Dominate Content Game Using Multiple AI Models | Dustin Stout

33:27How do you make sure you're not losing some of your natural skills by over-indexing on the AI so that you can still keep that human spark? Yeah, I think it first starts with an awareness, a desire to grow. I think anybody who does want to grow will be highly sensitive to that. I know I am. But, you know, think about it like almost like our physical world. I could sit here all day and be great. I've worked from home since 2014. I could stay in my house all the time, but there's a part of me that starts to get antsy and a part of me that's like, I got to get out of here. I got to stretch my limbs.

34:11I got to go to the gym. And in fact, I'm missing my gym, uh, workout time today, uh, to be here with you. Uh, not that I can do some pushups together on camera. If you want, no, I'll go after this. But like, I start itching, right? I start to get that feeling because I want to be growing. I want to keep my physical, my strength, I want to stay physically fit. And I can tell when I'm not. Same thing is cognitive. So like when I'm being lazy, mentally lazy, I can feel it, I can start to feel my brain getting, you know squishy so i i can feel i have a sensitivity to it some people may not and one of the things that i think is is very true that lazy people are going to be lazy they're going to find more ways to be lazy whereas people who want to do their best work will find ways to do their best work and so i think there's there's definitely a an awareness that needs to happen yeah it can be a crutch so back in the day when we had to memorize everybody's phone number right how many phone numbers you have memorized today probably not that many not many not many I can actually you know what I realized the other day dude this is so interesting is I know the last four digits of a decent number of phone numbers because sometimes they come up right but I don't I could never tell you the full phone number anymore so like I recognize the last four digits because that's what the iPhone will show like next to someone's name but but I couldn't tell you what their phone number is.

35:42I'll just, you know, like my ex-wife is 3631. My dad is, you know, 9-1-1-3, you know, but I, I honestly couldn't even tell you what the first three numbers of their phone numbers are today. And I was like, that's a, wow, that's a really weird thing that developed with the technology. Yeah. Uh, well, you know, so back to what I was saying is just that, like, I think that if, as long as we keep our wits about us, we keep our critical thinking hats on. That's, I think that's the most important part of it is critical thinking. again going back to this idea that ai makes mistakes ai hallucinates ai isn't perfect we have to keep our critical thinking hat on and as long as we're doing that and and we're realizing that our most valuable asset is our unique combination of experiences that leads to those serendipitous moments that that god has only built us for i'm also a firm believer that god created us for a purpose on purpose and that there's nobody else in this world that could fulfill that purpose the way that he built us to.

36:44So knowing that, I think as long as we keep that at the forefront of our minds, like we're here to be uniquely us and AI can never be uniquely us, we can fight that urge to be lazy and fight that urge to kind of just use AI as a crutch rather than using it as a stepping stool. So I want to talk about your transition. You wrote this incredible article actually like two days ago or a day ago. And I was pulling up your website because I was learning about Magi and I want to get there because there's a bunch of really interesting things that you've done that I'm like really intrigued about and I want to talk about for the audience as well.

37:30But one of the things I thought was really interesting in this article, I'm going to link it up in the show notes, everybody. So whether you're watching or listening, just scroll down on whatever platform you're on, and I'll have the link there. But one of the things you talk about, dude, is this capture, right? You had always been a social media guy, a design guy. This is what you'd always talked about. It's what your audience knew you for. It's what your social media followers, you know, why they followed you, why they were connected to you, or for those insights. And now you see this opportunity, you're called to it, you fight it, but you can't get away from it.

38:02And you decide to make this move into, hey, I'm going to become an entrepreneur in the AI space and create this tool. And you said there was, you talk about it in the article, and I think you do a wonderful job. And that's why I'm referencing it and pushing everyone there, is the friction of people had always seen you as X. And you're being called to become Y. And there's a lot of friction in there, right? And it creates doubt and it creates probably some frustration and it creates some fear. How did you, I think that's really natural. Again, we talked a little bit before we went live. I've experienced a little bit of that in my own life.

38:44One of the places is taking this podcast from being an insurance only podcast three years ago to the last three years, expanding it out into broader topics, bigger topics. And I've had a lot of people email me and go, dude, I loved your podcast more when it was just insurance stuff. And it's like, yeah, except over a decade, I basically had said everything I could and wanted to say on that topic and I needed to expand. Right. Okay. So I think your jump is probably even more powerful because here you are really, you're becoming an entrepreneur. like your career, your income, your ability to provide for your family is now in this company and you have to make this transition.

39:25How did you, one, what were you really dealing with? And two, how did you fight through that fear and just maybe confusion, frustration, et cetera, to actually make the transition to where you are today? How long we got? Do we have time for the Lifetime movie of my life? I'll try to keep it short. So here's the harsh truth about being a social media expert. Many of the social media experts you know are broke because there's not a ton of money in being a social media expert. There are a fair few who become super influential and have really figured out how to monetize their social media expertise.

40:07but the vast majority of social media managers, social media consultants, they are not valued as much as they should be. People don't want to pay them what they're actually worth. And it's a real struggle to make a living, a comfortable living in the world of social media marketing and consulting. There are many who've done it, but many, many, many have not done well. And so I found myself, I was always an entrepreneur. So back in 2014, you remember I started the social warfare plugin with a couple partners. That did fairly well. That was my first successful SaaS business. Around 2020, my partners and I had grown apart.

40:49We weren't agreeing on pretty much anything. And so I decided to exit. And I had a couple ideas of two products that I really wanted to build. Both of them in the realm of social media still. One of them was an app that would produce social media graphics quickly and easily in every size you need. Think of it like Canva Lite. It's supposed to be like you fill out a few words and then boom without having to tinker and resize. The other one was a way to create articles or turn articles or long form content or web pages and shop it up into a bunch of social media posts instantly. Now, this is all before the onset of AI.

41:28In my mind, I wanted these apps to be AI oriented. And again, this was 2020. It wasn't until November 2022 that ChatGPT hit the scene. So I knew in my mind, AI is starting to become a thing. We had some copywriters out there, some, you know, the Jaspers of the world, which back then were called conversion AI, if you're that old to remember, and then copy AI, a bunch of clones. and I thought to myself, I want to create an AI app that will generate images for people and one that will generate copy for people at scale for these social media jobs. Long story short, I had to raise money to build these apps because it was expensive.

42:10I didn't get anywhere close to the AI features because it just wasn't accessible then. I would have needed$100 million to make these products the way that I wanted to. I didn't have$100 million. I had about$30 ,000. so the mvps came out we built them took us about nine months and 65 000 later i'm broke nobody wants these products i'm i'm probably in the worst mental spiritual and financial state i've ever been in my life i've got a wife and three kids to take care of a mortgage that i don't know how i'm gonna pay and i realized like my career up to this point has been great until it wasn't and everything i was doing seemed to be failing around me.

42:55And then ChatGPT came around and it was almost like, it was one of those moments where I used it for the first time and I thought to myself, this is going to change everything. Like it was, I hate to be overly charismatic, but it was almost like God brought me to my lowest point. I mean, I was suicidal many times, thought I had nothing to give to the world. I was praying that God would take me home. And it was in that moment that I used Chachi Petit that God almost gave me a vision of like, this is going to change the world. And here's what it's going to look like in about five years. You need to pay attention to this.

43:40And I had such clarity in that moment. I had taken a couple clients. I was doing some web design and some social media for them at that point. So I started using ChatGPT. And as I used it more, the more I realized like, oh, this is great, but why doesn't it do this? And why doesn't it do that? And I wish it did this better. And why do I have to have five other subscriptions to have these image things and the video things and the audio thing? Like, why can't they build? And it was almost like God said to me, this is how I built you to find all the gaps, to find everything that's wrong and to build something better.

44:14And so I in eight weeks, when I finally decided I had this idea for Magi, I start to finish, built it by myself. No capital, no team, no resources, just me sitting at my desk going. I got to figure out a way to pay my mortgage next month and make sure my kids don't go hungry. and by the grace of God, I'm not a developer. I know enough code to be dangerous, but it was like when the student's ready, the teacher appears. God put all the teachers in front of me and I learned what I needed to learn and built the first version of Magi in about eight weeks, launched it March 31st, 2023, just four months after, or five months, I guess, after ChatGPT hit the world.

45:00Dude, it's incredible. And I believe, It's funny. You're in very good company, dude. I had something very similar happen to me with my last startup, found an insurance agency, grew it, sold it, unfortunately sold it to the wrong people. watched my baby kind of get torn up and shredded. And, you know, and I made good on the numbers that I needed to make good on. But like the first moment that I could hit an exit, I exited. And it was this moment where now, you know, I was used to having this company and I had 25 employees and all this stuff and was making money. And now all of a sudden, you know, like, yeah, I had a little bit of money in the bank, but I had no job, like no, it was the first time in my life that I ever woke up and didn't know what I was supposed to do, right?

45:48Similar, I knew and have felt the same thing with marketing, even as a CMO, and I've been a CMO a couple of times, you know, you're always put a peg below, right? Because you're the marketing guy. And I was like, that's, I got into marketing because, you know, charismatic, willing to create, and it felt like an easy path to move up the rungs. The problem is, and you talked a little bit about this, like you get pigeonholed as a marketing guy. and it's like, I'm not even a good marketer. Yeah. You know, like honestly, like if you were to put me head to head with like a good marketing person, I would get slaughtered.

46:24I just, you know. I really can't worry about myself. Yeah, like I'm like, I'm not just, and you know what I took it as and it's so, it's just, I love that story and I want to get into some of the technical aspects of what you're doing because I do, I want the audience to take some technical stuff away from this and I think your approach is really unique and really interesting but I just want to sit here for one second because I think too often, whether we're intrapreneurs, whether we're sales professionals in an organization and kind of living that salespreneur life or we are true entrepreneurs, et cetera, wherever we are, I think we look at these moments and we feel like we're being punished.

47:03And that's the way I felt at that moment. I was like, I had these enormous plans for this company and it just didn't happen. Yep. And I was like, why would you do that? Like, I gave up two years of my life. I got divorced during this time, not necessarily because of the startup, but, you know, like all these things were happening. And I just was like, what the fuck? Like, man, I try to be a good guy. You know, I'm a believer. I pray. I try to live a life that, you know, I believe Jesus would be proud of. And I think if I have to stand in front of St. Peter, I can justify. And fuck, why is this happening?

47:38And you go down this spiral. And and if I think and this is why I advocate for for for having a relationship with God, because in that time, like you said, you open your if you're willing to open yourself up and you allow yourself even moments. you start to you start to hear things and maybe it's god maybe it's just you're providing praying creates space in your brain for you to actually have some some original thoughts or or really get to the core of who you are but i what i heard was you you you're not a marketer you're a leader and and you need to get better at the business of leadership so that you can do this again for another company, which is what I'm doing now.

48:23I'm the CEO of a, of a AI startup in the insurance industry and like, uh, different from what you're doing, but, but, you know, just interesting. And like, I just hope guys, like if you're listening, you listen to Dustin's story, you know, you listen to mine, whatever we all have our own, like even God loves you, whether you're, you believe in him or not. So even if you're an atheist or whatever, or creating that space and act as if. Jordan Peterson, I think, said it best, act as if. And there's a lot of opportunity for us to hear and find what we really want to do. So you find this idea of God telling you, hey, you're here because you see the gaps, you see the spaces, and you create Magi.

49:09And so one of the things that I was immediately captured by is this concept of in context, the ability to use different models. One, I'm very interested in where that idea came from. And then maybe you can expand upon the power of being able to do that. Like what is the actual power of being able to start in open AI and then touch Gemini 2.5 and then move to Claude or whatever process makes sense. Where did that come from and why does that matter and how does it add value? yeah so it's funny transitioning from the pigeonhole conversation um i'm getting ready to pigeonhole some of these ai models uh but but to speak to that a little more like i i think anybody who's listening who feels like they have this pressure from the people around them to kind of stay in their lane um just know that sometimes you have to make the decision to remove those people from your life and stop listening to the people who want you to stay in your lane that's what I had to do.

50:10People wanted me to stay in my lane, were keeping me broke, miserable, and not fulfilling my purpose. And so you really have to have that bravery to just jump out of that pigeonhole and say, who do I want to be? In our case, who's God want me to be? And listen to that voice rather than cave into the peer pressure of, this is who people expect me to be. now to answer your question with ai models they they do have very explicit strengths and weaknesses the more that you test them the more that you explore them and this is also another reason i created magi because i wanted people to be able to explore it's so difficult to have to switch to this tab and try this one and then switch this other tab and try try that one and then switch the you know between the tabs and compare back and forth while at the same time the underlying technology is not giving you a fair shot at actually exploring their capabilities because they're all doing backend engineering and prompts that you can't see.

51:08So you don't get to explore. But when you do get to explore, you get to have a baseline for all these models. You'll find very quickly that you like how one model does this type of prompt versus how this model does another type of prompt. And you find very quickly you have favorites and you have preferences. For me in particular, I love GPT's creativity. I think it has the most out-of-the-box ideas. And whenever I'm looking for a perspective that I can't get to on my own or maybe an analogy that I just need a better analogy because the one I'm coming up with is just not hitting, GPT comes up with some really great creative thinking.

51:45Whereas when I'm writing marketing copy or sales copy, GPT falls kind of short. I just never really buy into it. But I hand that same task over to Claude and Claude writes copy better than I ever could. So there are definite strengths and weaknesses of every single model. You know, go to Google Gemini structure. I mean, think about Google in particular. It's all about structuring the Internet, right? So it makes sense that they've engineered it in such a way that it's really, really good at structural stuff. stuff and so when you find these different strengths and weaknesses of these different models and you you realize well i can let's say a blog post for example you know i need to write a blog post about this keyword let's brainstorm the the titles with gpt because it's super creative and then let's brainstorm that outline with gpt but once i have that outline and i want to do some actual writing i don't want gpt to do the writing i want to switch over to claude now how would you do that without magi you would have to copy that outline head over to claude log in start a new chat paste that in say hey we're gonna build it and then give it all the background right but with with magi because we've put them all in one place you can essentially say all right thanks chat gpt now i'm gonna bring claude into the room and claude's gonna take over from there as if it was in the room the whole time so think of it i think of it like having a room full of experts who are just sitting there waiting to help you out whenever you need them and then you can just say all right, you're in, you know, almost like a, like a, like a sports team.

53:17You know, like, hey, hey, chat, JBT, you're on the bench. Claude, come on in, you're up next. So that not only gives you a more seamless experience, an unbroken experience, so you're not breaking up your flow. But what's really fascinating is you find that these models, without going into like super technical how it works basically when when you start with one model and you bring in another one the new model sees the entire conversation from you and the other model and it it's almost like it it learns from it so claude can actually learn from the creative choices of gpt earlier in the conversation and actually becomes more creative and the more you pull in those abilities the longer that conversation goes, the better these models get to where you can even use a dumb model near the end, one that's not very sophisticated, maybe an older model, and it creates the same level of output almost because it's learned from the earlier parts of the conversation.

54:19So it's really fascinating to see how all these models can not just be leveraged in ways that brings out all their strengths and none of their weaknesses, but they actually enhance one another as the conversation progresses. Yeah, this to me is the future of creation because I completely agree. I at one point, and I started deprecating them only because before I knew what you were up to, and this is gonna be my go-to tool now. I mean, 100%, I've been playing around with it already. Like it's, I was going ChatGPT, then I would do research in Gemini. Then this is before deep research with ChatGPT, then I'd pull the research in.

54:59And then like you said, the titles and like, like some of the hooks that GPT creates are phenomenal. Yeah. But I agree. You get into the body and unless I feel like GPT has a really good feel for narrative design. But like you said, marketing copy, sales copy, email copy. Like you said, it's kind of like blah. It doesn't, it doesn't hit. And then, you know, you're, you know, I played around with Grok for a little while, but Grok pre the latest release of, you know, Grok 4 was iffy. 4 seems a little over tuned, I think, for personality. Yeah, 4 is wild. You know, I think they may have, the pendulum may have swung a little too far the other way.

55:42We'll see. I agree. But I do think what you're describing, this workflow of using, it's like anybody else. Like if you have a plumber, come over to your house. It's not like they have one wrench and that one wrench does every function they need, right? They got a screw gun and a wrench and a socket wrench and, you know, a pair of needle nose pliers. And it's like for what they need to do, they're pulling out of their tool belt the best tool. And to have one place where you can pull them all together, to me, is an absolute no-brainer. Like this is where we have to go. Like do you see a time when, like to me, it feels like we're in an expansion phase, obviously, of AI.

56:24Well, I shouldn't say obviously. But to me, it's obvious that right now we're in pure expansion, right? We're just new models popping up all over. You have entire countries creating new models and all this different stuff. And it's really wonderful and wild. It's really like a true FAFO moment for the world. Do you see where we start to come back in and these models actually natively start to work together? Or do you think the business of AI will keep that from happening? I think the business of it will keep it from happening. Because these companies are all trying to, it's a land grab. right now and they they want to build their brand as the best of the best ai model and so they're going to build their little cages around their products and try and pull people into that ecosystem which is fine i think it's great it incentivizes them to to really make the better model because there's heavy competition right now well when i say heavy there's really like four competitors there's four top ones you know deep china out of nowhere comes up with a new deep seek model that is more efficient more costly and can do just as well as the gpt models but at a lower price and so we have real competition in the market so i think that's great now the thing is these companies are losing money hand over fist on subscribers so your chat gpt subscription your 20 a month you're paying chat gpt opening eyes losing so much money on these subscribers that they have to find other ways to monetize their models and that's where apps like magi come in because magi in relation to open ai is pure profit for them every time one of our users uses a chat gpt model we have to pay for it we are paying for every single transaction that happens.

58:15And so OpenAI needs people like me to create these multi-faceted, complex, comprehensive systems that utilize their technology in a way that's 100 % pure profit for them. And so I think it's a really interesting ecosystem, right? They have to build their own little castle on their own little hill, but at the same time, they have to make it friendly with working with other apps because that's where their real profit is. Now, in terms of capabilities, I think it is rapidly expanding, but the sort of truth behind the scenes, if you talk to any of the AI engineers, is the AI model growth trajectory is actually slowing down.

59:00Because if you think about it, once you've trained on all of the world's data, how do you get smarter how do you make a smarter model on the same data so now you know they're talking about creating synthetic data you know data that's fake created by ai so even with this you know you're watching these um capability uh charts and it's like you know shot up real quick and then it's starting to kind of plateau out and so i think we're seeing a slowdown of like model releases and model capabilities there's all these rumors of gpt5 coming out before the end of summer and that might happen. Um, but I don't think it's going to be the mind blowing hype of, you know, the, the first version of it, right.

59:40It's going to be an increase in capabilities, but it's going to be slowed down. And I think what that means is now this, uh, wild West is starting to get a little more tame. It's starting to stabilize a little bit more. And I think that's a good thing for everyone. Yeah. I think just like raw hype cycle all the time. We just all be burnt out. I agree because that stabilization allows entrepreneurs like yourself to start to create the towns, the meeting places, right? I mean, if we're using the Wild West analogy, right? Like you're able to actually put down some roots and build something because now the models aren't shooting off in every direction.

1:00:18You know, the iterations have less deviation in them, which I think is a good thing. Actually, the last, I don't know if you listened to the All In podcast, but the last episode in the most recent week here, which would be the week of July, I guess it's the 7th, it would be that week's episode, the week before we're recording this. And I'll try to link that up in the show notes too. They talk about this idea of synthetic data. And I know we're getting up against a number, but I just want to put this in front of you and get your thoughts on it because you're so deep in this world. So guys, the idea of creating synthetic data is exactly what Dustin just said, right?

1:00:55Essentially, ChatGPT and OpenAI have indexed almost all of the knowledge that they can get their hands on today. There are pockets that are still hidden, but for the most part, they have their hands on everything that's available digitally they've consumed at this point. And they're consuming as new as created as well. And And just read your private API contracts if you don't think that's happening. Unless you're a Magi user, then it's all private. Yeah, that's good. Some of these tools, you've got to be really careful because some of them, there are different versions of the API. And I think some people don't realize what they're putting in.

1:01:32That being said, the idea of synthetic data is that they take and start creating data that has no physical or any basis in the real world in order to train these models on increasingly difficult problems, on increasingly diverse problems to give them things to train on. And I guess my question for you is like, is that a good thing? Like, how should we feel about synthetic data? Because I caught ChatGPT in a lie the other day. I asked it a question about where it gathered a specific type of data that was relevant to our tool. And it gave me an answer, which I knew at face value was false. So then I pressed it and it gave me this like whitewashed version of the answer.

1:02:21So I pressed it again, and then I got the real answer, and the real answer was a little scary. It doesn't matter. But if we're training on synthetic data that is, we'll call it false to begin with or just not based in reality to begin with, how do we as business professionals, like, should we consider that? Should we consider models that are using massive amounts of synthetic data when, you know, and where do you see that part going? Because I do think that's a real question, especially for regulated industries. Like, this is a big problem in our space, right? Like, insurance is highly regulated, both at the federal level and individually at every state level.

1:02:58And right now, insurance carriers cannot use AI algorithms to do any kind of underwriting. They can train models today, but they can't deploy them in actual underwriting. So a lot of carriers are starting to play with models as test data, but they can't actually use them. and the reason is hallucinations and all this kind of stuff. But the only way, and this is kind of where this question comes from a little selfishly, is there's been a lot of talk specifically in the insurance industry around the use of synthetic data because everyone's policy information is proprietary and is in these private vaults.

1:03:38It's their competitive advantages. Who bought what for how much and how much did they buy, et cetera, et cetera. So if everyone has their little private pool and there's not one singular place for it all to come together, the thought process has been the only way to really get more sophisticated, more accurate and deliver faster results is to start to create synthetic data. But if you do that, are you corrupting the underwriting models? So like, where are you on synthetic data? How do you see this being used? And maybe if I'm a business owner who's all in with everything you've said so far, how do I navigate models that are using this synthetic data?

1:04:17I think it's really hard. And one thing you'll never see me do is pretend to be smarter than I actually am. At least I try not to. I do that all the time. So if you did, you'd be fine. so i i really try not to jump to conclusions because you know experts have been wrong before i've been wrong before and i i really can't make a decision on the whether or not the synthetic data is bad or good because we i just haven't tested it yet i mean so far there have been some new models that come out the new clod models have come out the um uh opening gpt 4.1 models have come out and they've been trained on some synthetic data and so far so good uh so i i am optimistic about it but i'm optimistic because i have this emphasis on critical thinking going back to this idea like we need to use our brains and we need to not just take what the ai gives us at face value like you immediately saw when it was wrong right because you were looking at it critically.

1:05:19And I think as long as we keep that critical thinking about us and we analyze everything that it gives us with a critical eye, I think we should be fine. And in my optimistic outlook, I think the synthetic data will be mostly fine. But I don't know. Could be terrible. We can only see because it's completely new territory. We just don't know. And I don't know enough about what kinds of synthetic data they're creating, what it looks like, what it reads like, how it works. So I'm going to remain optimistic and I'm going to remain, you know, the advocate for think critically, don't take anything at face value.

1:06:00Assume it's going to lie to you if you're not careful. Yeah. Trust, but verify, right? Trust, but verify, trust, but verify. Dude, It's been so great to reconnect. I'm so excited for what you have going on. Fan of the tool, user. Where can people dive deeper into, and we'll have links to anything that you mention here, especially Magi, just where they can go and sign up, et cetera. But if they're just interested in your takes and kind of staying with your ideas and getting deeper into your world, where should they go? Well, you can visit my personal blog, DustinStout.com. I'm pretty inconsistent on there.

1:06:36I've been trying to be more consistent writing recently now that I have a little bit more margin. Things have gotten a little bit crazy with my last podcast interview. But yeah, I'm trying to write more. But you can find everything about me there. Links to Magi as well. You can go to magi.co, M-A-G-A-I dot C-O to learn more about Magi. Dude, appreciate the hell out of you. I love what you're doing. I really love your story. So good to see you again. You too, bud. Be good, man. It was great. Good.

1:07:11Let's go. Yeah, make it look, make it look, make it look easy. Thank you for listening to The Ryan Hanley Show. Be sure to subscribe and leave us a comment or review wherever you listen to podcasts. Ain't anything for me. I never switched to no change in me. The only thing changing is... Happy holidays. Want to give your host a gift? Consider subscribing, rating, and reviewing the show this holiday season. It really helps the show grow. From all of us at Believe, have a Merry Christmas, everyone, and a happy holiday.

1:07:47If you like the show, please take a moment to rate, review, and subscribe. It really does help the show to grow. Thank you for listening.