was in the morning, I was reviewing it all, and then that's it, man. Send send out the text messages to everybody automatically, and it worked amazing. So there's a bunch of stuff like that that started blowing my mind, right? Then I made a Josh bot where I was like Josh Bot had all of my knowledge base. Yeah. So everything I've ever written, all the all the transcripts of all my coaching I'd ever done, um, both in AI and in um in attraction and dating stuff I used to do a long time ago. Um, that I started picking up again because now I can write books with AI. So I'm like, I've always wanted to release this book about this thing. So I did that too. It's super duper fun, right? So I grabbed all my old content and have all these guys in a in a group, you know, thousands and thousands of guys that are interested in talking to me and asking people questions, but I don't have time to answer all their questions. And like I'm actually more excited about AI than anything else, and I don't really want to be in there talking, right? But I want them all to get their questions answered because if they if I don't answer their questions, they're gonna go elsewhere, and that advice is going to be maybe nefarious. Yeah. And I want to actually guide them. So I was like, cool, how about this? Um I'm in here. If you don't hear back from me, um just uh just click this link and talk to Josh Bot. In the beginning, they were like, okay, you know, I'll I'll check it out. And they would they found some bugs and told me about it and I would fix it, right? And then after a little while though, uh I remember the first time this happened, I was like, a guy had a question. He's like, you know, hey, what's wrong with this? Like, what am I doing wrong here? And it was a complicated situation he was in, and I didn't see it. And the members are I was I was not present and I was out doing other stuff. And um, I was out doing AI stuff, right? Yeah. Uh and also, of course, we have an amazing AI training company now that that's where I was spending my time, right? Yeah. So I was I was like, oh, I didn't see this, right? I it showed up. It was like two days later, and this guy had no idea what to do for like two days, right? So again, I'm like, oh I gotta answer this question. I was stressed, and like, I mean, I only have a couple minutes to do this, I jump in and I see a thread, and the thread goes, um, hey, I have this situation. It was long, and just to read it was like too long for me to read it. And then I saw the comment underneath it said, Um, uh, hey, why don't you try Josh Bot, someone in the someone else in the group? And he goes, Oh, what's the link? Here's the link. He clicked the button and he goes, Oh, great. It perfectly answered my question. And the guy was like, What did it say? And he goes, boom, and he posted as a comment, and I was like, What have I done? Yes, like I actually took all my expertise and I gave it away with a link and it worked. Yes, it worked, and it worked complicated stuff. And it was only about two or three weeks later that I checked my my API usage for that, but for that bot, and I was like, Whoa, someone is going crazy in here. They're like, you know, talking to this thing all day, every day, and spending like $40 a day, right? And I'm like, that's that's a lot of API calls, yeah. Yeah. So I so I um reached out. I was like, hey, I'm I love that you love the tool and Josh Bot. And like, uh I see you using it all the time. What are you using it for? And he goes, Oh, I'm an investor, and I just put all my investing ideas in there, and then you tell me what to do. And I was like, dude, maybe you shouldn't do that. Like, you need to you need to probably not use my knowledge base for your investments. He's like, no, no, actually, most of it's about mindset, and I just I I get I get weirded out about my investments, and I I want to pull out early, or or I want to jump in with FOMO, and and he's like, You calm me down and you get me straight. So he was using it to do his job. And he would use make reports and come up with strategies and send them off to his clients, and he was doing his whole work in there, you know. And I thought that was so hilarious. So I was like, hey, you know, maybe don't do that, right? Like I don't want to be on the hook for you deploying making a bad decision with investments. Like, that's the one thing I used to be that hedge fund uh manager, and so I I'm like, um, you know, that's really cool, bro. And maybe don't do that anymore.
SPEAKER_01This is awesome because for anybody that's out there that has uh domain expertise or subject matter expertise and they're known for it, and people are like, oh, if I could get five minutes with Chris at the cocktail party or whatever, they don't have to wait for the cocktail party anymore.
SPEAKER_00That's the thing, though, is that our entire uh you know funnel, sales funnels for coaches and stuff is based on the cocktail party.
SPEAKER_03Yeah.
SPEAKER_00Right? So if I if I'm gonna get clients, then I need to not give away my stuff for free, right? And that's that's true. I do need to do that. Um if I make a bot that you can talk to anytime you want, it's gonna devalue all of my coaching, right? That's what everyone thinks. But it doesn't. It raises the value of your coaching like crazy. Because of access? Because of access, and because the wall they hit will get at some point, they they'll get all their answers, their questions answered from your bot, whatever that might be. And then uh they want to talk to you. They want you to help guide them through something. They don't want the the bot, right? It's just a preference then. It's not even like a necessity anymore. Yeah. They would rather pay you. Now, if imagine you had a money-making business. This is these are the people who do success in money making are gonna are gonna explode soon. They're gonna be doing killer with with their businesses. And here's why why I think it's gonna happen. Imagine that I am broke and I want to buy your money-making course. If I have money, I don't want to buy it. Yeah, right. So that's the always been the annoying thing about that industry. Yeah, however, now I can go, oh, I'm just gonna jump on to your bot, which you let me have for free. If I'm subscribed to your newsletter, I get it for free. So I subscribe to your newsletter. By the way, all your newsletters are written with this AI, right? Because it's all your knowledge base, right? Yes. Um, you just kind of look at it and scan through, make sure it's good, or have someone do that. Yeah. I read the newsletter, it reminds me to jump on the bot, I get on the bot, and I start asking questions. Then it answers my questions and helps me with a method to make money. Now I have enough to pay you and I can buy you. I can buy your time. I can buy your consulting, and I don't want to do that. If you have a mastermind group, I want to join it, you know, because I want to keep going and I want to have connections to other people. So these are the things that are always going to be valuable and they're going to become more and more valuable over time. Whereas your knowledge base is going to become less and less valuable over time.
SPEAKER_01So let's talk about that so that you don't find yourself in a situation to where people are like, oh, I don't need to pay. Because it seems like, oh, if I'm giving everything away, and it's there's intelligence behind the answers. It's not just like here's a bunch of FAQs, but here's a customized response to your scenario that the individuals might feel like that's the all the value that they needed. What but that's not the case.
SPEAKER_00No. This is one of the biggest misnomers in marketing is that um if you give away your best stuff, that people won't want to hire you. But the truth is that when you give away your best stuff, people always assume you have other stuff behind the scenes that you're holding back. Your but your real best stuff is actually somewhere else and you're keeping it, and the stuff they see is not even the best stuff. So imagine how good the good stuff must be if the if the stuff that you give away for free is already worked. Yeah. It proves your your authority and it shows people that you have uh more.
SPEAKER_01And karmically, you're like hooking up the world with great info.
SPEAKER_00Yeah. And so, but you do want them to do something in order to get it. You don't want them to just like have access, here, share this link, do whatever you want. Because the value, then even if the bot goes down and your information goes down, they won't use it. Right. Yeah. And you'll just pay for all these API usage, right? Yeah. Um, but uh but you can do that. You can just have a newsletter. Say, this is free to everyone who signs up for my newsletter. All the newsletters are in there automatically. All of my books, I have over 113 books I just realized or products, right? That I've written over the years. I used to have a publishing house too, but they published almost exclusively my content. So it was all this stuff was in there. And I'm like, okay, you can get all that stuff. I'm not even gonna say how much it is. You just you get all this, you can ask whatever you want, it has that entire knowledge base. You just need to sign up for my newsletter, and I'll give you the link. They sign up for the newsletter, you're like, hey, cool, thank you so much. Here's the link. Uh enjoy. It's in a members area, right? So they go in the members area and they access it. Done. And while they're in there, they interact, they ask questions, and at some point, they'll everyone's gonna have every question they've ever had answered. That's and it'll be answered whenever they want. When Google first came out, yeah, people were like, What? Yeah, like I'm never going to XYZ again, right? And now we do, we we Google everything, right? Most people Google everything. And it's gonna be like that. Where um it's a deeper level of I have access to all the information in the world now. I'm gonna get it faster, it's gonna be catered to my use case and my context, it's gonna be exactly what I'm looking for, right? It'll be step-by-steps, it'll be able to take actions, you know, for me. So soon then all that stuff is gonna be free. And the world is gonna see it as not valuable, right? Having access to all the information in the world is gonna be normal. Google's free, right? Yeah. You just jump on there and you ask whatever you want, and it'll tell you. Uh it's not always true, but it's pretty good. Same with AI, right? You're gonna want at some point, there's always gonna be validation that you're gonna wanna have. But besides that, you're gonna have access uh greater than any other time in history. It's gonna make the Library of Alexandria in your pocket, right?
SPEAKER_01My mind is spinning here. So so if if that becomes commoditized, right? Like if I can go, oh, everybody does this, everybody has their Josh bot, right? The point of distinction could be what you did with Coda, where it was like based on this particular session, the user session, it got that information. It was like, oh, by the way, Josh, Josh is gonna follow up with you, you know, throughout the month to make sure that you're doing what you're supposed to. That would be something that because our a past guest and uh um somebody that we do a lot of work with is Jody Cook, and Jody has a business called Coach Box. And Coach Box is it's for coaches, it's for it's the it's the natural language interface, AI powered, Gen AI powered, for coaches to introduce their subject matter into it and their IP so that their clients can have their Josh bot experience, right? But this idea that you did with Coda, which isn't, I would imagine, like not terribly complicated, you just have to be able to think that way. Yeah, not complicated at all. That's what made the yeah, it's great. I can talk to Josh once a month. In the meantime, I get the perspective that I need. Oh, and by the way, what I really wanted was accountability, uh-huh, and that's what I'm getting, right? Because that's what most people want with, they want the advice and knowledge with coaching, and but they get results when they know I spent money on this, or I don't want to be embarrassed because I didn't do what somebody that I respect suggested that I do.
SPEAKER_00They'll fall off because of that. Yeah. I mean, just the shame of not of not accomplishing something that you helped them set up perfectly and their resistance will make them make them fail.
SPEAKER_01And if you've got this automated mechanism that is just checking in with them, uh that's all like that's probably enough for a lot of people to end up getting results that they wouldn't have gotten without that one automated step.
SPEAKER_00Yep. Pretty awesome. And this is like uh the you know, I had to build a tool uh uh you know about RAG, right?
SPEAKER_03Yeah.
SPEAKER_00So for for people who don't know about RAG, RAG is retrieval augmented generation. It's this uh it's this way to um take a knowledge base and then let the AI grab stuff from it by searching through in a really smart way, searching through everything that you have in that knowledge base and pulling out just the pieces that were relevant to your question. They pull that into a big pool, and then the AI then looks at it and goes, Oh, here's the answer in here somewhere, and it answers your question. And it's the future, 100%. We're gonna see rag and everything because it's uh the knowledge base, it's not about uh people don't really want a tool that answers every question, everything. What they want is something that does one thing in one niche in some area that they know is trusted sources. Because if you tell me here's how to you know fight a legal case and you and you give me all the instructions and I don't know any different, then I might go, that's really great. But I also am gonna be nervous about you applying it because if I apply it and it's wrong, I lose the risks are really high. So what I really want is for a lawyer to say, here's all the case law for this particular thing, um, upload it and go, This is you know, this this is non-negotiable. This is a thing that's real. This is a case law, these are the laws. Um so now would you have a question? I want it to look at the laws and say, this is illegal, this is not, and why, right? So to help build your case, I should be able now to take a if I'm being served something, right? Someone's trying to sue me. They should be able to send me their paperwork, whatever it is, and I should be able to take that, input it into my AI, push a button, and now it reads all of this trusted source data with RAG, grabs out the information, it could be thousands of pages, grabs out the right information to answer just my query, pulls it in, and then answers my thing, right? And to give you an example, I had to use this. I had some not my current landlord, but a previous landlord that I had was just being a jerk. Like he he was trying to get me to uh to pay for something that I didn't have to pay for, right? As it was not in my contract. But there was some vague verbiage about it, right, in our lease. So he was like, Oh, this is look right here, this is what it says, and blah blah blah. So I took the tenant landlord code and I took our lease and I put it into the AI. Yeah. And then I took his email and I copied it and I said, Here's what my landlord said to me. Give me all the reasons why this is wrong and illegal. And it just boom, boom, boom, boom, boom, boom, boom. And then I said, Okay, write a really nice email to my landlord that says, um, in no uncertain terms, that this is how it's going to go down, and and ask him if he wants to do it another way. And it wrote the sweetest email ever, and uh, and it had it was like, you know, according to this statute of this thing, um, this is the thing, and here's the other cases in which this happened, and here's the results of those cases. And your cases, I would say, falls in this category. Would you agree? I think so too. And um, as a result of of um of basically going after me, here's what you're gonna lose. You're gonna lose, you know, not only this, you're also gonna put your property at jeopardy and everything else, right? And um, and then it said, Do you wanna do you wanna try this another way instead? And um about two hours later, I get an email um with five words. And it it said, Don't do this to me. And that's it. And I was like, Okay, great. Like let's let's just not let's just not worry about it. Like we'll split the costs and be done, right? And I got to be a nice person instead of being a jerk about it and like having to be vindictive and crazy and resentful. And so um he was like, Whew. Like after that, he did not mess with me about anything ever again, you know. And I think that that is the superpower. I can actually do that, and I only do it because it's using the information from from the from the code that we're supposed to follow. So trusted sources are going to be a major part of the next wave of AI.
SPEAKER_01And you know, if I so almost like specialized small language model environments where it's just contract law in the United States, uh huh. It's just maritime law and you know, whatever.
SPEAKER_00Or or it's uh, you know, business advice, uh-huh, right? Um, or whatever it might be.
SPEAKER_01Based on McKinsey's way of doing things. Yeah, exactly. Or whatever, yeah.
SPEAKER_00And and you should be able to select the reason why so I built a tool that did this, right? This is like a year and a half ago that I started doing it, and uh called it the librarian. And the reason why I built it was because I was like I was thinking to myself, well, what do we all have, right? We we really do actually care about our own trusted sources. We really do, and you can prove it because you have a library at your house of books. And maybe you have a digital one, maybe it's not digital, maybe you have a physical one. Most people I know that are especially business people, type A people, you know, they all have a bunch of books. Yeah, and they're all in their library. They keep those books. Why? Why don't you read a book and then throw it away after you're done reading it or give it away? The reason is because you might need it again, right? And a lot of this stuff you'll forget because you're doing your life and you have other things to think about, right? So you keep those books because you trust them now. You read through it, you vetted it, and you like it. So that's not gonna go away. Every every single person has a different library. You have a different library than I do. And every single person, no matter how similar they are, they don't have the same books. Yeah. And this is where it's a fingerprint. You know, your your trusted sources are your knowledge base. Interesting. And if you could, you would have way more books, probably, with a lot more trusted sources, but you just don't have time to get them all and to buy them all. Yeah. You know, I wanted to have a whole library where you walk around with a freaking ladder and go get the books down. I really wanted that. That was like my whole thing. But what I did is instead, I mean I have something like that, not a not a ladder, but that'd be cool.
SPEAKER_03Yeah.
SPEAKER_00But um, but I have uh, you know, a library at my house. I found all the digital versions of all those books online. You can search, you know, PDF, file type PDF inside of Google and find so many things, right? So I just downloaded them all and uploaded them to my librarian tool. And then I was like, I was blown away by what I could do. I I the first thing I did is I said, I have to write an email to and to an audience, it's a it's a promotional email, like a marketing email. But I don't like being markety, and I want them to um get a lot of content from it. I also need it to like do all the things that good marketing emails do so that they actually want to keep reading my emails. Like, you know, I love this uh this method by a guy named Andre Chaperon. You know? Yeah. And he had something called autoresponder madness. I love it so much. It's my favorite style. It's a storytelling and it goes from email to email. They call it the soap opera sequence. Yep. And um, but I I was gonna I was gonna have to sit down and read that book again and and find all the secrets again and then structure a template and then, you know what I mean, and write around my stuff. I also wanted to write about the 33 strategies of war as they pertain to attraction and dating, right? So I'd have to read that book too. Yeah. And then I and then contemplate mapping it all together. Yeah. I'd have to build some kind of a system, right? And then and then show it to everybody. And it would take forever for me to do that, you know. So I said, all right, here's autoresponder madness, here's 33 Strategies of War, and here's my book. What I want you to do is write using only the techniques in autoresponder madness for your writing style. I want you to uh write an email about 33 Strategies of War as it applies to the readers of my book, and I want you to speak to people who are the readers of my book, not to the readers of the 33 Strategies. And it blew me away. It was like, okay, boom. Here's the subject line, here's the email, and it was the best email I'd ever seen. I I've never written an email that good. It took one of the principles from the 33 Strategies of War, reframed it into dating, and then said, you know, you're basically your you are your worst enemy, kind of strategy. One of the strategies of war is uh to fight yourself, right? Fight your ego. And and it wrote in a beautiful email. And I was like, huh, write the next one. Because at the end of the email, it says next, in the next email, I'm gonna teach you this. Boom, next one done. And I was blown away again. Now everyone's got different writing styles, everyone's got different, you know, structures they'd like to use for their emails, for example. I just cranked out 50 of them in about 10 minutes, and then I scanned through them. I was like, Does this look right? Does it look right? All the information was great. And I just put them into a large doc and sent it over to my marketing homies, and they loaded them up and they went out. And it was awesome content. The people were just like, guys were just waiting for the next email to come, and I would love to do that with every single step in my business, with every single part of my business, but I couldn't before, you know.
SPEAKER_01So with this librarian, it's basically it's using the AI brain, but with a select set of context that you feed it. Yeah, exactly.
SPEAKER_00Yep. And I used to explain it like this like when you put when you put content into Chat GPT, you upload a file or something, what it's doing, because they have uh so many regulations, looking people looking at their bit at their business, yeah, yeah, yeah. Always hawkeyeing everything that they're doing, they have to um control the narrative of whatever's going on. Yeah, they can't let you just produce whatever you want, you'll produce literally illegal stuff. So um, you know, you can go in and be like, hey, how do I make LSD? And it will go, I can't tell you how to make LSD, right? Yeah. So But I know how to. But yeah, of course it knows how to. It knows everything pretty much, you know. Um it can explain a way, is it real? You won't know unless you try it, I guess, or like you know somebody that does it or you know how to do it already. But but it tries its best, you know, and it's it's very, very good at everything else. So um so they do that. They make it so that you can't ask certain questions. When they did that, uh, they also made it such that if you upload a file, your file, I explained it like this, it will bend your uh it'll wrap your file around the LLM. Meaning the AI itself is going to be core and exactly the same. It will make your file like a like a nice hat it can wear. It won't fundamentally change anything about the LLM. Whenever I built my system, it was as if our files were the core and it wrapped the AI around it instead. So now the AI is looking inside, going, what's in here? Oh, I see what's in here. My job as an AI is only to report the truth of what I see. It's not to determine the truth.
SPEAKER_01If you're enjoying this episode and want to learn more about how to start using AI at work, we've made it easy for you. For just one dollar, you can have full access to the Chief AI Officer community, which will give you additional training, custom software, daily training calls on AI tools, using AI automations, getting more from your Chat GPT sessions, and the business of being an AI consultant. Simply go to chiefaiofficer.com forward slash insiders to accelerate your AI journey. Now, back to the episode. So you built this? Yeah. With and you mentioned uh early order in uh your enthusiasm for open source. Yeah. So is this an open source model uh?
SPEAKER_00You can use an open source model with it. I use um so there's a there's a couple of open source systems. One of my favorite ones is a company uh called Grok. It's G R O Q. Yeah. There's Grok also, that's the the Twitter Twitter AI, which is gonna be the one of the best AIs in the world, I think, very soon, probably in next year. Uh it's gonna give a run, completely a run for everyone. Like everyone every other AI is gonna have a hard time catching up with you know, he likes uh not censoring very much. Yeah, he likes free speech a lot, which so it's gonna when he releases his API for that, a lot of people are gonna switch over to that. And um because they can won't censor their content, right, is in a way. So if it disagrees, and one of our clients uh uh got in trouble for a little while from from uh uh the media chase after him for something, that they were all wrong about it, but they did it anyway, and uh it tried to ruin his reputation. And um, if you put his name into the AI, it'll go, I'm sorry, I don't feel comfortable helping you know talk about this person or whatever. And then you're like, screw you, buddy, like you have you can't tell me who to trust and who not to trust. Yeah. So that's gonna make Grok with a K more valuable. But Grok with a Q is a chip company, and they make these speed chips for AI for large language models. And they started hosting them in in racks and then letting people come in and you know buy access to them. Yeah. And it is the fastest thing I've ever seen. It is, you put something in, it's almost like as soon as you hit enter, you have 10 paragraphs of an answer. Like instantly. It's the fastest. You've set this up for yourself? This is already existing. Okay. I just had to plug into the API. Wow. And it uses open source large language models. They're not an AI company that makes AI, generative AI systems. What they do, sorry, uh LLMs, yeah, they're um instead, they just make the chips. So they find they process it all. So and they're lightning fast and they're really cheap. So that's another thing. So all the open source stuff, like I have an I have an open source model, um, Llama, right? Yeah. One of the Llama models on my computer. I can go, you know, in on an airplane with no internet, or whatever I am in the whole world, I can say, I want to do this and it'll help me do it. I also can go, hey, I'm outside right now in the woods, and I don't have internet. I can go, I'm outside, I'm stuck in the woods, I don't know what to do. Uh here's the kind of stuff I see around me. Help me identify the plants I can eat and help me make shelter. Tell me what to do, and it will walk you through everything. Right? You don't need the internet at all. Yeah. It's like the ultimate survival tool.
SPEAKER_03Yeah.
SPEAKER_00But it also can help you do other stuff. You know, it can help you come up with strategies, you're gonna have to do the strategy for now, right? But it's only a matter of time before you don't have to do the strategy either. Where you say, hey, I want to I want to plug into this API or I want these systems to be here. I think APIs are gonna be major in the future, by the way. They're gonna be really, really important in the near future.
SPEAKER_01Because people will be able to come up with an idea and go, oh, this is one of those ingredients, this is one of those ingredients, and they'll be able to hodgepodge everything through the APIs.
SPEAKER_00Yeah. Okay. So, like, for example, I saw a kid uh just the other day on Twitter, and he's like, hey, we we're not it's called iX-ray. We're not releasing this information because um we only did it as a project for fun and because it's a security risk. But you know those glasses with little cameras on them? They're like look like regular sunglasses. Yeah, yeah. He uh he um got a hold of the live streaming data that was coming through that, that he then plugged into an API in his phone, and it scanned people's faces for facial recognition and found their social media and all the information about them while he was talking to them. So he'd look around and it would be like, here's this person, boom. Look at the phone, because he didn't have display yet, you know, in the glasses. He'd look at his phone and it would tell him, Here's it, here's what this person does for a living, here's where they work, here's any all this information about them. You know, you could do a background check, here's are there crimes that they've been committed, uh, that they've committed or been convicted of, or whatever it might be. Yeah, any any data they want about them, and and they did a test. It looked like regular sunglasses, right? They're not. You would never even notice there's a camera in there. It's flat. It's totally different, right, than anything I've ever seen. And uh they did a test where they walked into a subway in New York, I guess, and he was he walked around and he saw somebody, took off his glasses, looked at his phone, right, saw who that person was, and walked up and said, Hey, is your name Dave? Do you work at this company? And the guy was like, you know, he's sitting like this, and he goes, Yeah, that's me. Yeah, what's your name? And he started getting up and shaking his hands. Like, what can I do for you? Like, how can I help you? You know, and he goes, Oh, I just I love your work and I love what you do. He says, Oh, cool, you should stop by the office sometime. Like, he thinks he's being recognized by somebody. Somebody who's actually interested in what he does, right? This is gonna disrupt everything. Yeah, also I saw him approach a girl, right? Hey, uh, do you by some chance really like poetry? Yeah, actually, that's what I do. Like, I think I know your poetry. Like you had a poem about dogs, and it was this poem. She's like, Wow, you you know my poetry, right? And I saw that and I went, Whoa. Yeah, whoa. Instant rapport. Instant rapport. And you know, everyone was like, this is really scary, right? But I was like, this is amazing, right? This is amazing.
SPEAKER_01I instantly think about going to an event and sitting at the event, networking. Like, what do you do? That's gone. Like, just scan the scan the room, target, target, target, target.
SPEAKER_00Done. There's your networking. And you think about it because you're a good person. That's what you think about. But if you weren't, you'd be target in another way. Yeah, that's true. And that's the that's the hard part. Yeah. He did it with API calls that that a teenager can get a hold of. Wow. And it was all instant on his phone. Pretty amazing.
SPEAKER_01So one of the things that you've mentioned a couple times, and I'm very interested in it because I I I truly think that there is a lot of runway in the agent environment. Yeah. And to the point to where like we've got some services and things that we do for companies. And rather than having to train up a whole army of chief AI officers, if we can get agents to fulfill some of that, then there's less need for the humans to get involved, less human error, and those sorts of things. But you know, all the tools that I see there, they they call themselves agents, but it's questionable if they're really able to have agency and like make the next decision. What's your experience with that?
SPEAKER_00Well, you know, the word agency is is kind of loaded because it kind of assumes consciousness. Right. So that's a that the word agent itself is also kind of that, too. It's based on that route. So um I think that that's a cool rabbit hole to go down, but that's probably not really what you're asking. I think what you're really asking is can they do stuff and make decisions, right? And if you program them to make decisions how you want to, say you have if then uh directions to go in, um, you know, there's a there's a system, um, geez, am I drawing a blank on this right now? It is a tool selection system, Langchain. It's called Langchain. And they're a whole open source community as well. And it gets better and better. What it is, is it's an internal thought process for the for the AI. And it goes, first, do I need to use a tool to do this? And it asks itself, it doesn't tell you that that thought, right? And when I say it asks itself, what it really does is it prints that in the code, it says, Do I need to use a tool to do this? And that's the prompt first, and then it answers just that prompt question, and it doesn't display it. And then it goes, yes, actually I do need to use a tool to do this. Um, which tool do I need to use? And it says, here's all the options. And it goes, well, I know all about these tools because I have it all written right here. That's the best tool to use to do this. How would I use this tool? Oh, here's what I would do, right? I would first, you know, log in, then I would write, you know, open it if it says Gmail. I would uh, you know, create a new message, a new email. I would search through first actually, maybe all the previous emails I wrote to this person, and I would find how I'm talking to them, and analyze it, come up with a strategy and a style for talking, and then I would create a new email and I'd write it about this thing, and then I'd hit the button and send it, right? That's what I would do. So then it goes, cool, well you have access to the API of Gmail, do that now. Now it goes, I'm thinking, I'm thinking, right? And it goes, done. And then your email has been sent. When uh a year and a half ago, Langchain and Llama kind of Llama had its own Langchain, it was really bad, but they had a Gmail integration, and I built it into my librarian just to test it and see. And I could get this was a year and a half ago, keep in mind, right? No one was doing this at the time. I could um go in there and say, hey, give me the list of the last five emails that I had that were personal, that were not this. Boom. List them right there. Like, wow, that's already crazy, right? I don't have to go in my email. Yeah. I have five email accounts, I can pull them all right here into the chat. Cool. Now then I said, okay, um uh write a draft reply to each of these emails, and it would go, and it wrote a draft reply, and then I said, I could say save this as a draft in my email, or I could say send it, and it would go, done, sent. And then I went to my email and I checked, it hadn't sent an email, it was in the sent folder. And I was blown away. I was like, I cannot believe this. So when you say agents, when people say agents, they they really mean that the AI system has a connection to an authorized connection, right? To an API of some sort where it can take actions and do things, right? It can go search the internet, it can um send an email, it can send a text message, it can read things, it can get access to information. You have to let it have access to that information. You have to give it the codes and logins and all that stuff and tell it what to do. You also need to, you know, tell it which tools you want to get access to. But once you do that, you can absolutely then say, okay, um here's an if-then, you know, split in in my decision-making tree. And I what I want is I want you to read this thing, and if this happens, you do this thing, and if not, I want you to do this thing, and it will do it. And then you just turn it on and it will scan. Or you can turn it on or push a button to turn it on and it will what tool would you be using to do this?
SPEAKER_01Coda. Coda?
SPEAKER_00Coda. I I can do it with coda without knowing any code, you know. You have to learn how to how to write the functions in coda, um, which is like spreadsheet functions basically. They're they're not they're not serious. Yeah. And the cool thing is that I don't really know how to do any of that stuff. What I do is I go on Chat GPT and I go, I'm in coda, I'm trying to make this happen, and it writes the codes for me. And then I go back and paste it in there and I'm done. Yeah. You know, and it works.
SPEAKER_01So the part that's missing, so anybody can do the mechanical parts, but the part that's missing from other people doing this is the thought process of architecting this. Yeah. Oh, I need to do this, I need to do this. Is there a framework or what's the secret there?
SPEAKER_00Yeah, actually, my my business partner, Peter, kind of taught me this. He was like, he's very simple when it comes to frameworks. He's he breaks things down really, really well, right? You know. Yep. And um, and he goes, Okay, you just want what inputs do I have to put in, what uh outputs do I want, and then that's all I need, right? And so you have like a black box. You put in these things, and then inside the box transformation happens and then outputs these things, right? So if you know the outputs you want, and you know the inputs you want it to use, all you have to do then is say, Hey, take these inputs, and you write a prompt, right? Take these inputs, here they are, and you know, wherever they are, and you'll say, Um, my goal, the output goal is this thing. You could just say that, and then it will probably work. It'll do a really good job. You can hit the button and it'll take the inputs and transform them into the output. You don't have to say, hey, the objective is this and that and that. You can just literally say, Here are the inputs. Well, the output I want is a business marketing plan. And then, and here's an example to follow. Done. Now it takes that, it looks at the example, it looks at your inputs, and it transforms them into that, and you're done. So you there's no architecture required. No. Then you have a bunch of transformations. You can take transformations and stack them, yeah, right? And they'll go left to right and they'll work. Yeah. Like for example, I wanted to make something that would check uh comprehension. Like you're talking to someone, it's a chat, and you want to know, did they actually learn anything? And think about this. Like you went to college, right? Ish, yeah, be too ish. I went for two years and then I stopped going. I stopped paying at least. I kept going for a little while for free, and that was fun. But I didn't do get a degree because you know I wasn't paying. And and I was I hated college because of uh I loved it too, but I hated it because it was not testing my comprehension or my application. It was testing my memorization and my conformity. And they were testing for conformity in college. And I hated that. I was like punk rocker when I was younger and I didn't want any of that stuff, like get away from me, you know? Yeah, yeah. And um, and so I left college and didn't want to deal with that stuff anymore. Uh but now, you know, you have if you have kids, you're you're like, I want to know, I want to train, actually teach my kids. How do I really teach? You can't teach for memorization, you have to teach for comprehension. So what what I did is I said, hey, when when you're talking to somebody, this app, this this is I'm gonna oversimplify it, right? Say you're in a chat like Chat GPT, and it's teaching you something, like a teacher. And then it goes, Do you understand this? And you and you go, Yeah. And then and then uh it says, well, if you understand it, then I want you to, you know, tell me what I just said, right? Yeah. Well, I mean, all you do is look up there and read and say it again. You don't have to understand anything. I there's a you know, there's all sorts of apps you can now take a picture of your screen, yeah. You're taking a class, and it'll tell you the answer right there. You just choose the right button, right? So it's it's gonna get harder and harder and harder to actually teach kids because the memorization was was the strategy. So what's gonna happen next? We're gonna start testing finally and teaching finally for comprehension. So I made an AI tool that would check for comprehension against a rubric. It would say, here's all the different levels of comprehension. Here's their answer to this question. Uh on a scale of one out of 15, uh, how well do they uh comprehend? And it would say, here's how well they comprehend. If the score was too low, it would say, it would say, okay, the score is too low. Um what I want you to do is repeat that stuff, but the parts they missed only, and go teach them the correct them and tell them exactly what actually is accurate. So then it'll go, that's really close to the to the the way that it is. That was really good. The only thing that's different is that you know, instead of the engine uh turning, you know, a magic crank or or gear, the engine's turning uh, you know, a piston or something, you know, whatever it might be. And and it's and it goes, and here's the difference. Does that make sense? Yeah, that makes sense. And then it'll go, okay, well, if that makes sense, then tell me how you would apply this to this, this, this, and this thing. So it has all these different tools or t techniques for comprehension. And they're amazing books on comprehension, right? But I don't want to read them. What I want is to upload them and say, here's all the stuff about comprehension. Use this super genius comprehension stuff and make me a really good comprehension thing that now will will guide and change as I as I grow. So um, so that's what oh I built one of those and and we patented it actually.
SPEAKER_01So on Librarian, is that commercially available?
SPEAKER_00Uh Librarian?
SPEAKER_01Yeah.
SPEAKER_00It is now we're going back to something different, but um, yeah, Librarian is uh under uh under the name IP Vault.
SPEAKER_01Okay.
SPEAKER_00You know Perry Marshall? Yeah. Perry Marshall partner partnered with us. Um he's a brilliant marketer. And uh and he was selling early access to this thing uh as well. And we're gonna release it. Ultimately, though, that that is about making Jarvis. Okay. Right? Like Jarvis, for people who don't know, is like a they probably all know, but yeah. Um Iron Man. If you're a dude, you definitely know. Yeah. It's Iron Man's AI. And what I want, I I was like, what can we do to make it, right?
SPEAKER_01And what I realized And it's personalized for everybody based on what they Yeah, of course it is.
SPEAKER_00Their library. Yeah, your library, exactly. Your trusted sources are your fingerprint, and you will change them. It's your agency to change them.
SPEAKER_01Now this is interesting because it without that distinction, it would be homogenized. Everybody would be even though I'm this way, the output that I'm getting from my tool isn't as calibrated to me as it would be in a like a librarian environment, an IP vault environment. So that would actually, even though I thought I was a unique snowflake, my responses would be fed by something that everybody else had access to.
SPEAKER_00Well, how are you gonna get an edge now with anything in life, if that's the case? So everyone would be exactly have exactly the same uh opportunities, same knowledge, everything else, right? But that doesn't create uh evolution that creates equity. Yeah.
SPEAKER_01Right? So if somebody wanted to test out something like IP Vault without building it themselves, is it like is it available?
SPEAKER_00IP Vault? Yeah, you can go to IPvault.ai and check it out, right? Um you can do that, and if you want to test something that's kind of simpler than IP Vault, but it's a similar concept, it's a pain in the ass to use it, um, in my opinion, because they don't let you just bulk upload all your files and then stay, keep them and select which ones you want to talk to. And it's like you pick it, you pick your files, there's an environment, it's called a notebook. Yeah. You know what I'm gonna say? And then you have in that notebook, you have all those resources, and then when you go in there, you can talk to it. You can't change the LLMs, you you're stuck with their LLM and stuff too. But that's called Notebook LM. And Notebook LM is like a version of this. It's a Google product. It's a Google product. Yeah, Google bought it or built it, I don't know. Um, but it's a rag. It's yeah, it's exactly the same thing, but uh you you have to choose the books and the resources and then they're locked in. That's static. Yeah, it's it's static. So when if you're in the middle of a conversation, like if you ever had a really great brainstorming session or something with somebody, then in the middle of the middle of the conversation, you're like, you know what? Hold on, I got a book about this. Let me just pull this up. And you're like, yeah, here's the answer. Cool. Well, let's keep going. But you know what? Now that we're talking about this, I have a book about that too. This is evolving, right? The conversation. Let's pull that in and let's talk about it. Here's the answer. So that that like evolving book resource thing is how you naturally think and how you naturally evolve conversations and knowledge. So that means that if I if I have a static environment, I have to now create a new chat with that book separately. I don't get the crossover between these two things so I can see where the relations are.
SPEAKER_01I have to just And it might also be pulling in information that's not completely relevant just because it happens to sit in that notebook.
SPEAKER_00Exactly. So I don't like I don't I don't know anybody that uses notebook LM for that. Yeah. Hilariously. Everybody uses it to make podcast interviews and listen to them because that's really fun. Yeah. Right? And that's why everyone's talking about it, because the voices are so great and they sound so real. Um, I saw this one guy and posted, he used, he uploaded a document that just said poop and fart. And it said poop and fart over and over again for like, you know, 15 pages. Uh-huh. And then he made a podcast. And it was the most amazing podcast I'd ever heard. It was it was them going, you know, one of the users uploaded this file, and we're looking at it here in the studio. You know, it's like not really happening. Yeah. And and it's just the words poop and fart, repeated over and over again for pages. Like, wow, yeah. You know, and then the girl goes, you know, but I wonder if um if this is really just us trying to find meaning in something that doesn't have meaning. Don't we do that? Don't we look for patterns when they don't exist? Yeah, they we totally do. And it made this philosophical, deep discussion about poop and fart. Yeah. And it was the funniest, coolest thing I'd ever heard from this thing, right? But the power of this tool, that's really cool because you can now make a podcast and listen to it about any topic you want, right? And from the resources you decide, but people aren't using Notebook LM to actually talk to their files. Yeah. They're using it to make podcasts.
SPEAKER_01So with IP vault, I I can toggle on and off which context I want to be included in this particular chat.
SPEAKER_00Yeah.
SPEAKER_01Okay, that's the distinction.
SPEAKER_00And you can also do something. This it took us a while to figure this out. But AI doesn't write like you. And if you've ever tried to get to write like you, you know it's like pushing a snowball uphill. Yeah. You know, it's not, it doesn't really work very well. And uh the first attempt that everyone does is make a writing profile, writing style profile of you by giving it a sample of work. Yeah. And then saying, write like me, and it still doesn't do it. It's pretty terrible. So what we figured out was that uh if you want to actually make a writing style, we call it personalities, okay. You need to be able to turn it on and off in the middle of a chat. Say, like, okay, I'm talking to you, but now I want you to write this email, so you need to write like me, right? Or now you're writing a book, yeah. So a chapter, so you need to write like me, but but while we're brainstorming, I don't need you to. So that's the first thing. You need to toggle on and off. So we made something called modes for uh IP vault and librarian. And you upload your work, and instead of trying to make a profile out of it, it finds uh things that you said, phrases and stuff, um, about something similar to what you're writing about. So if you have a whole bunch of content about um, let's say it was uh medieval uh carpentry, whatever reason. And you that's all you did. It was medieval carpentry, and you talked about it on podcasts, and you had a thousand podcasts. You upload them all in, and then you say, uh, also, you had a bunch of emails, newsletters, stuff like that. You upload those all in, right? So now you got all this content. You want it to write an email. You don't want it to write like you speak a podcast, because you don't write like that.
SPEAKER_03Yeah.
SPEAKER_00So, but you do want to write about this subject matter, whatever it might be. So you're talking about, you know, chisels, for example. What it does is it'll find all the times that you talk about chisels, and it will find the sentences and the sentence structure, pull that in, and say, This is how this person talks about chisels. Because you talk about chisels and write about it differently than you write about dogs, differently than you write about family and friends. Right. You have a totally different tone, and it uses them as examples instead of trying to make a profile. And because of that, I can now say, here are all the examples of how this person writes about this, here's all the content, so they're only using that for style, here's all the content um they believe in over here. Now write like them, and it'll do it. It actually works. This is fascinating. Yeah.
SPEAKER_01Um, you know, I was at a thing here just the other night, the Texas Venture Fest, and I was talking to a guy who had put his tax returns into Notebook LM. Uh-huh. Because he didn't really understand uh what was happening. Yeah. And after Notebook LM did its thing, he completely understood like his tax situation and the strategy behind it. It was crazy. Yep. So it's again going back to what I I was asking you earlier, how do you think about these things? Like, I think that's the the number one, these tools are out there, and the only limitation is what you can come up with of, oh, I wonder if I can have it do this.
SPEAKER_00Yeah.
SPEAKER_01That's fantastic.
SPEAKER_00Yep. And speaking of the tax stuff, I mean, again, you you I like the idea of having a sovereign system that I can work with and I can choose and I can switch between them because these things sometimes they get dumber and I don't want to use the the LLM anymore. Like when ChatGPT released GPT-4.0, they were like, it beats everything. Yeah. And then I got in and I was like, this is pretty good. Then about a week later, I was like, this is terrible, actually. It was so bad. Yeah. And and I was sorely disappointed. However, they already had hit the benchmark with it. They got righted and everything, and then they made it dumber. They made it dumber because it was too expensive to run it.
SPEAKER_03Oh.
SPEAKER_00They want to save money. Yeah. So now I'm using this thing, and I actually don't know if it's good or bad. Yeah, yeah. Every time I log in and I talk to it, I'm wondering: is it something that they've, you know, they've deprecated in some kind of way, right? That isn't working anymore, or is suddenly gonna censor me, or whatever it might be. Like I'm not allowed to think about certain things, right? So I don't want that to happen. So I like to have my own systems in my own, like I like to have stuff on my computer too. One thing that I did um though, because I thought that it might say I can't help you with that, yeah, is upload, uh download, first off, which was a pain in the butt, the hardest part was downloading this, the entire tax code for the United States.
SPEAKER_01Wow.
SPEAKER_00So I have the entire tax code in my librarian, in my IP ball.
SPEAKER_01Uh-huh.
SPEAKER_00And I was like, okay, I want to fill out my taxes. Here's the I just I wanted to take a picture, right, and upload it and say, here's all the fields. And and it pulled in all of the related tax code. It meant that I could find write-offs that that were otherwise really hidden in thousands and thousands of pages of documents. And it would tell me exactly how to go about it. Also, it let me know if I wanted to start a new venture, what's the best situation to set up, you know, all the different types of LLCs or uh entity formation types. And and it walked me through them all. But I had a hard time with the files. There were way too many files, right? I tried to upload them all and I kept running into errors, they're all too big. This is because uh they don't want you. The United States uh IR the IRS doesn't want you to um actually know all the write-offs. The write-offs because they want you paying taxes. Yeah. And there are laws saying that they can't overtax you, but there aren't any laws saying that if you don't care, yeah, you can't overpay. So you'll just do it, you know? And this is gonna change. As soon as people hear this podcast, first time they're gonna be like, what? I can do that. Yeah, all they have to do is go online and search for a US tax code and then find the download buttons and download them all, you know, and then try it. You need to upload all of them at once, yeah, but it's your right to do this, you know. You have a legal right. Another thing is laws, right? Yeah. This is what's messed up, I think, in general, right? We have a ton of laws, tons, ton, tons of laws. Every single country has a ton of laws. They're all different. They're all different state to state, they're different county to county in certain ways, and you have a constitution, but you have a lot of other things, right? And it makes it very, very, very hard to understand. If you screw up and break one of these laws, whether you know it or not, if you didn't even know it was a law and it was illegal, you still go to jail. Yeah. And no one in the United States, not a single person, knows all the laws. Yeah. Therefore, we all live with this sort of looming thing over us that we might be doing something we don't even know is illegal and we're gonna go to jail. Right? No one in the whole country knows all the laws until now. Yeah. Now you can download all the laws and soon in different systems, and I'll probably do this inside of librarian and and other places too, and build it to different stuff, right? As soon as you uh as soon as you download. all that stuff and you look at it uh with AI, you now can actually navigate piece piece by piece, moment to moment, knowing all the laws. Now you're not gonna be like walking down the street and it's like, hey don't jaywalk, because I see you're at the corner, right? You know, don't do that. It's illegal. Hey don't don't shoot that person who's over there because that's illegal too, right? That's not helpful. But but when you're in a situation where where you're you need to navigate kind of illegal waters and you and you're and you're setting up, this is definitely going to change. Yeah yeah that's gonna mean that everybody in the world knows the laws of wherever they are. That's gonna change everything. You know same with tax code it'll empower everyone and that's really where this is all going. It's gonna it's gonna balance out so that an individual person has the ability to do anything on their own.
SPEAKER_01It'll make you totally sovereign if you know I guess you don't even know need to know the questions to ask you to say what questions should I be asking.
SPEAKER_00You need to know what you want. I want to not pay too much in taxes. I want to pay the right amount only the fair right amount. I also here's a really great worse or use case here for for um workflow is I want to buy property right um I want to buy a property and I want to buy in the best place right I also want to get as much support with lending whatever it might be any chance that I can to just get the property without having to put any money down and without getting any interest to to bother me right okay cool well I can get a USDA uh loan for a farm for six hundred thousand dollars there's a ton of grants not actually not a loan loan it's a grant and tons of these things that are given out every year in the United States and the only people that get them are the people that apply. Well I don't have time to apply but guess what I can just use AI to do that. Yeah how long do you think it'll be before the USDA is uh aware that people are filling out um these applications with AI probably not fast yeah they're probably gonna be ten years yeah before they realize that people are using AI and then they're gonna have to change the structure but between now and ten years from now you can buy you know a million pieces of property you know and and it's it's not you know you have to keep them all up you have to have a plan but you can come up with that plan in minutes and you can apply for every single grant that exists in two hours for your business. You can get all of the entrepreneurial grants, all of the research and development grants, you know the I I throw my plastic water bottles in the recycling grant. You know whatever grants there are, you can get a list of all of them and then fill them all out using AI. And right now you don't see a lot of tools yet where you upload a document that's got fields and it fills them out for you but that is coming. Right?
SPEAKER_01It'll fill it in with the AI provided answers.
SPEAKER_00Not just that yeah but it'll do it itself. So like you get your taxes right and you see them there and you go and you can talk to it too hey I uh I uh you know my name's Josh here's my situation it'll ask you a few questions maybe and then it will go well this this if you're this is your goal and it's an AI it's an IRS agent basically uh you're like okay I'm gonna tax agent to help me fill this out it'll just go oh that's the field you're looking for boom and it'll write it in for you boom boom boom boom boom boom boom you'll see it start to pop everything in and it does all the research real quick pulls it back and then fills it out. Dang and this is stuff that you could build right now. Right, I know anyone could build it right now and make so much money selling it, doing whatever with it right you could totally do that right now. But everyone's sitting around and going I'm gonna wait for the for the you know for the water to settle like I'm gonna wait for the for the everything to shake down and whatever's still being useful you know then I'll jump in. No you won't at that point no one's gonna care. Everything that you have that you could build right now will be rolled up into one thing in the future. It's true it will. But that one thing in the future is going to have to buy those disparate parts and if you make one they'll have to buy it from you. And they don't have to but they probably will because they're moving too fast. So this is the greatest opportunity for business I've ever seen in my life if a person right now who has $20 a month can get ChatGPT can for $200 more can launch a business in six hours and get an offer to buy it for $500 for six hours of work think about that. That's the anybody who has $20 a month, that's almost the entire world has $20 a month. You know that's about to be our competition. Every single business that you have every bus everyone who owns a business anybody who has $20 a month now in time which you don't by the way because you're busy running your company they have all your secrets they can strip them straight out of your website. They also have all your models they also have all your mistakes they can see everything you did wrong and they can make a better product and they can put you out of business. They can come at you from maybe three four hours of work and replace your entire business. And do it way cheaper undercut you price wise and serve a larger group of people and be more efficient have fewer employees move faster and that's it. If you don't apply AI right now to your business you are going to be dinosaured the hell out of this whole thing really fast.
SPEAKER_01Man we could go down these rabbit holes for a long time is there any specific I don't know advice that you tend to share with people regularly who are on the journey is there any lesson that you need them to be paying attention to or is there kind of like some standard something that you say a lot that you know it's it's landing with people um yeah I mean there's there's a few things I mean one because we teach we teach businesses how to apply AI right to their businesses.
SPEAKER_00This is the I think one of the greatest places to it's my favorite business of all time this and it's because I love helping people like this but also because it the impact is immediate. Yeah and these and these businesses are they're good people doing great things and they suddenly are seeing this thing apply and they're they're 10xing and their competitors aren't doing it that fast. Their competitors are all sitting around doing nothing right waiting like like they they used to be doing so I love doing that a lot. In that world I wanted I wanted essentially the employees that everyone's scared are going to be put out of work by AI I wanted them to be AI programmers basically without without using code something very simple and they would learn to build tools that do their job for them with AI so they don't have to do that anymore. And the company owns those tools this person becomes more valuable so they're not going to get laid off the um now because now they can build AI tools so now they can AI fy your whole business right that makes them more valuable. So now the company doesn't is not going to fire this person because they are worth more money. The um person's not going to leave because they're getting trained in how to make AI tools and they have now they don't have to do the stuff they don't like anymore. It's being handled. Yep and then um the company if they do leave the company owns all of the AI tools now the company's worth more money because now the company isn't just going to sell its SOPs which are a list of on a spreadsheet or you know an Asana table or something or a Trello board. It's got SOPs uh that are AI fied. Now the SOP says in order to do this job you have to come here copy this paste this push this button and copy that and read it anybody can do that. So now a company wants to buy your company they can hire whoever they want to do these jobs not a subject matter expert nothing like that. They can scale it at ad nauseum. They can do it as much as they want right so what was happening is that these these employees are not telling they're all using AI right there shadow usage. Over 70% of employees are using AI right now. Forty percent of those uh or less of those uh uh employees that are using AI are telling their bosses yeah sixty percent or more are not and here's why because they don't want you to tell them to do more work yeah or to say I don't understand that therefore you can't use that anymore. Sure. I mean there is that's a that's a legitimate concern yeah the actual problem with that is that now they're using tools taking your company data elsewhere you have no idea if it's compliant you have no idea if you're leaking anything you're you can't guide them because you're too tight to actually help them you won't say I'm gonna spend some money on you and train you and we're gonna use good systems that I can grow my business with. They're not thinking like that. A lot of people aren't thinking like that. So when the employees are doing this thing called um there's a there's a group on Reddit called Overemployed and it has hundreds of thousands of people in it and they're working multiple full-time jobs. Yep and they're using AI to do their jobs and they just sit back. Yep they have whole entire threads of how to have two Zoom calls at the same time and be on two different calls and like oh what happens if they both ask you a question at the same time oh well this is what I do you know I just you know I hit this button here and it makes my audio break you know and it and I can't talk or it makes my camera and then I and then I type uh here um I have that already written the type in the in the chat uh one minute got to reboot and then they just have the other chat over here and then when they're done they wow they push it over here and that's what they're doing. They they have the entire system set up for this and they're helping each other do it. Yeah. Because it's the only way they're gonna get ahead.
SPEAKER_03Yeah.
SPEAKER_00But what's gonna happen as a result is that these people are going to be the ones that end up building these companies out of nothing real quick because they're training themselves in AI and they're gonna put you out of business. They are already in your company working for you right now. They're going to put you out of business they see everything that you're doing. So you need to train them. That's the thing you have to train them come up come out with it you know come up off board with it let them actually learn and train them on how to use AI to do their jobs. If you don't do this you're if you're thinking oh I'm gonna wait till AI replaces everybody uh in my company you have a you have a rude awakening coming big time because what's gonna happen is the people that you you're gonna lay off they're gonna go out and build competitive businesses and they're gonna undercut you. Because they understand the domain yep they understand the domain and they understand it better than you do. Yeah because they're at the granular level exactly now you might be thinking a lot of people are thinking but they're not entrepreneurial. Yes they are they're getting two or three jobs at once. Yep that's that's how it always starts you know and that's it once you start seeing that you can make more money and you can work less you can't have a boss so for those of you listening up you're taking notes.
SPEAKER_01Hey so this has been a fantastic interview I'm yeah like it was kind of uh random that we were in the same city at the same time like this so it was awesome to do. Yeah total totally fortunate for me too for those of you who aren't familiar with what Josh is doing though keep him on your radar. He was a huge fan favorite of our audience at the chief AI officer community um we had him come on to do a guest episode one night of a training call we did and I think it lasted two and a half three hours. Yeah it was great and that was months ago and people still talk about that call. So excellent yeah I love that where can people find like if they want to plug into this type of thought from you where is that?
SPEAKER_00Yeah I think that it you know if if you really want to work with me the um the best way to work with me is to let me train the team and let me get them all ready with AI is all that we do is that and we do that with at uh the website's real time it's rltm.ai okay real timeai and which is hilarious because uh openai just released an API called real time api and it's spelled exactly the same and everything is like all right you know that's great but we're not that that's not openai's um go to rltm.ai you can scroll down and you can um you can actually apply there for uh for a training okay and people can you can join and it's honestly do it now I mean I'm not trying to get your money I don't care I I mean I love I love doing this for people yes and of course I love being able to run the company and it's great and it's growing and it's wonderful but it's it's not about that. This is it's literally you have to do this right now. So I think that's the best way. Go there you can you know apply you can reach out to us through that um you can email me uh through that too but I think the best way is just to fill that out and talk to us you know and I'll get on the phone with you and I'll talk to you about whatever's whatever's possible there. So I love it and um yeah the the working in this industry as you know has been the the coolest yeah freaking thing I've ever seen in my life and every day it gets cooler. If you're on the sidelines and you're thinking about joining uh joining in and playing a little bit just just jump off. Yeah and jump off and jump in and and play. Don't worry about it you know I used to look at it like it was a river. It's moving so fast if you try to jump out of the river to monetize something, you know, like build a build a tool you discover something amazing in here in AI and you want to just stop now and build this tool for the next year and a half like I did right yeah I wouldn't do that. Yeah right I would just keep learning keep learning spend all your time learning and nerd out about it as much as possible and apply as much as you can and if you do want to build a tool that's great but um I think that's really really fun and I I'm doing it myself. I think it's awesome but um but don't think that you're gonna monetize it and sell it or whatever right don't do it for that do it because you love it and and because this is something that you want. So yeah I would love that if they want to get in touch with me that we put a link you know I will in the show notes for sure and then um and IP vault check it out yeah you know it's a high end product it's all about your IP it's not like a you know little thing that you do on the side this is like a real real business program you know and we update it all the time but we are going to continue and we eventually will get Jarvis and it whoever gets Jarvis first um the group of people that get Jarvis first are going to be people who do it on their own. Yeah and those people if you don't know how to do that kind of stuff yet those people are going to be I mean they're gonna they're gonna get a a two inch head start that's gonna widen and widen into 10 miles. And you're not gonna be able to catch them. Yep and they may slow down they may lose it after a while that advantage they may screw up of course but there'll be some that don't and the ones that don't are going to be changing the world and you can be one. Yeah you know so that's awesome advice.
SPEAKER_01Yeah cool well thank you so much bud thank you appreciate it thanks for tuning in to using AI at work don't forget to subscribe for more conversations about how to use AI at work. And a special thank you to our sponsor Chief AI Officer for empowering businesses with AI education and training. Visit their website for free AI readiness assessment and AI strategy guide to help you get started using AI at work. That's www.chiefafficer.com thanks to our producer Evan DeSolnier for making this episode possible follow us on Twitter at the handle usingAI at work and visit www.usingai at work dot com for free resources to help you harness AI in your role