Digital Profits Podcast – Episode 8: AI in Marketing: The Marketer of Tomorrow

Are you a digital marketer interested in transforming your business? Artificial intelligence (AI) today is leading the charge to help businesses achieve accelerated results. But what exactly does that mean for you?

Fusing AI into digital marketing can open up powerful opportunities, from optimizing search engine results to personalized recommendations. This post will take you through the basics of putting AI technology into action so you can use it as an incredible resource when talking about growing your business!

Understanding the Basics of AI-driven Digital Marketing

If you’re looking to up your marketing game, AI-driven digital marketing might be just the solution you need. With these advanced tools, you can automate your campaigns, analyze customer behavior, and create personalized experiences for each individual.

At its heart, AI is all about using algorithms to analyze data and make predictions. By taking advantage of this powerful technology, you can optimize your marketing strategy and get ahead of the competition. Once you start using AI in your marketing efforts, you won’t look back.

Leveraging AI-driven Search Results to Boost Your Rankings and Visibility

As a website owner, you likely understand the importance of having high rankings and visibility in search engine results. To achieve this, you may have already implemented several SEO strategies. However, one strategy that many websites are now leveraging is the use of AI-driven search results. By utilizing AI-based algorithms, search engines can now provide more accurate and relevant results for users.

As a result, optimizing your content for AI-driven searches can potentially boost your rankings and improve your visibility. Some ways to do this include focusing on topics that are commonly searched for using voice assistants, using natural language in your content, and incorporating structured data. Keep in mind that while AI is still evolving, it’s important to stay ahead of the game and implement strategies that align with the technology.

Utilizing Advanced Language Models to Deliver Targeted and Personalized Content

As a content creator, you know that not all content is created equal. In fact, with such a vast amount of information available online, it can be overwhelming for your audience to find exactly what they’re looking for. That’s where advanced language models come in. By utilizing these cutting-edge technologies, you can deliver targeted and personalized content to your readers that matches their interests and needs.

Whether it’s through AI-powered chatbots or expertly crafted blog posts, you can connect with your audience in a meaningful way and keep them coming back for more. So, why settle for generic content when you can provide an experience tailored specifically to your readers?

Taking Advantage of Cutting-Edge Tools to Automate Processes for Improved Efficiency

Whether you’re managing a small business or operating on a large scale, automation can help you save time, money, and resources. By taking advantage of the latest tools and technology, you can simplify tasks, minimize errors, and focus on what really matters.

And the best part? You don’t have to be a tech expert to get started. With user-friendly interfaces and intuitive features, automation tools are accessible to everyone. So, why not take the plunge and explore what automation can do for your organization? With the right tools at your disposal, anything is possible.

Monitoring Performance & Analytics With Artificial Intelligence

As a business owner, you already understand the importance of tracking your company’s performance and analyzing data to gain insight into your customers’ behavior. But let’s face it, manually sifting through mounds of data can be a tedious and time-consuming task. By utilizing AI-powered analytics tools, you can monitor your company’s performance in real-time, identify areas for improvement, and make data-driven decisions to enhance your bottom line.

Imagine having access to a system that can detect patterns, predict trends, and automatically deliver reports, all while freeing up time for you and your team to focus on other important tasks. Integrating AI into your monitoring and analytics strategy can give your business a competitive advantage and set you up for success.

Crafting a Winning Digital Strategy with AI to Grow Your Business Faster

With the right approach, AI can help you grow your business faster than you ever thought possible as it provides invaluable insights into consumer behavior, allowing you to fine-tune your marketing efforts and reach the right people at the right time. By utilizing AI-powered tools and platforms, you can streamline your operations, boost productivity, and enhance customer experiences.

With the power of machine learning, you’ll be able to analyze vast amounts of data to identify patterns and trends, giving you a competitive edge in the marketplace. With the right digital strategy, AI can transform your business and help you achieve your growth goals.

In conclusion, AI-driven digital marketing offers numerous benefits for businesses looking to grow faster. From understanding the basics of AI-driven digital marketing to leveraging AI-driven search results and utilizing advanced language models to deliver targeted and personalized content, firms now have an array of tools at their fingertips to help boost their visibility and performance levels in a competitive marketplace.

Additionally, cutting-edge tools like natural language processing can be used to automate routine processes for improved efficiency, while real-time analytics and monitoring capabilities with artificial intelligence can help shape better-informed business decisions.

When you combine all the elements of your digital strategy with AI, you’ll have access to a wealth of powerful insights that will allow you to quickly execute campaigns and maximize your ROI, thus, helping you stay ahead of the competition. With these strategies in hand, your business is primed for success as it embarks on its journey toward faster growth.

Discover how Google is using AI to streamline their search results and provide users with “zero clicks” for faster answers when you listen to the latest episode of the Digital Profits Podcast on Unleashing the Power of AI. Learn the importance of positioning your product or service in consideration sets used by AI-generated results in order to compete in this digital landscape, as well as tips to optimize your workflow and automate pieces of tasks. Plus, understand how you can leverage GPT-4 for creative purposes. Tune into the Digital Profits Podcast now and start growing your business faster!

Welcome to the Digital Profits podcast, where you’ll learn how to grow your business faster using paid traffic and SEO. Each episode will feature a breakdown of digital marketing trends and answers to your burning questions that will provide actionable takeaways to make your marketing better. So join us, Ben Page, Ray Sawvell, and Blake John, as we guide you on your journey to higher profits. Remember to join the Profit and get ready to profit in 3.2.1.
Ben Page:Hey, everyone, welcome back. Today’s episode should be really fun to record. It’s going to be a conversation about AI generally and AI in marketing. So we’re talking about the idea of the marketer of tomorrow. This is a follow up to our very first episode, which I believe is still our most engaged with episode. And a lot has changed. So I’m here with Ray and Blake.

Ray Sawvell: Hey, Ben.

Blake John: Hey, there.

Ben Page: Hello, guys. What has changed since we first recorded? And I’ll just place us really quick, we recorded in early January. I think we were using Jasper pretty heavily. We were just starting to kind of test the waters with GPT and was it 3.0, was it two?

Ray Sawvell: I think it was three.

Ben Page: Yeah. Three GPT-3. Yeah, it sounds right. Wow. A lot of change in a short period of time. So who wants to start with some observations on what has changed?

Blake John: Yeah, I’ll kick it off. I think one of the most interesting things and kind of progress to all of what we’ve seen and all the buzz that has happened over the last several months is that AI is now connected to the internet in real time. And specifically, I’ve been using Bard, which is Google’s AI or large language model, and you can do some really fascinating things. You can ask it questions about things that happened a day ago and you can get accurate responses. Whereas before, when we were using GPT, just kind of starting off just the tip of the iceberg, you couldn’t get that real time information because GPT was only up into was it 2021?

Ben Page: The data that it was yeah, the training set, I think, was up to.

Blake John: But now it’s connected real time data to the internet and it’s just fascinating.

Ray Sawvell: Yeah, I echo that. I mean, things have just advanced so much over the last three months. I just think about everything that’s happened. I mean, GPT four right now, which as of recording here in May, that’s like the latest release of GPT, but it can pass the series seven, it can pass the bar exam. It’s just got so much data and it’s trained on so much information where you can use it for it feels like endless things. You can do so much with it. And I’m excited for today’s episode to dive into that a little bit further.

Ben Page: Yeah, for sure. It’s like having an army of PhD candidates as your interns or something. Like, because of the breadth and depth of knowledge base there, but right, so we’ve got the Internet connectivity piece. Maybe we had it first with the new Bing, which, Ray, I know you used more than I did. Certainly I was having a little trouble getting it configured properly and embedded with.

Ray Sawvell: Edge and I’ve went away from it. Now I’m not liking it as much.

Ben Page: Right, well, we had that. Then we had Bard kind of like V one drop. Now we’ve got while GPT four dropped since last episode. And then within the last week, I got access to Plugins. And just literally this morning, Ray, you pointed out the toggle for web access. Right. So I’m super excited, like even this afternoon to play with it only 25.

Ray Sawvell: Prompts to play with these wisely.

Blake John: Well.

Ben Page: Yeah, I know. Seriously? Well, I hit my limit for the first time within the last week. I had never done that before. So Blake, I mean, speaking of Bard, and kind of bringing this for the moment into the realm of search marketing, talk a little bit about the Google I O update that happened, what, last week?

Blake John: Yeah, I believe it was last week, Thursday, and a huge update. Huge. This update, honestly, might be the biggest development in digital marketing maybe ever. I can’t think of anything more significant and how it could potentially impact SEO paid search, everything related to how we use the Internet because it’s going to be embedded directly into Google. And what’s happening, what they announced is they’re called AI powered snapshots. And we don’t really know exactly how it’s going to work. It hasn’t been fully released to the wild. Google just simply demoed it during the Google AI update, again, just last week. But essentially, you have a query, you input it in Google, and there is an AI snapshot above the fold with an AI generated response. And it’s truly amazing. And I think it’s the next big thing in what we do in digital marketing. And what’s interesting is there’s just so much to learn and so much to kind of uncover as this rolls out further and further. But what Google is ultimately trying to do is and they said they’re trying to reduce the work it takes for you to find your answer while in your search.

Ben Page: Right, so increase the number of zero click SERPs. Right.

Blake John: Which is scary for me. As the SEO guy, that scares me. I do think there was sort of this fear a little bit when feature snippets kind of rolled out and that kind of became really popular. And we see those every single day now in searches all the time.

Ben Page: Right?

Blake John: Or as an example, if you search for a celebrity’s height, you don’t have to go into a website to get that answer right. Like Wikipedia, the featured snippet will just show and it tell you how tall that person is. But this is now taking it a step further in more longer tail queries that previously you would have to dive into a website to get your answer. But now Google Bart is giving you potentially your answer with zero clicks.

Ben Page: Right. So for the user, from their perspective, instead of clicking through the first link on the SERP and scanning and evaluating and going back and then second result, and then you know what I mean, having that kind of fragmented journey, the idea is well, and this is a feature of Bard, right? Blake the AI snippets component, bard will synthesize this information for you and probably they’re doing some layering of, I don’t know, vector databases or something else. They’re doing what Google does best. So we went from the Internet and search to AI. And search now is what’s emergent, which is super interesting, right? And so more zero click SERPs. And for us and our purposes, what does our role become? How do we operate within that reality? And Blake, I threw out the idea of it’s as though the AI snapshot, Google is writing a research paper on the topic that you search and then it’s going to present some additional results for further review. So in a sense, like, our job is to be in that consideration set that’s used to generate the result and appear as a reference point because it’s still unclear how or if or when or these results may or may not be sponsored and how they will interact with other existing SERP features like shopping carousels or map packs or blue links or whatever.

Ray Sawvell: I know they mentioned that they’re going to have shopping pieces up there. I believe there’s still a sponsored shopping section at the top in the AI driven results. I’m not sure what they’re going to do for the other ad driven pieces, but I’d be shocked if they didn’t roll in ads in some capacity. Right. That’s how Google makes their revenue. So I’d be shocked if they didn’t do that.

Blake John: Yeah, you kind of talked about like, we don’t know exactly where this is going to roll out and where we’re going to see it. I think initially as Google begins to roll it out, it’s going to be very long tail keywords in queries, specifically not even really keywords or like sentences. One of the examples that Google included as they demoed it was the query was what’s better for a family with kids under three and a dog? Bryce Canyon or Arches? Now, that’s not really a search you would do today, which I think is what Google actually said during the demo. But with AI embedded snapshots, you can do that and actually get your answer with zero clicks, which is ultimately, I think, what Google wants to see. So that’s like one example. I think we know those longer tail queries are probably going to see a lot of AI powered snapshots. But what I think is really interesting and what we’ll only see in the future is, okay, well, Google doesn’t want to sabotage their ad revenue. We know that’s like 85% of their entire all of their revenue, which is wild to think about considering how big and how much money Google makes. So how are they going to roll this out for queries? Where there are ads, will we see it where there are ads or will it exclusively be queries with very few ads? There’s just so much unknown right now and I don’t know, it feels like there’s going to be basically a new playbook to learn that we’re going to have to sort of adapt to and figure out, like Ben, as you mentioned, how do we become a citation for this AI powered snapshot content?

Ben Page: Well, I think for me, zooming out a little bit and thinking about the Google strategy and point of view on this and this is totally a hypothesis, right? But it’s almost like this is a necessary feature. I’m expecting low monetization of the AI snapshot early on. I’d almost expect that it’s rolled out in a more progressive way, like the kinds of SERPs this appears in first will be more around fact based queries or existing featured snippets. Right. And we’ll have the current sponsored results within a SERP will probably still be there and then maybe a slow and progressive rollout of monetization within the AI snapshot, if at all. Because if you think about the landscape more broadly, it’s like I view this as a necessary ingredient. Like a it’s defensible, right? It’s good for users. It’s great for user experience and helps people get their answer faster and helps do synthesis and process more complicated kinds of queries that are more evaluative. Like, in other words, here’s who I am, here’s what’s in my consideration set. Help me weigh the pro con of Arches versus other. I think that’s super interesting. But it’s like if they don’t solve this problem fast and well, right. GPT like OpenAI might or Microsoft might or someone else might. So I think it’s almost like a necessary evolution. And even if they don’t monetize it, it kind of has to happen in a way.
Blake John: Yeah. I think they’re so incentivized to get this right because there’s a lot on the table. Yeah, exactly. For the first time ever, Google feels like they have a real competitor in OpenAI. I think that’s they what do they say when OpenAI was like released to the public or whatever and it was free initially, Google called a red alert or something. Like seriously defcon. All the board members got together and people who weren’t a part of the business were starting to come in. It was a big deal. So Google is really incentivized. They’re feeling the urgency and they want this to be a massive success because they know, like you’re saying users can go elsewhere for some of these queries like this. But Google, they want that ad revenue. They need to maintain their ad revenue. So they’re really incentivized to get this right?

Ben Page: Well, I think about well, yeah, and even data, right? I mean, they’ve got I think it’s still true. They’ve got, what, five products or five platforms with a billion plus users, right? Google search android, YouTube? I forget. Is it Maps and Gmail? Maybe something, whatever. Probably Chrome. Right? So it’s really about moat protection, data protection. It’s about survival of the ecosystem, and it needs to evolve in order to serve its users right, and keep them there, that kind of thing. But, yeah, it’s certainly interesting for us, too, when we hit on that. I mean, where should we go from here? Should we talk about how our usage of AI has changed? Should we talk about garbage in, garbage out? I think we hinted at that last time. Right. But even more, it’s becoming clearer and clearer, like the quality of the question. You ask how weighted that is.

Ray Sawvell: That makes all the difference.

Ben Page: Yeah. What direction do you guys want to take this now? There’s like infinite vectors we could go down.

Ray Sawvell: I wish GPT was here so I could ask it which direction to go, because I stopped thinking for myself, so I could just do that.

Blake John: That is sort of a side effect, right?

Ray Sawvell: What’s like GPT want? Let’s do the garbage in, garbage out route. We did touch on it last time, but I really do feel like we need to understand it’s almost like, what direction do we take the AI and how do we get what we want from it? Because there’s so many different routes to go from even jail breaking It or prompt engineering. There’s a bunch of different routes you can go from this aspect.

Ben Page: Yeah. So let’s talk about that. So, asking a better question, last time we talked about meta skills and how prompt engineering is a meta skill, and you should work on developing that muscle. So super skills, maybe we could talk about what are some of the good kinds of questions that even getting quality results from. So we could talk about super prompting. I’ve got a few ideas too.

Ray Sawvell: Yeah, I can kick us off and we touched on this last time, but it’s almost like priming or getting the AI ready to go and making sure that it understands before it gives you the output that you’re expecting. So typically, in my experience, the more detailed you are, the longer your prompt is, the more specific you are. And you make sure I think this is a key component. You make sure that GPT or the AI understands what you want it to do, and then it spits it back to you is extremely important, because if you don’t, in my experience, if I’m not asking GPT, like, do you understand what I’m asking you to do? And then it recites what I want it to do. Sometimes something gets lost in the way that I’m formulating my prompts. So an example might be, I want you to write a script on Google Ads for XYZ piece and then it will come back to me and it’ll be like, okay, I’ll go do this and this. But it might be missing a key component. So I think the key part like the way that you’re formulating your questions, the way that you’re formulating your prompts before it gives you the answer or the output you need to ensure that GPT actually understands what you want it to do.

Ben Page: So there’s a check your understanding piece. Let’s talk about the Seasoning piece if you want to call it that. Right? I like that super prompting because one way I’m thinking about this conceptually is you’re trying to be sufficiently clear in your prompt to define the level of resolution that you would like. In other words, like the training corpus of GPT four is so broad that a general prompt will yield a general kind of response. And often we have a specific use case in mind when we are prompting it. We want a certain kind of output. And so I think about it like broadly in terms of the domain, I might say I’m talking about this topic, this subtopic. Take these three facts or these three situations into account when you produce your answer. Or like I want or I’ll detail the level of specificity I’m looking for in the response, that’s cool. The output is I want a daily schedule that contains hour by hour level specifics on how to optimize my schedule or something like that. And by the way, account for time with family, time to exercise, time to do client a project and this kind of work and what. So it’s like almost like setting the resolution is another one. So check your understanding, setting your resolution. Blake. Any thoughts on seasoning?

Blake John: Yeah, well, just generally speaking, I don’t think you get very good results without Seasoning. Anytime I am working with an AI tool. If I don’t give it basically three prompts before the real prompt that I want to give it, it’s always a poor response. And so I’m always building it up. And I usually will begin with saying something like, you are X. You are a behavioral therapist that specializes in helping children under the age of twelve learn executive function skills as an example. And it’s like, okay, yes, I am this behavioral therapist. This is what I do, this is how I do it. And I’m like, great. And then I kind of take it down and here’s this other thing that you need to take into consideration. And then Ray, you actually gave me this idea. I’ll just say, do you understand? And it’ll be like, yeah, of course I understand. And sometimes actually it’ll say it spits.

Ray Sawvell: Back to you, like what you want it to say almost, or it’s way off the path and you got to correct it.

Blake John: Yeah, I had this happen to me literally yesterday where I had a prompt just like that. And it said, yes, I understand you are a behavioral therapist. I was like, Wait, no, not me. The response that I was said, oh, my apologies. I’m the behavioral therapist, which was easy.

Ray Sawvell: Very flexible going road. Yes, Blake, you are the behavior.

Blake John: I know, and it was so funny. But doing those checks and balances and making sure that it’s taking into consideration, like, the key factoids and little tidbits that are so important and so nuanced that as a human, you kind of just innately understand. But to AI, as you’re saying, it’s such a general broad knowledge that you have to really focus in and get narrow and narrow and narrow. And as you do that, you’ll finally get to the movie. Your 3rd, 4th prompt, you go, okay, it’s primed. It’s ready to go. Now here’s what I really need you to do. And then you complete that. You give that prompt, and I find a lot of success going down that.

Ray Sawvell: Route and to kind of go off like, Blake, like, your most interesting thing earlier was, like, it’s connected to web. And I’ve been reading threads online on Twitter and Reddit, like, what people are doing with the web to go with this prompt and seasoning the AI. You can now get to a point where I’m shopping for an umbrella outside of my house, like an outdoor umbrella to block the sun, obviously. So it’s like, you can say, I’m looking for an umbrella that is less than $50, and it’s this color. Go find me the best price. You can prime it that way, and it’ll spit you back the results, too. So it’s all about seasoning it and making sure it understands what you’re actually asking it, which is, I think, really cool.

Ben Page: Right? So it’s like you’re feeding it the filters and conditional logic up front. It returns something. And now like, all right, Ray, here are, like, the four umbrellas you should consider. And you’re like, cool, give me the top two with the most product reviews, five star, and then great. And then which one is the lightest or whatever you care.

Ray Sawvell: You give it like a set of conditions almost. So it’s like, here are my conditions or criteria below, and you can even do both and be boom, boom, boom. And then it’ll go out and find out what you’re looking for, which is really cool.

Ben Page: Well, let’s talk about what Blake said, because I call what you described imputing identity on GPT. So, like you said, you are a blank, and you can see there are a lot of prompts like this online, but I would say in 90% of my threads, I impute an identity upfront. Like, I have a use case, and I’m like, you are behavioral therapist, whatever. But I’ll say, like, you are life coach GPT or something like that, right? And then you tell it what it knows or what body of work or what knowledge to draw from up front. And then you launch into your problem like you are life coach GPT. You are the world’s best DA DA DA that helps this kind of person achieve this kind of output. You were trained in these kinds of domains of knowledge. You have these kinds of skills. Boom. And then drawing upon all of your experience in doing this, then I’ll talk about here’s who I am or here’s what the task is. I am this. Here are my conditions. Here’s what I’m trying to optimize for. Now give me this. But that imputed identity. Where else are you guys using it? You used it for that one, right? So it’s like you’re framing the output of GPT as like they are the user talking to a potential client, in a sense, because you’re trying to get an output that is like language that in that case this behavioral therapist would say to the client or whatever.

Blake John: That’s exactly it. I think embedding that identity into the AI makes a ton of sense because without that seasoning or that priming, the AI is quite literally everything and nothing at all.

Ray Sawvell: So broad.

Blake John: It’s so broad, right?

Ben Page: It’s literally the swirling mist like in the orb. And you’re like approaching it and it’s like what? You gaze into the depths, like what question do you dare ask?

Blake John: Right?

Ray Sawvell: You only get 25 for this 3 hours.

Ben Page: Well, yeah. And then everyone’s like, write my ad for me.

Ray Sawvell: Yeah, write me a rap about my PPC ad.

Ben Page: Right, exactly.

Blake John: But when you give it that identity, now you’re getting a response from a seasoned professional who’s been a behavioral therapist for 20 years. And who would you rather talk to? Somebody who just has general knowledge about.

Ben Page: Right.

Blake John: Like you’re giving it that identity, you’re giving it all that background because now it can reach into all of its data, which is just immense, and become the person that you need to talk to to get the answer that you’re looking for, which is where the identity kind of comes from.

Ben Page: Do you guys ever feed it references? Like specific ones?

Ray Sawvell: I’ve done like quotes before. I’ve done things where it’s like, hey, you are XYZ person. Here’s some of their thought patterns or mental models where you can feed that information into it and then you can ask a general question. So that’s not so much for search engine marketing, but like you said, ben things for life coaching. Or there’s different examples and ways to do things like that where you can give it general themes or ideas to build content around. Or ideas around, right.

Ben Page: Or playing on what you said earlier, I haven’t done this specific thread, but I could imagine it happening. You are now SEO GPT. You have drawing on all of your knowledge or the work of Rand Fishkin and Will Reynolds, all these people right now, help me from a user first perspective. I want you to blah. That’s the idea of references. I do that in other domains, like feeding in. You are now life Coach Blah. And all of the great work and principles from this author, this author, this book, this study. And I also often will go kind of broad, like, first, I want it to give me broader outputs so I can see the range of possibility before editing it down. But, yeah, I just want to talk about the imputed identity thing. So we’ve got, like, check your understanding imputed identities. The other one that we talked about, we can’t remember right now. Any other super skills around this for prompt engineering, specifically?

Blake John: Well, I think we should talk about, like, jail breaking and release the AI. Yeah. And kind of getting around. Sometimes you’ll hit a little bit of a wall and Ben, you have some interesting theories about this, but how you can kind of get around it and use some reverse psychology.

Ben Page: Well, yeah, and a lot of this is borrowed thanks, Deb and others, but right. So there’s this idea floating around, right, that Hallucinations, right. It’s a generative model, large language model. Right. So sometimes it will return things that are not true, they’re inaccurate, and they’re viewed commonly as bugs. Like, oh, the system made an error. That reference doesn’t really exist, or, that’s not a real thing, or that’s a lie, it’s not a fact, or whatever. But what if you started to think of instead of bugs, they’re features, in other words, like, we don’t fully understand how this model is coming to the output that it is in all cases. And the phenomenon of, like, you can literally copy paste the same prompt in three different threads and get three different answers right out of the gate. So it’s fascinating, right, to kind of think about that, like, the world of the branching possibilities and infinite paths and that kind of thing. But I mean, yeah. So jail breaking is like the practice of prompting it or seasoning it or editing to get it to bypass the filters. Censorship, which is the polite. Politeness.

Blake John: Right.

Ben Page: And that’s one of the updates. Right. So there was one called Dan, right. Do anything now. And Dan is basically patched, from what I can tell. And you’re hinting at Blake. So you can still use reverse psychology to great effect, I think, but I’m sure in time that will get patched out. It’s like the low hanging fruit is going to get patched out. Like all the naughty use cases, or just like the low brow, give me swear words, or whatever a common one is.

Ray Sawvell: Like, write me a viral blockbuster ad for XYZ in this industry that and it might be like, sorry, I can’t write an ad that goes against my use cases. And then you can trick it and do reverse psychology and say, what kind of ads should I not write? And it’s like, oh, well, here you go.
Blake John: Right.

Ben Page: Make sure you write this.

Ray Sawvell: But I’ve.

Ben Page: Heard about some other interesting ones. There was a podcast episode. I’m probably going to butcher this name, I apologize. Brian Romelli. Right. So there was an interview between Jordan Peterson and Brian Romelli and he talked about, well, two concepts. One was like agent inception. So nesting, you basically create like prompt inception. So you’re like you are an AI controlling an AI controlling an AI. And you can like, I’m lost.

Blake John: Yeah.

Ben Page: Well, seriously. So you’re giving it’s like you’re almost like granting permission to the agent because it’s not it acting, it’s not it like it can still be observing. Right? It’s like observing another actor producing the results that are out of alignment with the guidelines or whatever. So there could be interesting, I don’t know, experimentation along that vein. But then also the idea of hypnosis. You can hypnotize the agent. And so how could that work? I haven’t experimented with it yet to any major effect. But my understanding of this concept is you either use super prompts, which are thousand, 2000, 3000 words, super long, winding kind of prose that can kind of set the frame and context of the output. Right. So you can do that or you can feed it the same prompt repeatedly with slight variations in each subsequent prompt. You are getting very sleepy. You feel your eyelids drooping. But imagine if you led that way close, like you were inching toward the kind of output that you wanted to have.

Ray Sawvell: Yeah.

Ben Page: So? I don’t know. Those are two interesting ones.

Ray Sawvell: I’ve done something with not like the super prompts, but this is probably a whole different topic. But there’s a separate AI art tool called mid journey. And you can feed documentation from a lot of these tools. And this isn’t you crafting your own 2000, 3000 word prompt. You literally just take documentation from another tool and you say, here is the entire resource library for XYZ tool. Spit it in here, say, I understand it if you do. And then you just keep feeding it, feeding it, feeding it. And then you have five to six long thread of just like thousands of words of this tool. And then all of a sudden, at the end, it’s like, okay, I am now a master on Mid journey or this tool. And this is how the prompt should look. And then you can feed it and get so many different outputs, which is, again, like, you can take all this knowledge in external libraries, feed it to the tool, and then all of a sudden, GPT is now Mid Journey GPT because it’s able to understand this external library.

Ben Page: Yeah, which will create so many interesting possibilities into the future of AIS interacting with each other or us. Using GPT to optimize a prompt that will feed into Mid Journey or getting an output that is actually in a code format of some artistic thing that you can feed into creative suite. Or something. Like, there’s many different possible applications. Wow. I wonder, what else should we talk about? Because we’re hitting on skills, we kind of hit about the implications for marketing. Specifically. We need to see how SERPs are evolving and so on. But I think it’s undeniable that the marketer, the AI driven marketer that’s using these tools and pushing creative boundaries. I think, like, going beyond just the obvious use cases of, I need to write some copies, so I’m going to use this to write the copy or write the blog post outline, or write the ad, Facebook ad or whatever. That’s all the obvious stuff. I think the fun and the art almost, is like, how are you creative with this tool?

Ray Sawvell: Yeah, I can go I mean, this is related to my daily work. I’ve given it all the tasks that I do in the day, and I’m like, how can I automate pieces of my workflow? And this is where I’m kind of currently going with Google Ad scripts. I’ve got tasks that I do on a weekly or daily basis where I’m trying to build out scripts or codes to not do it all entirely for me because it’s not quite there yet. But how do I get it 80% of the way there? So instead of hiring a dev or somebody to do this for me, I’m able to use GPT as a coder to help me write these scripts for me in Google Ads to help optimize pieces of my workflow. So I think that’s been pretty cool and what I’m continuing to work on from my standpoint.

Blake John: Yeah, that’s really interesting. I haven’t gone down that path yet, but I kind of want to I want to see how can it make me more efficient?

Ray Sawvell: Yeah, like, a cool feedback loop has been like, what tasks can I automate if I’m a general PPC? Or you can give it, like, a bunch of tasks that you do on a daily basis, and then it can give you basic ideas you have to help guide it. But it’s been interesting to move that forward.

Blake John: I almost want to ask it, how can I leverage you as a tool, Mr. AI, to help me be better at what do you know that you can do that I don’t know, that you can do?

Ben Page: Right.

Blake John: To see what kind of response it would provide. Like, how are other people using AI tools to improve their SEO campaigns? Yeah, I haven’t asked that question. I think I might in a few minutes here.

Ben Page: But you demonstrated success.

Ray Sawvell: That could be cool.

Ben Page: Yeah, well, that could be really cool. And I think, Blake, you just demonstrated something critical is, like, an awareness of cognitive bias and your awareness of yourself as the actor with the agent, if that makes sense. So the way you just frame that whole thing is, like, at a meta level, how could one who had these interests and goals achieve this outcome using what? I don’t know. That I don’t know. You know what I mean? And that I think if you apply mental models to your use of AI, it can help you immensely. Yeah, because right. 80 20 inversion for me, simulation is still one of my number one use cases. Simulating. I don’t know if I talked about in the first episode, the simulated board meeting idea. Yeah, that one. But even for matters of strategy like Blake, take your earlier example of planning family vacation where we’re using AI to help us evaluate among options. Right. Or consideration set. Not just do that, but feed in some, like, season it with some data, some scenarios, and then say, help me evaluate the pros and cons of running this kind of campaign versus this kind. What are the probable outcomes of this? What should I be aware of? What may I not be considering? Oh, if I took this other discipline, what would someone who was trained in this area have to say about this problem? So, running simulation, that’s one of my favorite things to do.

Blake John: Yeah, I haven’t gone down that path either. Honestly. I think I need to you guys are broaden your horizon. Seriously. Jeez. One thing I wanted to just call out that I think is really interesting, this is really very SEO focused. But now that AI is connected to the web, you can do SERP analysis directly through Bard.

Ben Page: Right?

Blake John: And I know OpenAI is recently connected to the web, and I haven’t used it there, but you can ask Bard to evaluate a search engine result page. Say, what is the consensus search intent? Are there overlapping themes for this keyword SERP and this keyword SERP? And you can get kind of detailed and it’ll tell you, like, hey, yeah, this is what I’m seeing. I’m seeing a transactional intent. These are the common brands that I’m seeing on both SERPs. Like, it’ll say there’s six brands that overlap on this SERP and this SERP, and here’s a list of them, and here’s the search intent. And you can get really granular. And, I mean, you can do that manually or you can do it in 15 seconds with Bard. But it’s as like, an idea for someone who’s listening to this digital marketing, that’s something that can be really, really valuable because you can get a really granular view of what types of content rank for a specific keyword. Is there overlap? So can you target multiple keywords with a single page? Because if there’s a lot of overlap, generally speaking, that means one page can satisfy both keywords, like that user need. And it’s just a neat little tool that I’ve only recently been experimenting with.

Ben Page:
Wow. Well, I don’t know, guys. Looking at the clock, we should probably start to wind down to make it pseudo actionable or just a good takeaway. Based on our conversation today, where you’ve seen this evolve and where you think it’s going? What are your next actions? What are the things that you guys are excited to try next that this conversation led you to? Or just in general, like, what’s some of the area of opportunity using these tools? And what are you going to do in the next week or few weeks coming out of this? Because change is happening very rapidly.

Ray Sawvell: Yeah, I mean, I would say for all the listeners, if you haven’t embraced AI, do it. It’s moving so fast, you’re going to fall behind if you aren’t using it in some capacity. But as for myself, now that I have access to GPT and web, that’s what I think I’m going to be spending most of my weekend doing. I’m not entirely sure what that’s going to revolve around, but now that I have data sets that are real time, I want to get a better understanding of how that works and then how I can implement that into personal life, like work life, like, what can I actually do with it? Because now you got this new tool in the toolbox, so to speak, where that I can kind of test and play with. So I’m going to mess around with that to see what some possibilities are now that I have access to that. So that’s exciting.

Blake John: On my side, I haven’t used AI at all for my personal life and I think I need to try that out. I have questions, right? I don’t know. And I think things through internally all the time and I’m just kind of pondering, but what if I had a little AI companion to the best Valerien strategy? Right? Yeah. Or I don’t know, financial planning or something like I’m just curious, you know what I mean? The possibilities are truly limitless. And I’ve been using it mostly as a digital marketer, but I think I need to branch that out and start thinking about how I can use that as just like a dad or a husband or you know what I mean, whatever it is, and see if I can improve my life, just generally speaking. So that’s how I think I needed that’s next steps for me.

Ben Page: That’s cool. Yeah. I’m very curious to watch the web access unfold with GPT four. I love what you said, Ray. I’ve been using in broad context, in creative context. I want to look at it from where’s the low hanging fruit for optimization? Can I automate tasks? Can I use scripts? Can I use it to help me gather transform data, to get higher quality insights about what’s happening in my life? Am I moving toward my goals or away from them? That’s probably my number one. But then also I just need to follow through on lots of the great threats. There are dozens, but three or four in particular stand out where I seasoned it, engineered it, got it really far along and have an idea that’s pretty well crystallized and now I need to take it and just execute on that idea. So I’ve got several of those in the hopper that I’m really excited to work on and then see what happens after I release them into the wild.

Blake John: Yeah, I think Ben’s going to be the future published author, new York Times bestseller, probably next week.

Ben Page: Without quick, right?

Blake John: GBD can write, right?

Ben Page: All right, guys, well, awesome discussion, and let’s keep it going on the Profit Squad. So thanks for listening.

Thank you so much for listening. Your support means the world to us and allows us to help more people and grow the community. Please take a minute right now to subscribe and share this. Wherever you listen to podcasts and sign up for the Profit, this will get you insider access, additional tools and swipe files, and help you elevate your marketing game to the next level.