Loading summary
Transcript67 lines
- [00:01]
Michael Stelzner
If you're like me, you're feeling a little bit of overwhelm right now with AI and everything that's going on. The good news is I have a solution. Social Media Marketing World 2025 has an entire ticket dedicated to AI marketers just like you. Imagine learning how to create three months of high converting content in a fraction of the time, or discovering 25/AI tools that will revolutionize your workflow. I've recruited some of the top experts in the world, AI strategists who will demystify everything from AI image creation to advanced data analysis. Here's what attendee Sarah Heilbronner said. Quote the AI speakers were amazing. Every bit of information they shared was completely new and valuable to me. Unquote. Don't get left behind in the AI revolution future. Proof your career by visiting social media examiner.com/aicon and grabbing your AI ticket today. Don't delay. The event takes place March 30th. I can't wait to meet you there.
- [01:09]
Mike Alton
Welcome to the AI Explored podcast, helping you put AI to work. And now, here's your host, Michael Stelzner.
- [01:20]
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 Mike Alten and we're going to explore how to create an AI assistant, someone who can help you in your work. Someone that can kind of be your CO buddy, if you will, that can give you another perspective. I think you're going to absolutely love what we talk about with Mike. If you are not yet using AI to help you do your work and you kind of like the idea of having another brain that you can spar with or that you can help with your work. Mike's going to make it very easy for you to understand how to do that today. Also, if you are new to this podcast, be sure to follow this show. We've got some great content coming your way. Let's transition over to this week's interview.
- [02:17]
Mike Alton
With Mike Alton, helping you simplify your AI journey. Here is this week's expert guide.
- [02:27]
Michael Stelzner
Today, I'm very excited to be joined by Mike Alton. If you don't know who Mike is, he's the founder of the AI Hat, a consultancy that helps St. Louis businesses put AI to work. His podcast is the AI Hat and his ebook is called the AI Marketing Primer. Mike welcome to the show. How you doing today?
- [02:45]
Mike Alton
I am doing fantastic. I am so thrilled to be on here with you. You and I have had very similar journeys, which I know we're going to talk about, but it is exciting just to have this conversation with you.
- [02:54]
Michael Stelzner
Well, I'm stoked that you're here today, and Mike and I are going to explore how to use AI assistance to improve your work. But before we go there, Mike, tell us a little bit about your story. How'd you get into AI?
- [03:05]
Mike Alton
Yeah, so, as you know, I've been in digital marketing for over a decade. I've been blogging about it, writing about it, and I tell people all the time. One of the things that I'm used to doing is telling businesses not to jump on bandwagons, not to jump into the latest social network. And I was treating AI that way in 2022, 2023, I was keeping it at an arm's distance. But then late 2023, probably November ish or so, I was listening to a mutual friend, Paul Raitzer. He was talking about AI, and he was specifically talking about something that Sam Altman, the founder of OpenAI, said, which Sam implied that 95% of marketing jobs are going to be replaced by AI in the coming years. And that stopped me in my tracks. I was immediately afraid for myself. But I was also af, you know, mutual friends and colleagues in this space who I knew. Many of them, like me, were not thinking about artificial intelligence. They were not thinking about how they might need to upskill or reskill or transition their roles. So I started reading more, started listening to more podcasts, started writing more. I started to devote myself to creating content. And by mid 2024, launched a whole new podcast and have had a really great time talking to, gosh, 60, 70, 80 people, experts and fellow marketers like yourself, who have really embraced AI and really illuminated for me what AI can do to help me and the other folks that I'm trying to help.
- [04:32]
Michael Stelzner
So kind of bring us up to speed on what you're building now. What are you doing these days?
- [04:36]
Mike Alton
So I started off by publishing a lot of content to the social media, which is the blog that I've had forever. And at first I was talking about how I was applying AI to marketing and I was bringing other marketing folks on. And I decided over the holidays, late 2024, I needed to really refocus myself on helping businesses here in St. Louis, where I'm at, and go beyond just pure marketing because I saw a lot of businesses that had opportunities to bring in AI to make themselves more efficient and increase their ability to compete in their market. And there wasn't a lot of people here in St. Louis helping those kinds of businesses. So I rebranded the podcast, I launched the new site the AI Hat, and that's one of my focuses.
- [05:22]
Michael Stelzner
Right, Outstanding. Okay, so there's lots of marketers and entrepreneurs that are listening to this show and obviously creators as well. We're here to talk about an AI assistant. If it's done well, what is the upside? What is the value? Why should the people listening today consider using AI assistants?
- [05:41]
Mike Alton
Well, let me throw it back at you for a second and ask you, have you ever paid for a consultant?
- [05:48]
Michael Stelzner
Of course. Yes.
- [05:49]
Mike Alton
Right. And when you did that, how long were they working with you?
- [05:54]
Michael Stelzner
Typically, it depends, you know, but usually it's a short term deal for sure. Like most of the consultants that I've been working with are like 30, 60, 90 day kind of situations.
- [06:04]
Mike Alton
Yeah, that's been my experience. I'm sure everybody listening has been the exact same way. They've probably been in that consultant's shoes. Right. To turn around the other way. Right. We've all been consultants to others and we're not available 24, 7. Those consultants weren't available to us 24 hours a day, 7 days a week. There was a finite window in which they were available to us. There was a finite amount of information and assistance that they were able to give to us. And that's one of the things that I keep coming back to that I like the most about having an actual AI assistant, because I can turn to them day or night, any question I have. Now we're talking about, and we're going to get into the details of how to do this, but we're talking about an assistant that's been trained on our business that is like a consultant. They're an expert in the industry, they're an expert in our role. But we can turn to them anytime we want and we can ask them questions. And unlike human beings, they're not going to get impatient with us and frustrated because we keep asking the same questions over and over again. They're happy to help us, happy to re answer those questions. And this costs what, 20 bucks a month depending on what LLM and system you're using. So that's one of the things that I like the most about this idea of having an actual AI assistant. Not just an AI chat that you talk to, but one that's been customized and personalized. And for the business owners listening, just imagine if Every single person in your organization had an AI assistant walking alongside them. A consultant, a mentor that they could turn to whenever they needed help, whenever they wanted, not only to do something in their business, but when they wanted to think strategically about where they could go with their role with their business in the next six, 12, 18 months. Because I don't know about you, but most of the people at my level and below, because I'm not an executive, I work full time for to Gorapulse. I've got to just. I'm a doer. Right. I'm a content producer at Agorapulse. It's been really hard for me to think strategically about what I'm doing long term because I'm so enmeshed in what I'm doing today. But when you've got an assistant who can take some of the busy work off your plate and then sit alongside you as a strategist, that opens up so many doors.
- [08:11]
Michael Stelzner
Yeah. And I want to add some thoughts to this as well. Depending on who you are that's listening right now, you might actually be really strategic. Like, I am very strategic. It's my number one strength and strength finders. And I often tell myself this story. Well, I'm pretty strategic. I can figure it out. But the reality is that I've got so many things that I need to do as the CEO and founder of this company, and I've got all these people that work for me. I actually often am kind of out of energy when I'm needing to be strategic in my thinking. And having a sparring partner, as I've referred to it, or having a strategic ally that I have trained up that I can interface with just kind of helps me, for lack of better words. Like, it gives me like a pick me up. I don't know how else to say it. Right. It's like a boost. All of a sudden, the synapses in my brain start connecting and I'm like, oh, that's a really good idea. And I just keep working it and I find it's really valuable. My gu is when you did your rebrand, you were using AI to help you with that. Is that a fair assessment?
- [09:15]
Mike Alton
Oh, absolutely. I was having long conversations with Gemini back and forth. We were talking about my strengths, we were talking about my goals and objectives for myself personally. We were talking about the problems that the people I wanted to help. What kinds of actual problems were they having? Not things at a high level, like, oh, we're overwhelmed by AI. That's not really a problem. What the problem is they are trying to produce more with less within their business. They're trying to stay competitive and profitable in an era where that's really, really hard these days. And we had really deep conversations back and forth, and I loved it. It was so much fun.
- [09:54]
Michael Stelzner
Well, there you go, folks. The upside to having an assistant or multiple assistants, as we may be talking about today, could be very valuable. So let's set the groundwork. Like, where do we actually begin if we want to create something like this?
- [10:06]
Mike Alton
The first step, I think, is you have to understand how to have a conversation with the AI with an LLM. We're talking about ChatGPT or Gemini or Claude, something like that. And if you aren't experienced in doing that, I recommend that you use a framework and we can walk through the Rice framework, R I, C, C, E. And the concept there is that you don't necessarily have to use a framework in every conversation, but like training wheels, it helps you get started on the right path and make sure that if this is something that's new to you, talking to an AI, giving it the information it needs, giving it context, that framework will make sure that you know exactly what to give the AI through that conversation. This is important for two reasons. One, in every conversation we have with AI, whether they're trained as an assistant or not, it's still important to know what we need to tell them and what we don't necessarily need to tell them. But then secondly, we're going to talk through how to actually build your own AI assistant. And in order to do that, you have to talk to the AI. It's not going to necessarily inherently know what you want to accomplish. So you've got to do some of that groundwork.
- [11:13]
Michael Stelzner
Okay, so let's get into some of the things that we talked about when we were prepping for this was to identify some of the kinds of tasks that AI can do. Talk to me a little bit about that and then talk to me about this concept of a custom instruction.
- [11:25]
Mike Alton
Yeah. So most of us are probably familiar with some of the what I would call basic uses of AI. Creating blog posts, creating social copy, creating email campaigns, answering questions. Some of us may be even using systems like Perplexity instead of Google for search, where instead of a whole list of results, we get the answer that you can do with Most of the LLMs out of the box. But as you start to personalize and customize them, then first of all, the results that we get can be so much better. Instead of a vanilla blog post that sounds Like AI wrote it, because it's using all kinds of high volume words like Dell, that we weren't really using colloquially before. It can be personalized to use our brand voice and our brand style. And it can know not only how to speak through that blog post, but it can know who we want to talk to and help us address those target audiences, those pain points help us focus as businesses and marketers on benefits over features because it's really easy to fall back into, hey, let's just talk about how great we are. When you've got that AI, who's been trained to know what you want to accomplish and who you're talking to, it can give you those guardrails. Like we talked a moment ago about how I'm focusing on St. Louis businesses right now. Well, my AI assistant knows that. And now every time I go to brainstorm an email campaign or a blog post, it is inherently trying to interject, oh, Mike, we should talk about St. Louis businesses. We should talk to St. Louis businesses. We should find examples of St. Louis businesses that have adopted AI. That's a great reminder. It's not always appropriate. If I'm sending an email to my wider newsletter audience, most of them are not in St. Louis, so we're not going to do that. But it's still that helpful reminder that's sitting next to me digitally, right? And it's telling me, look, these are some of the things that we need to do. So those assistants can be brought on to do much higher level tasks than just create a blog post, create a tweet, right? They can help us map out entire marketing campaigns. They can help us translate content into different languages, maybe even push that content into different systems. Those are the kinds of things that I want us to be thinking about when we talk about AI assistance. Again, like I said before, those strategic partners that we could sit next to before we even get to the point of creating a blog post, let's strategize. What should our content Strategy be for Q2? Who should we be talking to? What have we done before and what worked? What didn't? How should we think about and approach then our next quarter's content? What might that look like? What are some example blog posts or what can we do that's not just a blog post? Those are the kinds of high level conversations we could be having and then drill down into the actual machinations of creating those marketing assets.
- [14:16]
Michael Stelzner
Talk to me about this concept of a custom instruction that we were exploring when we were prepping.
- [14:21]
Mike Alton
So this is where we Start to personalize the AI. You can do this in a prompt. You can open up ChatGPT or Gemini and you can tell it who you are, what your business is, what your goals and aspirations are, what your challenges is, who your target audience is. You can give it all that information in the prompt if you want to, and then have it help you draft a blog post or an email or whatever it is that you're trying to accomplish. But imagine if you didn't have to do that every single time. Imagine if you could personalize the AI so that it already knew who you were, who your audience is, what you're trying to accomplish, or any other specific things that it should know about the task that you're trying to do. So all the major large language models, ChatGPT, Claude, Gemini, they all have this ability in the settings where you can go in and you can create what are called custom instructions. Now, in ChatGPT, they're called custom GPTs. It's something you guys have talked about on the show before. In Claude, they're called projects, which I know you guys use and talk about. In Gemini, they're called gems. They all work the same way. You give your instructions, a title. You can describe it if you want to. You can even give it an icon if you want it to stand out from the others. And then you set about writing the instructions. Now, the cool thing about AI that's different from everything else we've done as marketers, technology wise, if we don't know how to do it, instead of going to YouTube or reading a book, you could just ask the AI. So if there's something we don't cover today, you can open up ChatGPT or Gemini and say, hey, I want to create a set of custom instructions. Can you help me? This is what I want want to accomplish. I've done this so many times. So in the custom instructions, you're going to tell the AI what the role is that you want it to serve as. Is it a blog writer, is a copywriter? Is it an email strategist? Is it something more than that? You are going to give it all the context it needs to know who are we writing for? What is this that we're doing right? What are the channels that we're exploring? If it's a marketing role, you know, maybe it's a product marketer. Well, then you're going to give it all the information about your product, links to the website, all the features that you have, your pricing, you're going to give it all that information so that it knows and then you can set about giving it any other specific instructions. So these can be used for very specific tasks. Like, I have a podcast, so I have a showrunner GPT for every single one of my shows. I've got four with AgoraPulse and then my AI show. And each one of those has been trained to go step by step. We have a new guest, what's their name? I give it their name and then it asks me in return what's their LinkedIn. And I do that and we go back and forth and we build together a pre show document and I programmed those steps in the GPT in the instructions. Or you could just have it be more general. Like, I have an AI chief of staff that knows everything about my business and what I want to accomplish and it just has been built with that information and I can ask it anything I want and it can be a strategy assistant to me and a consultant.
- [17:21]
Michael Stelzner
What I love about this Mike is that. And we're going to get into the tech stuff in just a second, folks, and we're eventually going to get back to this Rice thing that he talked about. So what I love about this is that most people are using just the threads, for lack of better words, inside of chat GPT or whatever thing, and they don't really understand the value of creating what we're calling a custom GPT or a cloud project or gem. And we're going to get into this kind of functionally in just a second. But it is a absolute game changer. And I would imagine if you don't really know how to write a good set of instructions, the AI, as you mentioned, could assist you. Like, the idea of a chief of staff is cool and like, this is what I want everybody to start thinking about. Like, you could create an assistant for quite literally any role you can imagine. And if you've never had that kind of role, is it true that you could have AI write the description of what something like that is like it is, right?
- [18:13]
Mike Alton
100% absolutely.
- [18:14]
Michael Stelzner
Cool. All right, so now let's talk about the tech stack. Now let's talk a little bit about the platform. You kind of already hinted that you're using Google Gemini, so let's kind of dig in on that because I've not really dug in on that with any of my guests before.
- [18:26]
Mike Alton
I love Google Gemini. I'm literally using it every day, similar to how other people would be using ChatGPT every single day. And as LLMs go, it is very similar to ChatGPT so if you're already using ChatGPT a lot, don't change. There's no reason to. But if your organization is a Google workspace organization, you're already paying for Google, then you should be using Gemini. You don't need to pay extra for OpenAI's ChatGPT because Gemini does almost all the same things that ChatGPT does. It's very similar, I think, to social networks. You and I, Mike, we've been watching social networks evolve over the years, and, you know, we go back five or 10 years. They were all very specific in what they did. Instagram, you posted images, square images, and that's it. Twitter, you posted text messages, and that's it. And I've been saying for years, you know what? They're all going to become ubiquitous. They're all going to have video, they're all going to have multiple image formats and live videos and communities. And sure enough, they pretty much all do. The AI LLMs are the exact same way. So don't get enamored with one particular feature. Like OpenAI recently released Deep Research, which was cleverly named after Gemini's Deep Research tool. They're very similar. OpenAI has some additional features and benefits, which is pretty cool. But if you're already using Gemini, don't worry about it. If you're on ChatGPT and you see something that Gemini can do, don't worry about it. OpenAI is going to release that soon.
- [19:51]
Michael Stelzner
Let me just make the case as to why everybody ought to pay attention to Google. First of all, the concept of a GPT was invented by Google, okay? They were the ones that wrote the original white paper. OpenAI was the one that went and exploited it, and Google didn't know what they had. In addition, Google is a massive company that owns YouTube, and everybody forgets that, okay? Third, their Gemini model has been trained multimodal from the start. What that means is it's been trained to see, to hear, to speak, and all the different mediums, right? So even though today it seems to be just text, I was listening to an engineer at Google, and the new model of Gemini that's coming out called 2.0, is going to have the ability to do audio, video, images, text, all the things, okay? And there's a lot of smart people that we follow, like Chris Penn, who said, if I had to put my money on something, I'd be putting my money on Google. We might even look back a couple years from now and say, oh, do you remember that thing? ChatGPT? So how do we get to Google Gemini and let's talk about what the heck a gem is.
- [20:57]
Mike Alton
Yeah, so it's just gemini.google.com. you'll also increasingly see to your point. When you're in any other app of the Google ecosystem, you'll see the little bluish star thing in the icon. That's Gemini. And they're increasingly building it into other apps. It's still limited. It's not like you can just click the Gemini and say, I want you to do this. That's not necessarily true today, but there are certain things that it can do and I love that it's integrated. Just the other day I was in Gemini and I wanted to come up with a playlist to play during a workshop that I was doing. And there was going to be like a 10 or 15 minute break where people were literally going to be workshopping. And I just wanted kind of AI inspired music to play while they were silently doing their stuff. And so I asked Gemini for some songs that kind of made you feel like artificial intelligence in the future and that kind of thing. Not only did it give me a list of songs, it automatically, without me asking it, I didn't even know I could do this. Created the playlist in YouTube for me.
- [21:54]
Michael Stelzner
Oh, wow, cool.
- [21:55]
Mike Alton
It gave me the link in the chat.
- [21:56]
Michael Stelzner
Wow.
- [21:57]
Mike Alton
And there it was on YouTube. I was like, oh, that is pretty cool. And we'll keep discovering those kinds of things. So you open up gemini.google.com and in the left sidebar, this is going to be very similar to ChatGPT. You'll see a link towards the bottom that says Manage Gems, I think is what it says. And you click there and that's where you can create a new gem of your own. Now, here's one of the big differences and distinctions between ChatGPT and Gemini. ChatGPT has an entire GPT store and you can go and you can see other GPTs that people have created. Some are free, some you can pay for, and you can also share GPTs that you create with the rest of your team. You can't do that yet with Gemini, but I think that's another example of something that will probably change fairly soon. So when you go to Manage Gems, you'll only see your own, and they're completely private and exclusive to you. But that's where you would create a new one. And again, you just put in the name of that gem, whatever it is you want it to be, and you'd start to fill in the instructions and you could already have another chat going on with Gemini about how to create the gem that you want to create the custom instructions and then just copy and paste whatever it comes up with with into this new window.
- [23:06]
Michael Stelzner
Are your competitors already using AI to create better content and faster? Are you struggling to keep up with the pace of the ever changing AI landscape? If so, Social Media Marketing World 2025 is your solution. Our dedicated AI track brings together the brightest minds in AI marketing to help you master the latest and greatest AI tools. You'll learn from experts like Matt Wolf who will share 25 AI tools to transform your content creation forever. Chris Penn will show you how to use AI as your second brain for analyzing data and building winning strategies. Molly Mahoney will reveal how to quickly create months of high converting content with AI. As one attendee Melanie Miller said, quote the AI teaching was mind blowing. I realized had I not gone there would be a good chance my business would have been obsolete. I'm so far ahead of so many people because Michael Stelzner brings only the best to teach. Thank you Melanie. The most affordable way to attend Social media Marketing World 2025 is with our dedicated AI ticket, which gets you the AI track and access to all of the keynotes in San Diego this March. Visit socialmediaexaminer.com Aicon today and future. Proof your marketing career with our AI Ticket. Don't delay. The event starts March 30th. I hope to see you there. Is it functionally very similar to cloud projects or custom GPTs? Like, are the fields virtually identical? You're just putting in like the various things. Like once you create the gem, you're giving it a title, and I would imagine you're giving it some sort of a description, right? And in the description is all the instructions that we've been talking about that you've referred to as custom instructions. Is it that simple? Is that all there is to it?
- [24:49]
Mike Alton
Yeah, basically. Two other caveats I would add. First is that I'm sure that they all have a file field option and they're going to call it different things, but this is basically where you can attach files to the custom instructions. And this is important to know because if you're creating a custom instruction that has to do with creating marketing copy, for instance, well, then you should attach your brand style guideline and your voice guideline in a PDF that talks about your target audience or your target icp. That way it's automatically informed by that information. You don't have to copy and paste that data into the instruction field, just attach the files. But one of the big difference is because I know you guys use Claude. You talked about Claude. Claude doesn't have Internet access, so you'll want to think about that before you start to dive into creating Claude projects. Because depending on your need, it might not have that ability to access the Internet that you might want for that particular task.
- [25:43]
Michael Stelzner
And I'm sure it's coming out, because I've heard from their founder that it's coming out. But what I'm curious about is, can you attach a Google Doc to it? And if the Google Doc changes, is it smart enough to know to reference the document? Does that make sense, what I'm asking?
- [25:54]
Mike Alton
Yes.
- [25:55]
Michael Stelzner
Okay, talk to me about that.
- [25:56]
Mike Alton
So it's not going to automatically know that you changed the document. Right. But if you've got a document, a living document that has information about your business or clients or whatever the case might be, and then you go to use your gem in a conversation, it's going to reference that living Google Doc, whether that's a sheet or a presentation, slides.
- [26:18]
Michael Stelzner
See, that's where the magic happens, because this is where I could see it being really valuable. I know that Google integrates with Claude as well, but I have not experimented with it. But the idea that there could be a living set of data in a spreadsheet or in a Google Doc that the gem is essentially referencing. Right. And if those things are dynamically being updated, all of a sudden, that's kind of magic sauce for any small business, because I'm guessing 95% of small businesses are utilizing the Google workspace ecosystem. So when you start. Start adding in all that stuff to it. I mean, that's kind of magical.
- [26:54]
Mike Alton
Yeah. The trick is you need to tell the instructions what it is, because you're not going to attach a Google Doc. You're going to put the link right in the instructions, because it's just a link and you would want to tell it in the instructions. This document is our brand voice, or this document is everything I need to know about my clients or whatever. Or this sheet is all of my past podcast episodes. You know, whatever it is that you're trying to attach. That way it knows when to reference that in the conversation that you end up having in that chat.
- [27:27]
Michael Stelzner
Interesting. So why are you using it more than chatgpt? I'm just curious, like, what is it about the gems that has gotten you all in outside of just using ChatGPT? Is it just the fact that it's all in one system, or is there something more beyond that?
- [27:40]
Mike Alton
That's a big Part of it, I'll be transparent. I started using it because I was doing an AI workshop in Philadelphia back in November with my partner Chris Carr. And Chris is a huge, heavy ChatGPT user. He's in there all day. He's doing some really fantastic advanced automations with ChatGPT. But we were teaching a room full of business folks from HR and all over the departments, and some of them were using Copilot, some of them were in the Google Workspace environment. And he said, hey, Mike, can you learn one of those other systems so at least one of us can speak to Gemini and Copilot? So I said, sure, I'll use Gemini. Turned out I was already paying for a Google account that I just needed to change it to a Google One AI account. And now I had full access to the most advanced Gemini model. So that was the reason why I dove in to begin with, and I immediately just fell in love with this system. First of all, Google knows what they're doing when it comes to UI. So even though it's similar to ChatGPT, there's just some subtle differences that just make it a little bit nicer to use. And then I started using NotebookLM, which I know you guys have talked about on this show in a past episode. Folks definitely listen to that, if you haven't already. And I started using the original Deep Research, Gemini's Deep Research. So for folks who don't know, this stuff is freaking amazing. You open up Deep Research, which is just a toggle in the chat, so you could use one of the regular models, 2.0, 1.5, or just choose Deep Research and you tell Deep Research, hey, I want to know everything about the businesses in this addressable market, for instance, or I want to know everything about something else. And you give it that research topic. You go in as much depth or as little depth as you want, and it's going to read what you said, and it's going to come back with a research plan that you can review, it's going to repeat what you said, and then it's going to add some other ideas that it had based on what you asked it to do. So I did this over the holidays last year, 2024, when I wanted to rebrand, and I live here in St. Louis. I've lived here since 2007, but I've always worked remotely for Agorapulse and then a different remote company before that. So I've never not plugged in to the St. Louis area. I don't really know St. Louis businesses and St. Louis industry. So I use deep research and I said, hey, we're going to start working more in St. Louis. Please help me understand, do some research and understand what are the top Industries in St. Louis. What are the businesses, some example businesses that work in those industries. What are some of the challenges that they're having that AI could resolve? What are some of the associations and organizations that represent those businesses? And it repeated all that back and then also added things like where these businesses are located. It said the top 10 industries. There were a couple of things that it thought, oh, I should probably research this too, based on what Mike's asking. Things that was like, oh, wow, yeah, that's a really good idea. You should definitely research that. And then it went off and scanned over a hundred websites, scraped them all, pulled all that information back in. It took about five to 10 minutes and then eventually came back with a complete report of what the business landscape is like in St. Louis. It identified finance, healthcare, manufacturing as the top industries here in St. Louis, which is correct. And some of the top businesses in those industries, like Barnes Jewish, is the number one hospital, number one provider of healthcare, and went on to talk about all the different things that they are going through, the issues that they have as a business, like we talked about before, that aren't necessarily specific to AI. But if I want to help them with AI, I need to understand, okay, what are their challenges and pain points today that maybe AI can actually help them with. Gave that all to me in a report. Now, for those who are familiar with ChatGPT's canvas, it's going to look very similar where on the one pane, on the right pane you have the conversation that you're having back and forth, and on the right pane you have the output. So initially it's a report, multiple pages of report, including all the sources cited. But then you can have a conversation with it, you can ask questions, you can tell it to go in different directions, ask it for follow up, export it to a Google Doc, which you can then download as a PDF, which is what I did. So now every time I have a conversation with my AI Chief of staff, one of the reference documents, one of the knowledge files, is that PDF of research that we did way back when. So all of you listening, imagine you could do this not only for yourselves, but your clients and have a really in depth research report on the competitive analysis of your clients. Something that you might have spent a week or two preparing before. No, now you can do it in.
- [32:13]
Michael Stelzner
10 minutes, a couple Quick things. First of all, the other advantage to Google is it has a very big context window, which is like a million tokens, something humongous. So you can put a lot more files in there than you can, for example, with claude, which has a definite limit. Anybody who works for a business where they're hosted on the Google workspace, like my company, some of these features have not yet moved over to these workspace accounts. For example, the Deep Research, I have to go and use my personal account to be able to do it. However, Google recently announced that all these AI features are included. They just ended up increasing the price by $2 a user. So you don't need with a business account to pay anything to get access to Google Gemini. But the Deep Research stuff as of this recording, is not yet available to the corporate accounts. But I'm sure it's coming. And this is one of the big challenges that a lot of us are like, why do we have to have a personal account pay for that to get access to these things? But the good news is, like, it's moving really, really fast, and I'm really excited to see what's happening there. All right. Earlier, Mike, you talked about this Rice framework. Yeah, talk to me a little bit about that, because I think that will be helpful when people are trying to put together kind of the instruction set on how to optimize their assistant.
- [33:23]
Mike Alton
So these are essentially training wheels, right? When you're first starting out, folks like you and I who have been using AI constantly, we're no longer thinking in frameworks, but if you're not familiar, the idea here is we want to make sure that we're giving the AI all the information in the prompt that it needs to proceed and give us a great output. So instead of going back and forth because you realize you forgot to tell it something or you weren't happy with the original output because it's just plain vanilla, you can address that. So the R in rise stands for role. This is a really interesting concept that has evolved with artificial intelligence. This idea that if we tell the AI that it's acting like a role, like a job, that it helps it and it really works well. And this is because while the AI doesn't have the entire sum of human knowledge, it essentially might as well have the entire sum of human knowledge for all intents and purposes. And within all that knowledge, a small subset is what you need for the task at hand today. So let's say you want to write an email campaign to your audience. Well, you want the AI to Help you as though it's this expert email strategist, right? Somebody who's really familiar with how to create terrific email campaigns. You don't need it to act like a lawyer. So let's just tell it. Don't think about that stuff. Just think about what it's like to be a really smart expert email analyst. And I'll tell you, there's actually studies that have shown that there's a difference between act as the role and to tell it to simply be. Be. You are the world's best email strategist. That's the role you want to give that at. And you can give us some additional information if you want, but that's what it boils down to. What is the job that you're asking it to do? Then the next is the instructions and the information you want to tell it, what it is that you're trying to accomplish and what you expect to get out of our little session. Today I am doing an email campaign. I want to reheat my audience. You know, a good portion of them haven't been opening emails for a while, so I want to target them and get them to re engage with my email overall so that they're now opening my newsletters and that sort of thing. And Mike, I know you guys have been having issues with email, so I'm sorry if I'm triggering you with this example.
- [35:35]
Michael Stelzner
You're not. But I do want to ask a question because this is where I think a lot of people have a lot of questions. The rule thing makes total sense. Like you are a world expert in emails that convert or whatever, right? With the instruction set, there's a lot of people that do a lot of things with email. Do we want the instruction to be set at the high level or do we want it to be set at the prompt level? Do you understand what I'm asking? Because like, if we tell it its only job is to send emails to our customers and then we ask it to send emails to prospects, it's going to not know what to do. So do you find giving it a broad set of instructions at the gem level makes sense? And then to give it a detailed level of instruction at the command level makes more sense.
- [36:20]
Mike Alton
Yes. And you're absolutely right. If we were using this to create a saved set of custom instructions, I might not be as specific. Right. If I want an AI assistant in my company who's an expert email strategist and writer. Right. I'm not going to tell them probably in. In the instructions who the emails are for, because that might vary from day to day, but in the actual prompt, like, if I need to literally create this set of emails today, I'm going to tell it who the emails are for. I'm going to get very as specific as I know. Now, that might be a conversation we have back and forth, but I'm going to be as specific as I know in the actual instructions. A little bit of advanced prompting here is that I might tell it in the instructions that we're going to go step by step through this process. I might tell it what those steps are, I might not. But I want to be clear what the first step is. Because the first step in sending an email, as the marketers in the audience will know, is, well, we need to identify who we're talking to, and we need to identify, you know, what is the angle, what's the point, right? So we need to think strategically about the email campaign before we get down into drafting the email. And this is important to understand because AI is programmed to be as helpful as it possibly can. And if you just tell it, hey, we're going to do an email reheat campaign, you know what the first thing it's going to do? Write your email if we don't want that yet, because we're not ready to write the email. As marketers, we need to think more strategically. So that thought process needs to be part of the instruction.
- [37:49]
Michael Stelzner
So in this case, it sounds like you would train it, Mike, to say, if I do not tell you who the audience is, ask me. That would be one of the sets of instruction, right?
- [37:58]
Mike Alton
Yeah, that's one of the really, really core mindsets that we have to keep going, which is that we need to tell the AI to ask us questions, because it won't unless you've told it to. So I tell it to push back on my assumptions and I tell it to ask me questions to clarify anything that I've said that it doesn't quite understand. And I'll rephrase that in different circumstances. But when you can get the AI to ask you questions, first of all, it's pretty funny. It'll give you a freaking questionnaire. It'll send you 10 or 20 questions to answer. Because as you read through it, you realize, yeah, okay, if I was really, like, invested in this role, whatever this role is, these are things I would want to know who the audience is, what their past open rate has been, what their engagement has been like. Have I previously segmented my email list? Those are questions it'll ask you, okay?
- [38:50]
Michael Stelzner
What comes after? So we got the R's rule, then instructions. What's next?
- [38:53]
Mike Alton
Yeah, so next is context. This is where you're going to want to tell the AI background information. So if you're just starting in a prompt, it's going to need to know, okay, who your business is, who your target audience is, those kinds of things. That's the kind of context it's going to need to know. If you're building a gem or a chat GPT, do the same thing in there. Use the same framework there, Give it that context. Say, hey, look, I've got an entire document prepared for my target audience. You'll find it attached in the attached files. Tell it that. Don't just attach it and assume it knows it's there. It'll read it, but it might not know what to do with it. So you tell that that in the context area of the instructions or the context of the prompt. Now, if this is your first time going through this exercise, just do this in a basic prompt, but later on you can build this into custom instructions.
- [39:42]
Michael Stelzner
Okay, keep going with this RICE framework.
- [39:43]
Mike Alton
So the next is constraints. And this is where you might want to add some specific boundaries that you want it to employ. And there might not be any, but maybe your emails are typically short, so you might give it a constraint of, hey, just so you know, when we get to the email creation phase, it's got to be 500 words or less. Or maybe you're like me and you want to have a long form stuff, because that's pretty much all I put out. Say, hey, we're going to do it. This is going to be a big deal. This should be at least 1500-2000 words. Because I want to speak at length. Give it that constraint so that it understands what your preferences are and any other constraints that you might want to give it. We're only English, we're only talking to our North American audience, or we're talking to a global audience. Anything that you can think of that again would help the AI come back at you with really great information. And again, you could do this in the prompt level or you can do this in your saved instructions, whichever makes sense. And the last, the E in RICE is evaluation. And this is almost more for us. It means that we need to know what success looks like. We need to know what good looks like. And you might tell the AI that. You might give it an example. Sometimes people use E, for example, there. But it also just means, you know, you and I, as, as we're creating the prompt or we're creating the custom instructions. We need to already understand what we're trying to get out of it. What will good look like so that we can evaluate the responses. Was this a great email or is this just a ho hum email? And so that requires a little bit of domain knowledge that might require some research if it's something you're not really experienced in, but that's important to know. Otherwise you might be tempted to just go with whatever the AI gives you as outputs. And it's probably not very good. Good.
- [41:23]
Michael Stelzner
Yeah. Some of the things that I've noticed mostly working inside of Claude is constraints is really important. Like I could tell it look, don't overuse emojis. Like that's an obvious one that you could put it as a constraint or avoid some of these kinds of words. Another thing that I really love doing, which maybe falls under context, but might fall under this evaluation slash example, is to give it examples of the best performing work that we've done, especially when we're writing, and then ask it to model or mirror that style. I found that to be exceptionally valuable when you're actually creating content. Now when you're actually. This is where it gets interesting, when you're creating a chief of staff, which I want to dig into a little bit here. What if you don't know what kind of constraints to give it or even how do you evaluate it? Because you've never had a chief of staff. Can you talk to us a little bit about what you did there?
- [42:15]
Mike Alton
For my AI Chief of staff, this is an AI assistant that I can turn to whenever I have questions, ideas I want to brainstorm for my own personal side hustle business. The AI hat and I have trained it to understand my target audience, understand my voice and style. It knows I like Star Wars. It knows I like history references. It knows all that. But it also knows because we had a conversation back and forth about this. It knows what my goals and objectives are for the coming year. It knows where I want to be by the end of 2025 and into 2026 with this business. It knows I want to help St. Lou businesses, like I mentioned before. So I gave it a role similar to my own, that you're an entrepreneur and you're a content creator and you're an AI enthusiast who wants to help other businesses embrace AI and not be left behind. You're an historian who understands that we're going through the fourth Industrial Revolution and this is different from say audio social like clubhouse. This is not something people can ignore. And then I filled in all the other details about me and I, I had a conversation back and forth with Gemini before I got to actually building the instruction set. And that's one of the beauties of AI, right? You can go into it and have a back and forth conversation. At the end of the conversation, then you can say to the AI, now that we've had all this, give me an output, tell me what the custom instructions should be to create this AI assistant. So one of the things, Mike, that I did this week, that was after you and I talked. So this is new to, to you, but I've been thinking a lot about this. How could I help other people go through this themselves? So I actually wrote an article and I called this your AI work body. Because maybe you're not a founder or a CEO. And so this idea of a chief of staff kind of all will sounds kind of grandiose. Our friend Paul Raitzer did a whole webinar about a co CEO. Well, if I'm not a CEO, why would I want a co CEO? Maybe my job is copywriter, maybe I'm a social media manager, something like that. And I thought, well, what if everybody just had a workplace work buddy based in AI? What if everybody had an AI assistant that they could go to that is like them in their role. They're trained, they understand they have the same goals, the same objectives, they understand their place in the organization, they understand what the organization is. And they can then turn to that AI work buddy anytime they wanted to and brainstorm, ask questions, help them prioritize that day's tasks, help them create copy if that's what they want to do.
- [44:48]
Michael Stelzner
Help.
- [44:49]
Mike Alton
Heck, they could even turn to the AI for empathy, which is super interesting. You could turn to the AI and say, you know what? I am just having one heck of a day. All this stuff has happened. And you could train it if you wanted to, to be empathetic and understanding and supportive. So I created an article. It's just theaihat.com aiworkbuddy all one word. And it helps you understand what this is and it gives you a prompt, just a single one paragraph prompt that you can put into Gemini or ChatGPT to get the ball. And then there's a PDF you can download. So I created the instructions that you can give to the LLM that will tell it what to ask you to create your own unique AI work buddy. So it's going to basically do that questionnaire one question at a time. It's going to ask you okay, Mike, what's your job title? What's your business? What's your role? What are your objectives? What are your key metrics? What are your OKRs? What's your structure in terms of your team? Who do you report to? Do you have people who report to you? And then they'll start to ask you things like, how do you want your AI work buddy to communicate with you? Should it be, you know, buttoned up and professional all the time? Should it be humorous, witty? And it gives you a lot of these examples. And like any AI, if you're not sure, you can have a conversation back and forth. And once it gets done asking you all those questions, it's been programmed to just give you in a complete summary. One Prompt output. Here are the custom instructions that you can just copy and paste into your new AI work buddy for whatever role you guys are in.
- [46:25]
Michael Stelzner
That's freaking cool. Where do we get that? Say the URL again, just for people and we'll make sure to put it in the show notes.
- [46:30]
Mike Alton
Yeah, so it's theaihat.com aiworkbuddy all one word. And I was talking to Chris Daigle about this on his podcast, and we were talking about, about this idea. Well, what if every single person in your organization had their own AI work buddy? What if everyone in the organization had an AI assistant, personalized, customized, and trained on their role and your business, that they could turn to this idea of a copilot, which I laughed because this is really what Microsoft Copilot should have been, because the branding of copilot is genius. That's absolutely what their AI assistant should be. But the rollout has been terrible. Nobody knows how to use copilot, and so they're rolling it back. But the this is what it should have been, right? That personalized assistant that's paying attention to what you're doing, helping you step by step every day, having all those conversations and then back and forth. It knows who you are and what you want to accomplish, and it's that assistant you can turn to. Just imagine if everyone in your organization had that today, and if everyone in the organization was increasing their efficiency and productivity by 10, 15, 20% minimum. I mean, bare minimum, right? The whole organization is going to benefit tremendously. And to your point, if you're using Google Workspace, it you what, an extra two bucks a month? If you're using chat, GPT is 20.
- [47:49]
Michael Stelzner
I mean, it's included. You don't have an option. So it's just standard. You know, it's like, hey, we're increasing your price, but it's included. It's free. It's not really free, but it's there. It's free. Mike, this has been amazing. If people want to connect with you on the socials, what's your preferred platform? And if they want to reach out to you and do work with you, where do you want to send them?
- [48:07]
Mike Alton
Yeah, so I'm active everywhere. The best thing to go is to the AI hat.com connect. That's where you'll find my social channels. You can pick whichever one you're active on. I'm on all of them. You can find the podcast, the blog, and if you want to actually have me come out to your business and do a workshop, I do a three hour workshop where I'll walk through what's capable and possible with AI today and then we'll spend an hour and a half just on prompt engineering and custom instructions and memory. And then we'll spend an hour workshopping individual solutions. You can find out more about it there.
- [48:36]
Michael Stelzner
Sweet, sweet. Mike, thank you again for coming on the show. This was amazing.
- [48:40]
Mike Alton
Thanks so much for having me. This has been a blast.
- [48:42]
Michael Stelzner
Hey, we took all the notes for you over@socialmediaexaminer.com a44. Be sure to follow this show on your favorite podcasting app. And if you've been a longtime listener, I would love a review on whatever app you're listening to this on. And would you let your friends know about this as well? And do check out our other shows, the Social Media Marketing Podcast and the Social Media Marketing Talk Show. This brings us to the end of the AI Explored podcast. I'm your host, Michael Stelzner. 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.
- [49:15]
Mike Alton
The AI Explored Podcast is a production of Social Media Examiner.
- [49:23]
Michael Stelzner
Are you ready to transform Your Marketing with AI at Social Media Marketing World 2025? 5 our exclusive AI ticket gives you access to cutting edge tools, proven workflows and strategies that drive real business results. Join industry experts like Matt Wolf and Chris Penn this March in San Diego. Visit Social Media Examiner.com Aicon to secure your AI ticket today. The future of marketing is waiting for.