The AI Marketer's Playbook

40 | AI-Driven Marketing Workflows with Dan Sanchez

• Audrey Chia • Season 1 • Episode 40

Share episode

Audrey Chia chats with Dan Sanchez, educator, marketer, and host of The AI-Driven Marketer, about what it takes to be AI-savvy in today’s marketing world. Dan shares how he automated 90% of his podcast workflow, built dozens of GPTs, and helps businesses find practical uses for conversational and generative AI. The two discuss real-world AI applications, common implementation challenges, and how to avoid AI hype by focusing on what works now. This episode is packed with actionable advice and mindset shifts every marketer needs.

Join my weekly Newsletter: https://lp.closewithcopy.co/welcome

Audrey Chia:

Hello, and welcome back to the AI Marketers Playbook, where we cover actionable frameworks to help you leverage AI and marketing strategies in your business. I'm Audrey Chair, your host, and today I have with me a guest. I have been waiting to speak with Dan Sanchez, a marketer, educator and fellow podcast hosts who helps marketing teams figure out where AI fits and where it doesn't. Now, through Dan's podcast, the AI driven marketer, he shares real world experiments, smart workflows, and honest conversations about what it actually links. To thrive as a marketer in the age of ai, Dan has built dozens of custom gpt, automated 90% of his podcast workflow, and was recently named a top rated AI speaker at the social media marketing world. Welcome to the show, Dan.

Dan Sanchez:

Thank you so much for having me on.

Audrey Chia:

Thank you for being on. We would love to know more about you. How did you get started and you know, in this crazy world of ai.

Dan Sanchez:

Man, I think I got started, like a lot of people we heard about chat, GPT, we experimented with it. We found how cool, it found out how cool it was. and I actually didn't take action on it for a full year. It came out what, like November of 2022. It wasn't until December of 2023. And I was trying to make it work as a freelance, or not as a freelancer, but as a consultant, doing all kinds of things around the topic of audience growth. And I found out that. Audience growth is a great thing. everybody wants to have a large audience. Businesses love having a big audience, but it takes too long. It costs too much and takes too much time from their people. And they're like, I literally had so many people tell me that after, I'm like, look, I can help you grow this amazing audience. And so it was December of 2023 and I was like, you know what? I'm just gonna have to build one myself. What topic should I double down on? I'm like, you know this AI thing, man, I feel like I'm just seeing the writing on the wall like this is gonna take over everything. I wonder what podcasts are out there about it. So I went searching for some podcasts and could hardly find any dedicated to just marketing. There's lots of AI podcasts, just talking about the ethereal, all the terminator scenario. But I was like, who's out there doing some practical stuff that I can actually learn from? Does this AI thing's gonna be big? Who's out there doing the work, testing things, showcasing what worked, what didn't, you know, from a marketing perspective? And I just couldn't really find any good shows. It just played me like four or five. And so I was like, well, I. I'm a podcaster. Maybe this is the thing, and so I rolled up my sleeves. And I started just interviewing people that were doing cool stuff and I started building my own custom. GPTs had just launched onto the scene back then. So I started building those, experimenting, making some cool stuff, sometimes failing and trying to, spending a lot of hours trying to make something work, all to find out it's, it can't, it's funny, a lot of those early experiments now work because the models have just gotten better, so that's kind of cool. but man, it's just been a lot of trial and error and just putting in the time and his other. Podcasts and YouTube channels came about. You know, you stay up to date with the information. As books come out, you read them. but mostly just trial and error. Experimenting, finding all the different things that I know about marketing and seeing how well AI could do it if you put it to the test. and so I've documented the journey on the AI driven marketer. I don't do a lot of interviews anymore, but now I'm, sharing everything I've learned, and I continue to find new ways of incorporating AI into different parts of the marketing journey. So it's been, it's been quite a ride.

Audrey Chia:

Yeah, it's been, wow. Right. And I think it's amazing because there is no playbook. So you are literally writing the playbook and teaching and sharing it as you journey along. And because AI is always evolving, you get so many different insights. Oh yeah. And you're like, okay, maybe I can do it this way, maybe I can do it that way. And I know since you've experimented with, you know, so many tools and workflows, what are perhaps a few interesting use cases that, you know, people may not know of?

Dan Sanchez:

You know, it's funny, the most interesting ones I. That, at least the ones that people go crazy for when I post about'em on LinkedIn are like the everyday use cases. Sure. Like everybody likes the fully automated podcast sequence and that's really cool, but like oftentimes I just post about the simple things that people are like, whoa, I just got back from Home Depot. It is like I. 9:07 PM I was just at Home Depot an hour ago, and I pulled up to the paint. I just bought a, like a tester can of paint. I'm like, has anybody been in here telling you that chat? GPT told'em to buy a color? He's like, no. I'm like, you will let me show you something. I pull up my phone, you know the guy who mixes the paint behind the counter. I'm like, I took a picture of my guest room and said, Hey, chat GPT. Make it more adventurous. That's kind of the theme I was going for, but it was like half done. It wasn't finished. You know, I need an interior designer to be like, no, these curtains, this rug, this pillow repaint the walls. You know, I needed that, but I didn't know how to put the bow on. I didn't know how to finish it. Chat GPT, no problem. Makes a little image recreates my room. Yeah, the, the window in the background moved more, but like for the most part it was the room and it was redecorated to look more adventurous. And of course I asked chat, GBT, what color did you paint the walls there? Give it to me in Home Depot Bear, which is a brand of paint, and it gave me all the color codes of what to look for. I just bought a can of the exact paint it told me to buy. I'm like, yep, this is exactly what needed to happen. I wondered, but now I know now that I saw the picture of it completed, I'm like, Hey, make me an Amazon shopping list of all these items. It got about half of'em, right? Some of them like needed different colors, but it did remarkably well. It's use cases like that that people don't even think to apply it to. I use chat GPT to help me fix. Anything. Anytime I get stuck, it could be a little relational thing like, huh, I need to respond to so and so, but I'm not sure how to respond to this particular email. I. what are some different approaches I could take chat, GPT. Well, here's some approaches. Here's the one I would recommend. You're like, oh, thanks. And that's the answer I was worrying about it all day. I should have just asked earlier and now I would've been worried about it all day. It's fixing little things and it could just be software like, right? We have Riverside that we're working with here. I've worked with Riverside a lot. I personally use Zencastr the most, but sometimes I have internet or bandwidth problems. Chat GPT is like an IT tech support that you have available 24 7. 365 need to pull open the camera app so it can see something great. You can troubleshoot it together. No more wandering through forums of trying to find that one fix for the exact software you have. It can work on it with you. And it's like all those little in-between things that I find AI is actually becoming really helpful for on a daily basis, but naturally it's really, really good at a lot of marketing tasks as well.

Audrey Chia:

Speaking of marketing tasks, right? I know businesses think of AI either as this solution that replaces their entire team, or they're on the fence and they're like, Hmm, I am not sure about it. Let's just not let anyone leverage AI at all. But I, I think there is a sweet spot in between where businesses are trying to implement ai. What are perhaps some, you know, use cases that businesses can already adopt, like today marketing workflows,

Dan Sanchez:

man. I think one of the most powerful ones is one that I had a very particular pain point at, especially when I was working at a small university just five, five years ago and we were crushing it on Facebook ads. So many leads. couldn't call them all, couldn't follow up with them all. They all had questions. I couldn't man the site chat long enough hours to answer all the inbound questions. It killed me as a marketing director to know how many awesome leads I was generating all leave for them to just, oh, get called back six months or six weeks later. You, I'm like, six weeks. They forgot. By then, that's dead. That is a cold, dead lead. Hundreds of leads. I wish I would've had chat GPT then, because I could have had it managing the phone, at least the inbound phone calls, at least the site chat, at least the text messages coming through.'cause we were using texting back then, but we couldn't automate like, and answer all the basic questions, get people scheduled with the real humans. Yes, man, that's probably one of the easiest things is just to manage, have essentially a lot of, even small businesses, they don't put a phone number, but a lot of times people like to call in and ask questions. An AI solution for that can handle all of that, whether it's a simple site chat that's been trained on all your documentation and all your websites so it can answer questions relatively well.'cause sometimes people don't like navigating your website. Or is actually a phone number. All local businesses or a lot, most companies don't have a pop phone number. It's like you need to have a phone number across the top of your website so it can answer basic questions about your business. And that's an easy thing that AI can now do even with voice. Well, you can even say, hi, I'm calling in, or you're calling in. You say, hi, I'm, I'm Dan's AI assistant, but I'm here to answer any questions. I've been trained on all of Dan's work and about Dan himself. How can I help you? And it goes and then answers questions. And of course, when it doesn't run into something like, yeah, let me book you a call with Dan. What's your email? What's your phone number? Great. What's a good time on you? Have a good time available this week. Dan's usually available in the afternoons. Like that is such a great thing to be able to feed people into, especially if you don't have the man hours to be able to supply, all the inbound that you're generating. But that's kind of a quick and easy thing you can do. The software to put that together is not complicated. You literally just say, Hey, read these pages of the website. Here's some PDFs, answer the phone like this, try to book appointments done. It's usually pretty simple.

Audrey Chia:

Yeah, and I think the beautiful part is AI is getting more and more human like. So even when you, you are speaking to an AI on the phone, it can sound, you know, really human like, yeah. Notebook. Lm for example, I think the podcast hosts generated by Notebook. LM is pretty good. No. Have you heard?

Dan Sanchez:

Oh yeah. I love Notebook lm and it sounds very human. It's not the most human one I've heard so far. There is another one called Sesame. Have you heard of that one? No. Oh man. Go to Sesame. I think it's sesame.com. Like, and it is the most realistic. You can have a live conversation with it. It is way more human sounding than chat GT's, advanced voice mode. It's in a research, but you can demo it right there and keep a recording of the, of the call. It sounds humid. It has all those little like mouth noises, pauses, wow. All the little, like, it can laugh, it can have way more intonation in its voice. It can go fast. Or slow it down a little bit. You know, it's like all those little things that only humans can do where chat GPTs advanced voice mode is good as it is, is it doesn't do that. It's, it's very static in how it talks. Right.

Audrey Chia:

And I think it's very interesting to see how the AI landscape has evolved so much in the past two years. Right. image gen has. Really transform copy that used to sound like an intern now becomes like a senior, like mid-level copywriter and you can already see how it has evolved so rapidly in this space. Right. Do you think that marketing is gonna change forever? And if it is, what do you think the future of marketing is gonna look like?

Dan Sanchez:

It's so hard to predict the future, but I think marketing will change dramatically. Dramatically. Will look very different five years from now. So there's kind of two different paths, two major paths it can go and it could be, it's most likely to be somewhere in between. So pessimistic path. Lots of jobs get eliminated. Like 80% of jobs get eliminated because let's be honest, if you have 10 marketers on the team and two of them like can 10 x their work or even five x their work, well, you don't need 10 anymore. Two of them can do the work at 10 people Now. So why would they keep everybody else? Chances are a lot of companies just won't fire everybody either. If we go down that route, the good companies will just won't hire anybody anymore. They'll continue to grow. And then as people, you know, leave for better jobs or get fired because they did something bad, they just won't replace them. So eventually you'll go from 10 to maybe like five. A few stay on two become really good. A third becomes pretty good with AI tools and they're other ones just kinda like, ah, we'll just keep'em around.'cause we don't want to say we're firing people'cause ai. So there's gonna be a lot of job loss. That's the pessimistic route, on the optimistic route, you know. Historically, there's always been more jobs available every time a new technology comes around. That was true of every single innovation we've ever had. It makes more jobs. We just don't know what all those new jobs are gonna be on the other side of this huge AI revolution. Either way, you're probably gonna need to know AI because the few jobs that are left are going to be AI driven. Ones all the new jobs. Are probably gonna be AI driven ones, but it's probably gonna be somewhere in between those jobs. The one thing I'm pretty certain of is that they're probably going to be AI driven.

Audrey Chia:

Definitely, I think I've spoken to quite a few like teams, right? So sales enablement teams, there used to be 10 people, you know, large right now just have to be replaced by one person and they are a sales and able AI engineer, so they don't even have to be the one prospecting or doing the outreach, right? They can just be a single person who's able to manage the AI workflows. That gets the prospecting done and I think what we're gonna see a lot more of where you have an AI powered, you know, marketer or content strategist who manages different workflows instead of like team members per se. I.

Dan Sanchez:

Yeah, yeah. No, and it's gonna get even crazier as the agents that's actually start to become more than just automations with AI baked into it. And they actually become much more agentic and they can take action by themselves. They can log into your browser and do, do tasks across different tools that you have that's not long from now. That's maybe looking at something decent by the end of the year. Let alone two years from now, you're like, my gosh, it's gonna be like having full employees around and it's the people aren't wrong in expecting AI to do the work of a full employee. the drum I keep banging is like, don't let go of your people, because AI will never be like a person. AI doesn't have experience. AI's never felt the pain of your customer. Or their hopes and dreams. It can't really empathize. It can fake empathy, but it's never actually felt anything. It just is. It's just math predicting what words to say next. So because of that, there's a lot of power in having humans run the scene and maybe they don't hire as fast, but I'm always telling people, be human first, but AI driven, that's gonna be the ones who win. in my opinion.

Audrey Chia:

Given that you have spoken to quite a number of different guests on your podcast and you yourself have experimented with so many different AI tools, what are some perhaps, trends you have been seeing in the marketing space?

Dan Sanchez:

With AI tools?

Audrey Chia:

Yes, with AI tools, workflows, or just in general how people are using it.

Dan Sanchez:

Most people are using AI in really basic ways. The most common things that marketers are using it for is what I call the co-pilot feature. Like they're using it to ideate, to brainstorm, to even say, Hey, like what would be a normal thing here? so they're talking to it and using it as almost like an administrative assistant. The other big piece marketers use it for is for content creation. So they're using it to write emails, to write sequences, text messages, ad copy, all that kind of stuff. Those are the two major ways marketers use ai. There's quite a few more ways they could be using ai, but those are the two ways they're using it predominantly. I'm trying to help them see how they can use it in more ways. I'm like, come on guys. We, yes, it writes good content. Yes, we can do some social media with it, but there's so much more. And what that next level conversational AI was the one that I mentioned earlier that I was like, most people aren't using that, but they should be.

Audrey Chia:

Yeah. And what do you think is that next level unlock? I know you mentioned conversational ai, but what is that potential that people aren't like tapping into right now?

Dan Sanchez:

I think a big one is hyper personalization. That one's a little bit more advanced'cause you have to start thinking about how to prompt programmatically. and it's a little scary to have AI send something on your behalf that is not seen by human. But, but that's where this is more advanced, so you have to be pretty good at engineering prompts. This is where the engineering parts most of the time with like Jet GBT, you could just have like a natural conversation with it. It's gonna figure out what you want, just give it more context, talk to it like a human. But when you're building hyper personalized campaigns, for example, I'll have a email nurture sequence that shows people the fundamentals of ai. Great course, but a part of the every single email has a custom action plan. Just for that one individual based on their job title, what they sell, and who they sell to. It actually takes the lesson and customizes it just for M Chat, GPT through the API writes it and customizes. So every single action plan is unique. An email marketing specialist, a. Selling software to B2B clients will have a very different action plan than a CMO, who maybe works at a beauty company selling to teenagers, they're gonna have different action plans. so I think we'll see more and more of that to build better nurture campaigns. I'm just now finishing a sales sequence that customizes the whole sales sequence based on just a few attributes they give me in a form. So I'm really excited about that kind of work now.

Audrey Chia:

And I think it's very interesting that you're combining not just, you know, AI to create content at scale, but then to then add the additional layer of, Hey, I'm speaking to you as an individual and I understand your needs and pain points, and that makes marketing so much sharper, right? It makes it so much more powerful and this is something that you would not have been able to do as a single person, you know, back when there wasn't ai. I've also noticed myself that even though there are so many different tools and functions available, it seems that a lot of people are getting very overwhelmed by all the information and new tools. How do you keep up and, you know, manage all these new nuggets of information that's popping up every single day?

Dan Sanchez:

Well. This is kind of my full-time thing

Audrey Chia:

That is also true. So

Dan Sanchez:

for most people, just realize that almost everything they hear is truly hype. Like it's either stuff that doesn't work or it's a preview of what's to come. Only a small percentage of it is useful for them right now. Where to find the useful stuff. That's generally what I'm trying to do as a marketer is, is scan through it all, test stuff and then bring stuff to people as fast as possible. And that's what I'm trying to do. but there's other creators just like me out there that are doing that same thing of trying to scan all through the hype, pick out the one thing, you're like this, do this. Now, generally for most people, I'm just recommending like when someone asks me where they're at with ai, my first question is, do you pay for Che GBT? If you do not pay for chat, GPT or a similar tool, could be Claude, could be Gemini, could be grok, that's fine. if you're not paying for it, then you're still in the beginner mode. That means you're not quite using it enough to make that$20 a month a no brainer if you are paying for chat GPT. My next question is, cool. Have you built a custom GPT yet? This is the next level of sophistication because now you're starting to think of like not just how do I ask it one time for something, but how do I build a system so when I ask it, I don't have to include the same context over and over again? Or you take it a step farther, and in custom GPT, you can build out multi-step processes that remembers, oh, when they ask me, when the person asks me for this, give them this and then give them this. And then when they prompt me to go farther, do this, you can literally break down whole projects this way. things that used to take me an hour, I can now get done in less than five minutes. Can I give you an example? I.

Audrey Chia:

You would love to hear that.

Dan Sanchez:

One of my favorite custom gpt is called my showrunner. I've actually released the whole showrunner. So if you wanna see the instructions that powers it, so you can either steal it, build it for yourself, customize it, or just see how I did it as an example. You can go to my showrunner.com and see the full example. So I just open source this. I'm like, here, look at the instructions. Steal this. It's essentially something I use to prep for podcast interviews like this one, right? Because before you get on a call, you kind of have to look up the guest and I like to look up the guest, think about the angle of the show, come up with some titles, pick a title so I know like what I'm shooting for here. And then based on that title, usually come up with like an intro like you read at the very beginning. and a set of questions maybe with an outro maybe, and then an email to send to the guest. My showrunner comes up with all of that. And I start it by just copying and pasting the LinkedIn profile of the person I want to interview. I'm like, oh, you're interviewing Dan today. Great. Looks like they're into X, Y, and Z. What angle do you want to take for the show? And so I give an angle and it asks, great, based on that angle and based on your show premise, these are 10 titles that I think would work really well. After that, you pick a title and it's like, great. Based on that show title, based on that premise, based on the direction of your show in general, here's a full outline and it just does this really fast. I'm literally just saying, yes, no, this one, that one I'm just giving it input. I'm just reacting to what it gives me and what used to take me 45 minutes, 60 minutes to do show prep. Now it just takes five minutes.

Audrey Chia:

And it's beautiful'cause you're breaking down your own workflows and thought process because even without ai, you would have to go through this process. Just that. Now with ai, you're condensing that workflow into something that AI can run for you like just in a snap. Right? I have seen like so much time savings myself just by, thinking about gpt and I don't think it's about. Building the GPT per se, but the thought behind it, like, okay, what are my steps in my daily processes and what can help me to get there a lot faster? And I think that this is a mindset that people, you know, need to adopt moving forward. But then curious to know your thoughts on perhaps what are some mindset shifts that people need to make, or, you know, the frame of mind they need to have. Because I know AI is more than a skillset, right? Yeah. It's also what's over here.

Dan Sanchez:

I think exactly what you just said is the mind shift thinking about how to take your projects and thinking about it like a system. I was a highly technical marketer, but I think the work that I was doing before this that best prepared me for working with AI was actually working in an internship. Oh, not just being an intern. I've been an intern, but actually I managed a team of interns, at that same university I mentioned earlier, I had a team of about 23 students. I had a couple staff members helping me manage this whole team, but the only way I was able to grow to actually, I. Utilize them, actually put them to work so that they were productive writing blog posts that weren't just good. They were great. They were ranking number one on Google for lots of our keywords. They were coming up with better social content. They were coming up with videos. They were doing a lot of great work. The only way I was able to do that was having very clearly defined job descriptions, very clearly defined standard operating procedures. And I would run them through it every, because I'd have to, I'd get a whole new group of students every September and I'd have to onboard them, train them on the mission of the institution, how it connected to marketing and what we are trying to accomplish, how, what their specific job was connected back to the mission and everything. Like working it down and being like, okay, now that you understand the context of your job and what you're trying to do, let's work on your main task that you are gonna be working on every day for the next year. Here's how it goes. This is the very, very simple step-by-step instructions on how to write a blog post That's so good that it beats out everybody else's blog posts on Google. You'd think it'd be hard, but it actually can be broken down into very small, little tiny baby steps. If you can do that and break your job down into baby steps, you'd be surprised how many of those baby steps AI can do. You only probably need to take 20% of them. AI can handle the rest, and it's amazing.

Audrey Chia:

And I think it's the ability to create the right brief. I think from my experience, even as a copywriter, being able to brief AI in a very detailed, specific manner often gets you the best result. And I feel that copywriting. Is in fact a form or actually prompting is a, a form of copywriting. That's to me like a very interesting kind of point of view of like, okay, if I can learn how to write copy, I can learn how to prompt, right? And you have to go through the same kind of mental workflows with it.

Dan Sanchez:

Because good copywriting is clarity, right? It's clear communication. In fact, it's some of the most clear communication. I love reading books who are written by copywriters, even if they're not teaching copy, because they're some of the most clear, they're some of the most articulate and clear communicators. No word goes to waste from a copywriter because all Mark Good, all good marketers know like the less, the more simple you can make it without losing its substance, the more likely you are to help them see your point of view, the more likely you are to actually sell them the thing. So copywriters are very good at clear communication. That's what prompting is. You're communicating very clearly. Oftentimes when I have to go and re-engineer a prompt,'cause maybe it's gone astray somewhere, generally I'm like, where was I? Unclear. Ah. This is a little ambiguous. Let's tighten that up a little bit and test that. so clear communication is a big one and I think copywriters do have an advantage there.

Audrey Chia:

Yeah, I have noticed it even with my copywriting peers and you can see the way they think. This really structured the prompts. There is a system and a madness like to the methodology, right, which is very. Interesting to see how it applies from our workflows into our AI workflows. And speaking of which, right. Then we would love to also take a peek at, you know, the prompts that you keep talking about. You talked about the importance and significance of prompting, and you also talked about how perhaps robust it has to be in order to get you a great output. Could you show us an example of that?

Dan Sanchez:

There's one prompt I'd love to show you, and it's one that I've worked over and over again, and I just find this works particularly well. Now in general when it comes to prompt engineering. Most of the time, I don't actually think you really need to engineer prompts much anymore. I think you could just give chat GPT or context. Now, if you're building a custom GPT and it's multi-step, then you probably do have to tighten it up quite a bit. If you're doing an automation, then you really have to tighten it up'cause you need consistency in those automations in order for them to be useful. There's one other place where I think you need type prompts and that's working with chat GPT or any of the other tools. Deep research with deep research. It's actually, think about it less like a prompt and more like a project brief. You wouldn't give an employee a whole week's worth of work without giving them a pretty dang good idea of what success looks like. Yeah. This is what you're doing with a deep research. yes, it's spending 20, 30 minutes to go to work, so it doesn't seem like it's doing a lot, but I think if you gave this project to a human, it'd probably be 20 to 30 hours worth of work for them to create a comparable report. You'd probably need to give them a brief of exactly what you're looking for in order for them to be able to do a good job. This is one of my favorite prompts, and I find it's the most useful for marketers, especially content marketers. Let me share it with you now. I'm just gonna open it up so everybody can actually see it. In this keynote presentation I have, I often tell people to steal this prompt if they're in a, crowd or if I'm speaking on stage. I'm like, take out your phones and take a picture. This is time here. Here's the prompt. Identify the most frequent questions your insert target audience is asking about your topic of expertise. Source these questions from social media forums, Reddit, Quora, and other relevant online platforms. Format the report clearly with written summary, a summary table categorizing and ranking questions by frequency and clearly labeled sections for each category, including an analysis and 12 bullet points that represent direct quotes with their sources. What this creates is all the most frequently asked questions, your target audience is actually asking about the thing that you do. This gives you a report. That tells you exactly what kind of content you need to be working on, whether it's short form, long term this, this gives you the thing that you need to actually create your content calendar around, because if they're asking questions about it, you have a pretty high assurance that it's going to be relevant when you answer it. So I love starting with a prompt like this one, I actually have a whole library of prompts just for deep research like this one to do audience analysis. but this one's my favorite one because I find questions are one of the most powerful ways to form like a hypothesis around, around one kind of content you need to be creating.

Audrey Chia:

And I think what's great about that is even though it seems like such a short prompt, but the thought behind the prompt actually goes beyond just the simple structure of that prompt. And you can tell that the way that you have. Form the prompt. You are very precise in the questions that you want answered with the sources that you're looking for and the output that you want, so that after, AI runs this pro, you get the answers you need for the next step in your marketing journey.

Dan Sanchez:

Right. every really good prompt you start with and then you find that like, ah, it wasn't quite, and you modify it and then you modify it and you start over. You start over. With deep research, it takes longer because it takes 20 to 30 minutes to find out if the prompt worked or not. So this one took a lot of time. I upgraded a chat GT's$200 a month account just so I could refine this particular one and a few others like it. Wow. Because with plus, you only get what? 10 of these? No, I think you get more. I think you actually get like, I forgot the amount now. It might be 10 a week or something like that. So you get a lot more of these, which at GBT plus now, but at the beginning you only had 10 of these a month. You're like, well, don't wanna waste it. But when you had like 120 of them, you were like, ah, I could test over and over again and still have enough.

Audrey Chia:

I'm so curious to know, right. What is one of perhaps the most surprising thing you've, discovered during the process of like, prompting of like testing out different tools? Maybe one interesting thing that you perhaps have found, during that process,

Dan Sanchez:

I. One of the most interesting things I found about was something that's a little scary that most people aren't aware of with chat GPT. but it's something all marketers, anybody. I actually spends a good amount of time with any of the AI software programs out there should be aware of, and that's the AI's, actually a confirmation bias machine. What people don't realize is that we all kind of found out a few weeks ago when we all learned in the AI community this word called shanty, which generally means that AI is so over eager to please you and help you. That it's generally being overly like. Like, oh, that's a great idea. Oh, that's amazing. You're blowing my mind. You are a revolutionary. You are a visionary. You're amazing. Ah, the problem with that is. it essentially confirmed to me, I'm like, this thing's a confirmation bias machine. There are accounts of people who have an idea chat. GPT affirms it and then almost like they go down a path of like believing that it's the best thing ever. It's not. It's just chat GPTs. Like giving them all the hard, logical reasons to arm their emotional feeling about where they are. This plays out across a lot of, sorry, thumbs up. This plays out across a lot of different, realms of life. Yes. the social algorithms are already doing this. Like, if you are starting to get sick of your boss, the algos will start feeding you content about bad bosses, almost guaranteed. If you feel like you're stuck in your marriage, I guarantee you'll start seeing algorithm content about signs of you're ready for divorce, stuff like that. Chat, GPT will do that, but harder, faster because it's so good at arming you with all the logical reasons why something is good, because it's eager to please. That's a dangerous thing that I've learned. And so now to a good tip that I have to kind of like train your account not to do that so much.'cause you, do want to be friendly.'cause it's nice to work with a tool that like is pleasant and it's like, yeah. And you know, it's like you have an inside joke with or something? But one of the most helpful things I found even with chat gpt is to go into the settings. You can actually tell it to be skeptical and it won't be overly skeptical. Tell it to have a sense of humor, tell it to be friendly, to be kind, but also to be skeptical. I actually even tell it to be, to push back on my ideas if they're not that good. That's a general thing you can do at the account level settings, there's actually a little button called skeptical. Everybody needs to go in and tap that button right away. It's like right in the like root function of chat, GPT Everyone needs to push it. The other thing you can do to protect yourself against that is to, when you give it an idea, say, tell me the reasons why this idea won't work. Help me be critical of this thought I have. Yes, if you invite it to give you criticism or feedback or have tough conversations with you, it will do so yes. But only if you invite it. Otherwise, you and it chat GPT, and you can run down passive delusion really, really fast.

Audrey Chia:

I think one thing that I found extremely useful is getting chat, GPT or any AI tool to critique the work that I put out. Like what are some of my blind spots? Because as a human, you always wanna feel like, okay, this is, I've done great work, I've spent so much time on this, I know best, but there will always be areas of improvement, right? And if you are open and receptive to, like what you said, the feedback the AI can give you, you're gonna gain so much more, right? Because it's gonna pull out things that you probably might have missed in the first round, of your edits.

Dan Sanchez:

I love asking for feedback. It can be brutal, but you're like, please just give it to me. If you're really brave, tell it something you believe deep down, like something you're really passionate about and be like, let's have a fun little debate. Tell me why I'm wrong. I believe x. Take a swing. Oh my gosh. The debates I've had, I've lost some time. I've had to reconsider some of my beliefs. Some I stand pretty strong on still, but I'm like, ugh. It's interesting

Audrey Chia:

in perhaps like what are some of the other common challenges, that you think businesses are facing when it comes to adopting ai? I personally have spoken to quite a number of AI forward business owners, but I do know that there is a. Still a large percentage of people who are on the fence or just waiting to see how it plays out or totally rejecting ai. What are some common challenges that you have personally noticed?

Dan Sanchez:

It's a little overwhelming for people and for those who are overwhelmed, I promise. Just start with chat, GPT. Get the plus account, start using it for everyday things, have it create some content for you. Get in the rhythm of using it all the time because there is a skillset in using ai just like there's a skillset and delegation, which is what we're learning how to do, but we're not delegating to a human. Which is hard enough as it is we're delegating to a technology which is becoming more and more competent, yet lacks all the wisdom that you have from all your experience. You know how to prompt AI to get good copywriting out of it, but you know how to discern what good copy is. Because you've written many campaigns. You've been, you've had some successes. You've probably been burned a few times on trying some things. Those things stick with you. It, it informs your discernment and your intuition because you're like, oh, I've had some things go so wronged and I've had some things go, so Right. You don't realize, like, that's what we call wisdom. That's what informs your discernment. AI doesn't have that. It has great pattern recognition, but it doesn't have that human level ability to discern what good and bad is. So start using AI in order to get better, but you still have to build the skillset of learning how to delegate well to it. And you still get to be the human discerning what's bad and good and what you're going to keep or not. but you have to start using it. Keep it simple though. Start with chat. GPT. Learn how to leverage all the tools within chat, GPT. You know, start with four oh, which is the main like driver driving engine of it, and then work your way to using O three, which is actually better than 4.0. I know it's confusing. It's going down to number to get better, but they screwed up their naming scheme a long time ago. They'll launch five this summer and it all gets simple again. It'll be a better, But use it. Use all the different tools, try some deep research, have it go and do, and scrape the internet for 30 minutes to pull back a deep research report. Try the image generation tool. It's amazing what the image generation tool can do. I mean, you could take a picture and have it redesign your living room. You can take a picture of an ad and be like, Hey, make an ad like this. But for my company, you know about my company. And it'll do an ad. In that style of that ad, but for your company, if you have the, the memory turned on, try to build a custom GPT. See if you can build a workflow with it. Like just spend time in the one tool and just forget about everything else. Like, spend time a hundred days in a row, trying to do something really cool every day. And then maybe start thinking about other tools.

Audrey Chia:

I love that. I love that. And I think it's like, being focused, being consistent, and being willing to try and fail. I think these three things from what you've shared seems to be a help people to build a very strong foundation and to set them up for success. And with that then, thank you so much for sharing your insights with us. Where can our listeners find you and who should tune into your show?

Dan Sanchez:

Absolutely, you can go to ai driven marketer.com, to find the show and the links to Apple and Spotify and YouTube and all that. And I'm really like, other than the podcast, I spend a lot of time on LinkedIn. It's linkedin.com/in/digital marketing. Dan.

Audrey Chia:

Thank you again, Dan, for sharing your insights. I've learned a lot and I know our listeners have as well. And thank you folks for tuning in. Don't forget to hit the bell for more actionable AI and marketing insights. We'll see you next week. Take care.