How to Get AI to Create Better Content

Happy 2026. Hey, how's this for a New Year's resolution? How about you master AI Marketing? The AI Business Society gives you live training with experts who are getting real results. No hype, no fluff, just proven workflows that work. And you've got access to an incredible library of training from many of the incredible guests that you've heard on this show. So secure your AI transformation by visiting socialmediaexaminer.com AI welcome to the AI Explored.

Podcast, helping you put AI to work. And now, here's your host, Michael Stelzner.

Hello, hello, hello. Thank you so much for joining me for the AI Explored podcast brought to you by Social Media Examiner. I'm your host, Michael Stelzner, and this is the podcast for marketers, creators and business owners who want to know how to use AI. Today I'm going to be joined by Dustin Stout and we're going to explore how to get better content utilizing generative AI. Much of what we're going to be talking about today is something you can put to work right away inside of your favorite AI tool, whether that be ChatGPT or Claude or any other kind of tool. But there is some stuff we're going to be talking about today that's a little bit more advanced that you may want to use a third party tool. For now, I will disclose to you that Dustin does have a third party tool, but we're not going to be talking about that at length inside of this podcast. Instead, we're going to be talking about the wisdom that he's bringing to the table and stuff that I think you can put to use right away. By the way, I'm at Stelzner on Facebook and at Mike Underscore Stelzner on Twitter, slash X. And if you're new to this podcast, which I'm guessing you probably are because it's only the second episode, be sure to follow this show and let your friends know about this show because we've got some amazing guests coming up and I think you're going to find it absolutely fascinating. Let's transition over to this week's interview with Dustin Stout, helping you simplify your AI journey. Here is this week's expert guide. Today I'm very excited to be joined by Dustin Stout. If you don't know who Dustin is, you gotta know Dustin. He's is a creator and entrepreneur. His latest venture is called Magi, spelled M A G A I CO and it's an AI tool that combines the leading AI technology from OpenAI Anthropic Google and others to help working professionals optimize their creative output using AI. Dustin, welcome to the show. How you doing today?

I'm doing great, Mike. Thanks for bringing me on. Always great to see you again and yeah, yeah, super excited to have a conversation.

Well, welcome to the brand new show. Today, Dustin and I are going to explore how to get AI to create better content. And Dustin and I have known each other for a very, very long time. But Dustin, what I would love before we get into today's content about how to create better content utilizing AI, I'd like to get a little bit of your backstory, like how in the world did you get into AI? Start wherever you want to start.

Yeah, so I guess we could start all the way back at the beginning. I mean, I've always had sort of a knack and an interest in technology. My grandfather on my dad's side, he actually worked for the Navy. He was what some might call a spy and was super into technology himself. And I, I just remember being in his basement a lot of times and he had all these gadgets and, and like really cool looking technology from that time period. And I was just always fascinated with it. So it was just a natural sort of progression that I was always kind of fascinated with. Technology got me into, you know, computers when, you know, the, the computer revolution came around and everybody started, you know, having desktops in their house and you know, I got online and social media became like one of my fascinations because I'm also an extrovert. So it was kind of a natural sort of fit for me to get into social media and then that led to marketing and all sorts of other things. But you know, being a sort of a student and a. Just generally fascinated with Internet and Internet culture and technology, when AI came out, I was just optimally primed I guess to be all over it and to be fascinated.

Yeah. So tell us a little bit more. Like obviously we met each other, I believe for the first time at social media marketing world, like way back in the early days. And I know you were coming there for many, many years, but somewhere along the line your fascination with AI led to you deciding to start a business. And folks, just so we're clear, we're going to talk about Dustin's business here, but everything else we're going to be talking about today really doesn't have very much to do with Dustin's business. And I just want to explain that because this show is a little different. I am getting founders of businesses on here because I think Dustin has A lot of value to add. But Dustin, you decide to start a software company, like, hello, tell us a little bit about the journey from how you got to marketing into software.

Again, I think a lot of where we're at today as professionals, I mean, you can't anticipate where the world's going or, you know, what's going to happen. Things, you know, things are moving at neck break speeds and, you know, culture changes. And so we kind of just follow the journey, you know, where it takes us a lot of times. And, you know, I've been grateful for both the highs and the lows that have led me to where I am today. I mean, it started as I was just a social media consultant. I was a marketing consultant building websites and doing social media marketing campaigns, content campaigns for clients. And so now naturally, for the last 12 years, I've been in the Internet culture and creating content online and specializing in doing that for businesses, specifically for growing their reach and growing their marketing. And what I found out very quickly is, although I, I was pretty good at what I did, at least that's what people told me, I was a great consultant. But I'm a bit of a perfectionist. So I'm the kind of guy that would spend eight to 12 hours on a blog post or an hour and a half trying to decide how round the button should be on a website. So that kind of proclivity doesn't necessarily lend itself to scalability. Right. So if you spend an hour and a half on a button, it's hard to really scale that kind of a business out and really, you know, reach higher levels beyond just your own personal effort. So I found that if I channeled that energy, that, that perfectionism into building products, digital products in particular, that can help many people at once. That was a much better use of my, the way that I thought about the world and the way that I worked. And so I got a digital products back in 2014, around the time that I started attending social media marketing world. I created my first product, which is a WordPress plugin. And then once I got that taste of digital products, I realized like, this is how I was built. I was built to create products that helped a lot of people. And so that just led me on a journey. Had a handful of other product ideas since that company that I left, first company they built and exited back in 2020 and a couple of really massive failures between 2020 and 2023 when I first started Magi, and I think the state of AI, AI has been around for years. And years and years, way before ChatGPT was a thing, you know, there were lots of products out there that were AI copywriters that us marketers were swarming all over the world. You know, the Jaspers, the copy AIs and all the million copycats of those products. But for me, they always kind of fell short. Like there was something about them that felt constricting. And I kind of narrowed it down to these, these two aspects. And the one thing was they were all predicated on templates. They, they had these libraries of like, use this template for this outcome. And it was sort of like a Netflix effect, right? First of all, you have to go in and sort through these hundreds, if not thousands of templates that somebody else crafted to have AI output something that hopefully looks like what you want it to be. And so you were in a sense presented with the paradox of choice, the psychological principle that when we have many, many choices, we tend to just not be able to choose. That's why we spend 20 minutes scrolling through Netflix and end up not even watching anything. But then the other principle of I have to fit my creativity and my goals and outcomes into somebody else's box, somebody else's box that they predefined and pre crafted. And that might work from time to time, but a lot of times it seemed to fall short for me. But when ChatGPT was released into the wild, it completely blew the doors off of those limitations. You could literally just tell the AI, this is what I want and this is how I want it, and it would do it. It was like magic, right? It was this like complete unlock. It lowered the barrier of entry. You didn't have to sort through someone else's way that they framed their problems and fit them in. You just tell the AI what you want and magically it produces it. And so this, to me, as a content creator, as a business owner, having worked with many different businesses over the last 12 years, I realized this is going to change the world. And so I started using it prolifically. And like any good entrepreneur business owner, I immediately started finding the problems like, well, why doesn't it do this? Or why can't I do that? And, you know, all these quality of life things that make my daily workflows and the things that I'm used to doing and producing made it a little bit more unpleasant to use. And so I set out to really solve my own problem and solve a problem that I saw a lot of business owners have that I've worked with over the years. And I wanted to create a product that just allowed them to access all the plethora of AI tools that are being flooded in the market and allow them to benefit from all of them in one place and with a more streamlined workflow, tools that help them to actually use what they want and organize it and put it into standard operating procedures and things that they're used to doing and work.

So what is magi then exactly? And who's it for? Just to be clear.

Yeah, MAGI is the all in one AI platform where it brings all the world's best AI models. ChatGPT, Claw, Gemini, all of the AI things. Image generators, video generators, and soon audio generators. It takes all of those tools that you would have to have a hundred different subscriptions to access, and it puts them into one beautiful interface so that you can access them all from one place with powerful tools that help you stay organized and collaborate with team members. So it's for mostly for working professionals. People are doing real work in a professional capacity and need AI tools to help them augment and enhance their work.

Awesome. All right, folks, Dustin has dove in. And by the way, when did you found magi?

We actually just had our first birthday. I started building it in late January of 2023, and then it was publicly released. We had a short beta period. I kind of just went crazy and just built it in eight weeks, start to finish. And then our actual first public signup link was released on March 31, 2023.

Very cool. Okay, so basically you got in within like four months of when Chat GPT launched for the Right. I mean, open AI been around for a while, but you started and you've continued to integrate all sorts of other powerful tools into this platform that you've built.

So.

All right, thank you for that story. I wanted everyone to understand kind of what you've been working on so that they understand why what we're about to talk about next is so important. So we're here really to talk about how to get better output out of the AIs, right? How to create better content. So let's start by introducing what exactly, why marketers and creators and entrepreneurs should even pay attention to the concept of generative AI? Because there's so many that see it as a threat. So what do you want to say to those who aren't sure about whether they should dabble with this concept of generative AI?

Well, I think one of the main reasons is if you've been working in any creative capacity for a period of time, any period of time, really, you realize that your creativity has a limit. You have a limit, and creative burnout is an epidemic. So I can say this confidently. Having worked with dozens and dozens of different businesses, there was one point where I was managing, I believe, 12 different client accounts, creating social media content daily for every one of those accounts. And we were doing it across multiple platforms. And so that means coming up with ideas regularly for creative content for different businesses, different genres, different industries, and your creativity really gets stretched. And it's difficult to continually and regularly come up with content at a high level. Eventually you're just going to get burnt out. And all creative professionals have to proactively attempt and try to prevent this creative burnout because it can be detrimental to your career. You know, if you start producing really poor quality work, you know, how does that reflect on you? And how do you. Once you've created a bunch of really bad work, it's hard to recover from that. So what I see AI as its most valuable. I guess skill is augmenting our own creativity, not replacing it in any way. I don't think in any way AI is going to ever replace us. There's a lot of worry and a lot of concern that AI is going to take our jobs, and that's really not true. The resounding wisdom of those who are deeply ingrained in AI and understand the full limits of its capabilities and where it's headed. We understand that AI is not going to replace you. Somebody using AI is going to replace you. And so it's up to us to really use this technology just like any other technology that came in and augmented our abilities and our. Enhanced our skills, enhanced our natural limitations. It's just a supplement to what we already have. And without us prompting the AI and guiding it, the AI is useless. If you create any piece of content where you have expertise in it, you'll know really quickly if you have AI generated that it has some limitations and it makes some mistakes and you could probably do a better job. But the value is that you don't have to start with a blank white page anymore. So if I was still back managing, you know, a dozen different clients in different industries, this would be a huge help for me to not, not burn myself out every single day. I could use the AI to generate ideas, to refine ideas, and to come up with variations of ideas very quickly and without breaking a sweat.

I like a couple things that you said here. Well, everything that you said, but particularly you said something that kind of spurred a concept in me. Technological innovation has an advantage to those who Use it, right? When the automobile came out, it allowed people to go further, faster than they could ever go before. When somebody invented tennis shoes instead of sandals, it allowed people probably to run instead of just walk, right? And if we start to think about back in the olden days when people were using swords, right, eventually somebody figured out how to put a little teeny sword on the edge of a piece of wood, right? And allow you to have an arrow so you could hunt at a bigger distance, right? And then eventually bullets inside of guns and eventually trains and planes and jets and eventually computers in our pocket, right, that allow us to, to take pictures and video where in the past we have to add separate equipment, right? So in the world of business and marketing and entrepreneurs, all this technological innovation has always given an unfair advantage to those who adopt it, right? But it's, it's only as good as the use of it, right? Especially when you're using something as open ended as gener AI. So this is a transition to my next question, prompt engineering. There's going to be people listening to this podcast that have no idea what that even means. So why don't we talk about first of all, what is prompt engineering? And then we can eventually get into some of the limitations with prompt engineering.

Yeah. So prompt engineering is this concept that you have to intelligently and with great intention craft a prompt for the AI that elicits the best possible outcome. And so you ask for something very simple. Maybe the AI won't fully understand what it is you are wanting from it, your expectations. And so like a simple prompt may not get you the best results, whereas if you spend a little more time adding some context or adding some specificity to your prompt, the AI will produce a much better outcome. And that is very much true. And I think over the, the course of the next five years or so, AI will get much better at interpreting simpler prompts. But in the meantime, it's up to us to really give the best possible input to get the best possible output. So the old adage is garbage in, garbage out, right? The same is true with prompting AIs and getting it to produce the best possible content for you. And that's essentially the idea of prompt engineering.

And you know, if you think about how you talk to a human, if you ask a human to write you an email about artificial intelligence, if the human isn't going to ask any clarifying questions, it's just going to pick whatever it thinks is relevant and it's going to output what you want, right? But normally when you talk To a human, there's back and forth. Right. You're giving them a little bit more. Like, AI as a concept is very broad. Right. So the more you can work to prompt the human, if you're art directing, a graphic designer, it's the exact same thing. Right. The more you can, can give them about exactly what you want, the greater the likelihood you're going to get back what you were hoping for. Correct.

Right. And most people, like at a high level, who have delegated any type of work, tend to be very good at this. I've. I've seen you be incredibly good at giving your vision to staff, employees, contractors. I've seen it. You are extremely good at giving them detailed explanations of what your expectations are and what the outcomes are. Most people who are good at delegating have that communication skill, and it absolutely is a skill that you can acquire, and it's 100%. How good are you at communicating what you want? And so if you can think of that in that same aspect of I'm telling a human exactly what I want that human to produce, if you apply that to your AI prompting, you'll get much better results. Because these AIs, they're trained to understand natural language and to interpret natural language in a similar way that a human would. And so, yeah, that's, that's literally all we do is just think of it like a personal assistant, a human being, and the details that you give it, it will take into account, and it will not take into account details that you do not give it.

Yeah, and this is really important for anybody who's experimenting with generative AI. If you give it a very simple prompt and you get a response that's completely off base and you just assume it's worthless and you walk away from it and you're not going to actually be really, you're going to have a. I tried it. It's stupid. It doesn't work. Right. So how do we then get the best possible results from generative AI? What do we need to be thinking about differently?

Well, I mean, it's, it's actually a little bit more deep than just, you know, becoming a good prompt engineer. Because being a good prompt engineer is one thing, but the challenge there is sometimes the AI still gets things a little bit off. And you might notice that a lot of AI generated content, sounds or looks like it was generated by AI because it's trained on vast amounts of data, like just an unthinkable amount of data, and it tries to filter your request through all of that. Giant pool of data, right? And it. So it has this sort of language that it has adopted naturally. And what you'll find is that language is very rarely the language that you might use. It might not be anywhere close to your voice. And so even if it gives you something that's pretty close to what you wanted, you still have to do all this editing and you have to like, change the word usage and restructure some of the sentences because it just doesn't sound like you. So the next level, I guess, of taking it from just prompt engineering is to understand how these AIs work and how they formulate their responses. They have this thing called a system message or a set of system level instructions. Think of it like a prompt that's given behind the scenes before the AI takes any of your requests. That's what a system message is. And in most cases, the system message for the AI, specifically for ChatGPT, we'll say, is, you are a helpful assistant. So it's given this very generic system message instruction to say, you know, be the best possible assistant. Which means it's going to be friendly, it's going to say things like, you know, I hope this is well for you, or, sure, I'll do that for you. Here's what I produced for you. And it'll have all these little nuances. But if you were to, say, be able to change that system message, if you were to be able to focus the AI in, say, a area of expertise or a informant of a specific writing style, you would be able to get much, much better outputs. Because no longer is the AI putting your prompt through these helpful assistant glasses. You're then telling it exactly what glasses to put on and what to filter everything through. And that's sort of the next level. Once you understand prompt engineering, the next level, to get to the next place of even better output and even more accurate output to what you want and how you want it to sound, that's the next level.

And you call this concept Persona engineering, right?

Yes. There are different terms, I guess, in the industry right now. ChatGPT refers to them as custom GPTs or just GPTs, I guess. But many other AI experts have referred to them, even before GPTs were a thing, as Personas or AI personalities or AI characters. Most popularly, though, they're known as Personas.

So what I'm hearing you say, Dustin, is that if you inform the AI to act as if it is a certain Persona, or basically it will put its self in that role. Right. And act as if it is like A doctor or an engineer or a marketer. Is that what I'm hearing you say?

Right, yeah. Or an expert copywriter or Hulk Hogan. One of my favorite things to do at the beginning when I was first using ChatGPT is some people were saying like, oh, just prompt it to talk like a celebrity or to speak like a certain person or character. That is well known. And it's amazing to see how the AI actually does adapt its language and it's, you know, how it outputs things to that character or to that personality. But when you put it at a system level, you know, it's one thing to prompt it, it still has to filter your prompts through the system messages. So you still get a little bit of a dilution of that instruction when you're telling it to take on a character or personality. But adding it to the system message completely eradicates any potential for like diluting that prompt. And so, yeah, you've, you see a lot of these other apps that have started to pop up before custom GPTs were a thing where they were giving characters and personalities to the AI and you get so much better outcomes from it.

Okay, so just so we're crystal clear, the AI systems like Claude and ChatGPT have a system level role which is be friendly and helpful. Something along those lines generically, right?

Absolutely.

And we want to essentially re engineer or program, if you will, degenerative AI solution to act in a certain way. And we're calling that Persona Engineering. Let's talk about filling the gap between AI potential and AI results. I'm sure you've tried ChatGPT. Are you seeing real ROI? Most marketers are barely scratching the surface of what's possible with AI. And it's because they're missing the training and the frameworks that will allow them to be very successful in their careers and in their future. AI business World allows you to get that training. When you attend, you get two full days and 20 expert practitioners, many of the people that you've heard on this very show, that will teach you exactly what you need to know to build the workflows and to succeed as an AI enhanced marketer. It takes place April 29th and 30th in Anaheim, right alongside social media Marketing World. And here's the good news. You can save $300 when you register by January 16th. Head to AIbusinessWorld live right now. This discount ends soon. How do we start with Persona Engineering? Like, what do we even need to be thinking about?

Well, the first thing, when you're engineering a Persona or you're engineering a GPT, we'll just call them Personas from here on out for clarity. When you're engineering it, you. You want to have several things in mind. You need to speak directly to the AI. So it's a little bit different than a prompt. In a prompt, you might say, do this, do that, write this, write that. But when it comes to a Persona, to get the most effective output, you want to speak directly to the AI, again as if it were a person. So really, you want to tell it who it is, what it does, what the user is going to give it, and how it helps that user accomplish its goal. A prime example of that would be I have an expert copywriter Persona. So whenever I need to write expert copy, I go to this Persona and I activate this Persona. And the brilliant copy that it comes up with is so good. But the Persona engineering is very minimal. So I'll just read you what I have on mind. It starts with, you are an expert copywriter. You are knowledgeable about effective communication techniques and have the ability to use words creatively to persuade and engage the reader. You help the user craft compelling headlines using concise and clear language and ensure their message is compelling and relevant to the reader. So very simple. Again, I said you are this, you do this, or you are knowledgeable at this, and you help the reader with this outcome.

Okay, so a couple quick thoughts on this. Are we just, like, if we're using ChatGPT or Claude or whatever technology, do we literally just. Just enter this in and type this in, and then the AI is smart enough to change this rule?

No, unfortunately not. So In a typical ChatGPT conversation, the AI doesn't have access to change its system message. It's really just sort of an input output interface. But you can create custom GPTs now with chat GPT plus, and it will sort of walk you through this process of like, it's basically a chatbot that helps you engineer the chatbot system message. But there are apps, such as the one that I created, magi, which will just let you create a system message and just type it out and save it to, you know, a list where you can easily swap it in or change over to that Persona. So different apps will approach that in different ways. And with custom GPTs, it's through the interface of a chatbot that you do it. You don't necessarily have as much control over it. It's sort of kind of interpreting what you want it to do and how you want it to do it. And then afterwards, I Think you can go in and edit the actual instructions yourself, but only after you've done the chatbot interface and sort of talk to it, Whereas some apps will just let you flat out write the system message yourself.

Okay, so just so we're crystal clear, OpenAI does have this option for anyone to create a. What's called custom GPT, I think it's called. Right. And the key distinction between. I think what we're talking about right now, which is creating these Persona engineering concepts, is that we're not just asking it to behave in a certain way, we're asking it to interact in a certain way. Is this the distinction? Because I would imagine I could still say you are an expert copywriter and here's what you're really, really good at, and here's the copy, and I want you to go ahead and create the output. Would that still work? It just wouldn't be a system level command.

Yeah, Again, you could do it in the form of a prompt itself, but again, the AI, a system message is the lens which it views everything through. So it's being filtered through a generalist sort of prompt. Okay, so you're getting this dilution, right? If, if you're starting with like the most general knowledge in the world and then having your prompt go through that, it's. It's going to dilute it in. In some sense. This is why you see a lot of people saying, like, why does ChatGPT add all these extra. You're welcome. And here's this. I just wanted to give me the thing and not give the pre and post preamble, so to speak. And I wanted to use better language. And so for brands in particular, if you have like a brand voice that you need to stick to, this is incredibly powerful because it saves you hours and hours of having to rewrite the content in your brand's voice. So you can specifically tell the Persona or write the Persona in such a way that says, this is our brand voice, or this is your brand voice. You want to be talking directly to the AI. Your brand voice is this. You use these types of words, you use this type of language, and the AI filters every prompt through that rather than filtering it through the generic helpful assistant.

So does this actually dramatically change the output?

It absolutely does, yeah. You find dramatically different results from using system level prompts as opposed to just a. A regular prompting.

Okay, so how do we even decide what we want the AI prompt to be? I mean, because I would imagine as marketers and creators, we don't sometimes even think this Way, like, do you have any thoughts on how we can even figure this out?

Absolutely. So going back to the idea of you are delegating a task to a human assistant, right. Or you're outsourcing something to a contractor, so to speak, somebody who's very good at delegation would give them a very detailed list of their. Their job title, their job description, their job responsibilities, right? So you want to think of the system message in that terms. You are crafting a job description for an individual who's going to complete a specific task for you. And so, number one, you want to have a specific task in mind, a specific goal. What is it you want this Persona to be, right, really good at. And number two, just go through and really be as intentional and thorough as you can in describing what that job is, how that AI is supposed to respond to what you give it, and again, just fill it with as much detail as possible. The more you give it to sort of chew on and to filter things through, the better the output is going to be.

This is a pretty meta question, but is there any reason you couldn't use ChatGPT to help you create these job descriptions and these, you know, all these things we're talking about?

You absolutely can. That's the wonderful thing about AI is that it really can. You can say, help me craft a prompt or a Persona or a system message, anything you want. Help me refine this job description in a way that is concise and accurate. And one little, you know, fun thing that I tend to add to my prompts is like, is there anything I'm missing from this? And then it will sort of help you fill in the blanks too. So, yeah, absolutely. I've used. Used AI to create AI Personas dozens and dozens of times, and it really does help the refining process, especially for, you know, one. One thing that I created right away when I unlocked this capability inside of magis, I wanted to create my own brand voice, and I wanted to train the AI on my article, blog posts that I've already written so it can help me write things in my voice already. And so, you know, you just feed it articles, things you've already written and say, read all these articles, and then craft a brand voice for this author. And that is a. Has been extremely helpful because it's great at summarizing, it's great at categorizing, it's great at identifying, you know, word usage and language and putting your voice into a succinct set of instructions.

Well, and I know people that have messed around with creating their own custom GPTs, which I'm pretty sure is free, but I'm not 100% sure it's free. But I know you can upload documents to kind of help the custom GPT to kind of have a knowledge base. Right. So I've heard of authors who have uploaded chapters of old books or workbooks or articles like you mentioned, and then I would imagine you could have that person essentially model you a little bit. Right. And if you. You know how you mentioned earlier how we don't all have a lot of creative. We're not always in the creative mood. Right. Well, if we gave it our most creative content and said, you are Dustin Stout, you know what I mean? And I would imagine it would learn your creative style. And then when you're not feeling creative, you could go to your Dustin Stout Persona and potentially have it come up with creative ideas for you that are in line with all the other ideas.

Right, Exactly. Right. Yep.

That's fascinating. Okay, let's dig in a little bit more specifically on, like, applications here. Like, how are you using this, and how are some of your customers using this to actually make their job better? Like, I would love as many examples we can talk through as possible.

Yeah, I mean, first and foremost, it's brand voice. Like, we work with a lot of businesses, a lot of agencies, remarkable amount of agencies actually, who they either have a strong brand voice that they want to generate content with, keeping in mind that brand voice, or they work with a lot of clients. And again, going back to my history of working with different clients and different industries, one of the hardest things for a writer is to adopt a brand's voice. At least it was for me and a lot of the customers I've talked to. So to change your writing style and to match that of the brand is really difficult. So being able to create multiple different brand voices for each one of your clients or each one of your projects or each one of your businesses, that is an essential. It's like usually the first Persona people create is they craft that brand voice. Persona.

Are you feeding examples of what the brand voice sounds like? And if so, like, how many examples are we talking about here?

In many cases, yeah. If you have examples of stuff you've written, social media posts, web pages, articles, white papers, anything that you can feed it to sort of help it craft that style and match it, the better. So, you know, a lot of people, that's what they do. They'll feed it these articles and then they'll just have the. The AI summarize the brand voice or the writing style into a succinct set of instructions. Other people don't have one. They don't have content. And maybe it's a new business that they're forming. And so they're still kind of creating that brand voice, so to speak. And so the AI is really good at. We actually have a brand voice generator Persona that, that I created to help brands refine their brand voice and to actually do that from scratch. A lot of other use cases, they have a brand book. They have, you know, some very detailed things that they want. And so they'll feed that brand book to the AI to come up with a set of Persona instructions.

On the brand voice thing, let's say that you are working for the company and you've developed this Persona, which is my company's brand voice, and you write something. Could you ask the Persona to refine it to make sure it is in alignment with the brand voice? Do you understand what I'm asking?

Absolutely. Yeah. Yeah. So sort of use it as a check and balance or as an editor in chief. Yes, yes, yes. Yeah, I use that all the time. So one of the great things about working with multiple different AI Personas in the same conversation is you get to kind of play them against each other. And that's one of the things that I really found a lot of use in and a lot of our customers are raving about is they maybe start with the expert copywriter Persona, have it write some expert copy, and then switch over, almost like pulling another expert into the conversation and say, hey, editor in chief of brand voice. Review all of this and rewrite it so that it matches our brand voice or tell us where in this it doesn't match our brand voice or tone.

Fascinating.

So being able to sort of play these different Personas against each other is a. Is another level of, you know, really getting the most out of it.

That's really quite fascinating. So that's okay. Brand voice. You're about to go on to some other examples.

Yeah. So brand voice, another one that I had a lot of fun with. And sometimes I just want to write, you know, funny content. So I created a stand up comedian Persona, one of my favorites. It's very simple. I just say, you are a stand up comedian. You make people laugh with your satire, jokes, humor. You answer everything in a humorous way to cheer the user up. Use satire and make fun of everything the user says in a positive way. So this is great for just, you know, having a little fun. You know, sometimes having a crappy day, right? We're entrepreneurs, we're on a roller coaster. And some days I'm just like, I need to vent and have somebody say something funny to snap me out of it. And so I'll pull up this stand up comedian and have it, you know.

Okay. I have to ask a question I'm sure is on the mind of a lot of people that are listening right now. ChatGPT right now allows you to have all these different threads, right? And it saves them all on the side, almost like they're almost like little documents, but they're not really documents, I would imagine. You can correct me if I'm wrong that you could try these prompts with ChatGPT. It's just that it's eventually going to lose its Persona. Is that right? Or I mean, like if you didn't have access, for example, to a tool like Magi, or you didn't have access to custom GPTs, because maybe in your, I don't know, for whatever reason, you just don't have it, you would probably have to prompt it every time with all this information. And this is. Yeah, right, like explain that a little bit. Yeah.

So there's actually two things you just identified that is a big problem with just using prompts outright. So one thing people don't realize is the AI has a memory limit. Every conversation you have with an AI model, whether that's chat, GPT or Claude or Magi, you will run out of the space in the AI's memory limit, so to speak. So think of it as a window. You can only see what's within the window and anything that's outside that window you can't see anymore. Right. The AI's ability to process text, it works in the same way. It can only process X number of characters in text. And so eventually, you know, that conversation starts to leak out the top, so to speak. If you're using a prompt, just a regular in chat prompt to tell the AI what it should do or what it should be or the language that it should use. You give your Persona instructions in that way. Eventually that prompt is going to leave the context window and the AI is not going to remember it, so to speak. And so your content will start to get diluted even further. Whereas if it's a system message, a properly inserted Persona, that system message never leaves the AI's memory. It is always first and foremost added to what the AI can see in the conversation. And number two, I guess we did already cover this, but you know, it's the most important instruction. So Everything will kind of be filtered through that and you won't need to reprompt it. And being able to save that Persona and access it from any new conversation or add it to any old conversation is definitely helpful because you don't kind of have to start from scratch all over again each time.

Well, and if we start getting creative, Dustin, and you and I are both creative, you're way more creative than I am. But I start thinking about all these possibilities. Like imagine I created my strategic advisor Persona, right? Where whenever I have really deep things that I need to process, I can ask it. You are a high end business consultant, right? And your job is to help make strategy for businesses. And your expertise are in these, these areas like growth and all these kind of things. Right. Then I could go to it as if I was going to a confidant and I could ask it specific advice and questions, could I not?

Absolutely, yeah. Again, filling it with as much context as you need for that given Persona is super helpful. You can give it details about your business and, you know, details about you personally that it would just always be aware of. And I actually have a business advisor Persona that I use.

Really?

Yeah, I do.

And how is it, how good is it in your opinion?

It's not perfect, obviously. And it does tend to, you know, once you use it a lot, you find the AI kind of, kind of tends to repeat itself. But for very specific questions. When I have a new nuanced question, it is really good at like helping me think outside of my own confines of how I usually think about things, you know, and that's really the big unlock again, going back to the creativity aspect, right. We have certain limits to our creativities. We have certain perspectives that we've developed over time. We have certain experiences that have shaped how we think about things, whereas the AI doesn't necessarily have those limitations. And you know, therein lies our strength and our weakness. AI will never have our combination of unique experiences and talents. But in the same way that means it could think outside of those typical thought patterns that we have. And so, you know, something like a business advisor will be of great help in helping me think outside of my normal thought patterns.

One of the things that we were talking about when we were prepping this is you can, when you do this system level Persona engineering, you can specifically ask it to ask questions of the users. Right. Can you talk a little bit about how that works?

Yeah, it's, it's one of my favorite tricks. So when I don't really know what I need to prompt it with or what information I needed to give. I, I will straight up ask it. Like, ask me all the questions you need me to answer in order for you to do this. You know, a great example would be coming up with a brand voice. Right? So I, I didn't necessarily know how to craft a brand voice. Many businesses don't. So I said to the AI at the beginning, I said, you know, I want you to help me with crafting a brand voice. So you give it sort of the instruction at first, and then I say, ask me all the questions you need me to answer in order to craft a brand voice for me. And that way you're prompting it to prompt you.

It's almost like programming, isn't it? Right. You're giving it permission to ask clarifying questions, which it will never do unless you tell it to do it. Right. That's really important distinction for people to think about. Right. This allows it to collect more information so that it can do its job better. Yeah, right.

And I'll give all the listeners a bonus pro tip. So take it one step further and instead of just saying, ask me all the questions you need me to answer, add this to the end. Ask me each question one at a time. I will answer the question and then you'll ask me the next question. We'll continue this pattern until you have all the information you need. Now, why does that have like a magical addition to it? Here's why. Every time you feed the AI another prompt or another answer, it's continually getting smarter. So as you answer more questions, that's going to continue prompting you with new questions. And that added context each time is going to make the AI's output better and better and better. So the questions are going to get better and then thereby your answers are going to get better. And it might think of some more questions based on your answers. So by having a conversational list, the AI by default is going to want to give you all the questions at once in one output. It's going to give you a big list of questions and you can go through and answer them. But if you use that back and forth technique, you're actually informing and making the AI's responses and future questions even smarter and even more laser focused on what your outcome is.

This, first of all, is really, really cool. I mean, like, I learned a bunch of stuff today that I did not know and I'm sure a lot of our listeners are like really excited about this. Dustin, if people want to connect with you on the socials what is your preferred social platform and if they want to check out your software or magi, where do you want to send them?

Yeah, so I'm at DustinWstout on all the social things. Find me on your favorite one. I'm there and you can find out everything about MAGI at M A G A I co Magi co Dustin, thank.

You so much for answering all my questions today.

Thank you Mike. Anytime.

Hey, if you missed anything, we took all the notes for you over@socialmediaexaminer.com A2A as in AI2 as in, well, second episode. Be sure to follow this show on your favorite podcast app by going to that, for example, on the Apple, clicking that little checkbox so that you get all of our future episodes coming in on your podcast player of choice. And would you please let your friends know about this show? You can tag me. I'm Elsner on Facebook, I'm ikestellsner on Twitter X. And do check out our other shows. We've got the Social Media Marketing Podcast, the Marketing Agency show show, and the Social Media Marketing Talk show. This brings us to the end of the AI Explored podcast. I'm your host, Michael Skelsner. I'll be back with you next week. I hope you make the best out of your day and may AI help you become more successful. The AI Explored podcast is a production.

Of Social Media Examiner.

AI Business World early bird pricing ends on January 16th. You can save $300 on your tickets. Register at AIBusinessWorld live today.