How to Think Strategically About AI Tools

0:08Hello and welcome to the Duct Tape Marketing Podcast. This is your host, Sarah Naye. Today, I'm stepping in for John Jantz, and I have a guest on the show, Dan Sanchez. So Dan is an AI marketer, consultant, and creator with a passion of diving into the latest tech innovations. He specializes in developing cutting-edge marketing strategies that leverage AI to enhance customer engagement and drive business growth. So welcome to the show, Dan. Thanks for having me on, Sarah. Of course, I'm excited to talk to you today. We first connected on LinkedIn because you had been posting about AI and thinking strategically about AI and speaking to marketers directly, which all of that resonates with me.

0:48But when I reached out to you, I was commenting about one of your posts. And I'm just curious, do you remember what you said in response to my initial message to you? I don't. No, I put you on the spot. I'm like, I don't know. I comment. I mean, I'm dropping 200 comments a day or a week on LinkedIn. Oh, I'm sure. They all blur together sometimes. And I'm like, I don't know what I said. No worries. So I sent you a direct message and you talked about how duct tape marketing was one of the initial blogs that you were following back when RSS feeds were a thing. It was Coffee Blogger, Duct Tape Marketing, and Seth Godin were the three.

1:20I was transitioning from graphic design to being a marketer and a marketing kind of mentor to me. He's like, hey, back when RSS was a thing too. He's like, go to create a Google feed account and subscribe to these three blogs. You need to read them every day. And so I did for a very long time. I love it. And I just brought that up because I think it's very interesting. Like you've obviously been in the marketing space for a while now talking about RSS and original blogs. And now today, the focus of this conversation is going to be all about AI. And so it's just interesting to think about the evolution that we've had over the last several years and the pace of the evolution that we're going through right now with all things AI.

1:56Well, great. Then that's what I want to dive into deeper. I've noticed through your posts and the content that you're putting out, you're talking a lot about approaching AI strategically to avoid overwhelm. And at Duct Tape Marketing, we've been saying strategy before tactics for 20 years now. At this point, we're saying strategy before technology because you need to have a solid strategy. But I would love to open that up to you. How are you advising? Because it can be overwhelming with all the tools that are being developed and all the stuff that's being put out there on AI. So how do you advise people to approach the world of AI strategically?

2:28You know, there's a couple of different approaches, but it's funny because I just got a DM yesterday and it was like, hey man, heard you did a talk on the 25 new tools for AI in your session recently. What tools should I use? And I was like, I don't know. What problems are you facing? It's kind of like that whole strategy thing. It's kind of like, well, there's lots of tools. They can do lots of stuff. And there's some general purpose tools that can cover a lot of different things. But what's the core obstacle you're running into your business with right now? where's where's the you know the choke point for your systems yeah what's causing you pain on a daily or weekly basis because those are the things i want to look for first as a consultant and see how ai might be able to help that it's funny because a lot of times people actually don't need ai they just need clarity you know in a strategic focus set but i do find that ai is changing the game because it's allowing us not only to automate and do things faster or even better, but it's helping us think better and more strategically if you kind of know how to use it as a co-pilot.

3:33So that's the first thing I'm kind of trying to help people understand is like this thing becomes a very good strategy thinking partner. Even if you can kind of, you just kind of have to give it a start. It's not going to proactively come after you and be like, Hey, so what's your plan for this? Hey, what's your strategy? Hey, what were you thinking here? But if you proactively ask it for feedback or for considerations or ask questions that it can ask you and then give you feedback on. It's amazing how much more strategically you can think when you start using AI as a co-pilot. Yeah, absolutely.

4:05And that was one of my early aha moments with AI is at first I was just using it or thinking about it as like a content creation tool. I was thinking of something that like helps take stuff off my plate. But when I shifted to thinking of it as a thought partner and started using it in my strategic thinking and planning, Like that's where my view on what AI can do completely changed. And I know you have a story that you talk about one of your early on experiences with ChatGPT. I think you call it your Mediterranean ice cream moment. Do you mind explaining what that experience was for you? Yeah, it's when ChatGPT first came out.

4:36I'd been a huge skeptic of AI before ChatGPT came out. I'd seen some of the early pre-ChatGPT stuff like Jarvis, which was using its like 2.5 API ChatGPTs back then. It was like a copywriting tool. and I was like, okay, it's starting to get things. But when ChatGPT launched, it woke everybody up, including me. And I remember sitting there and being like, well, is this thing just really good at regurgitating? You know, is it like, because remember before we had like Drift, you know, an AI chat bot and we had Intercom and like they were all pretty bad. None of them were good. So I was like, you know, can this come up with original ideas?

5:12Most original ideas are usually a combining of two different things that don't normally come together. And humans do it all the time to come up with new ideas. so I was like well let's find something that doesn't exist on the internet and just ask for it to create something so I figured recipes would be hard because I'm like well that's a whole different dimension it's got to understand taste and recipe and how it might like how things come together in order to form new flavors that's pretty tough and then I went searching I'm like what's a recipe that doesn't exist so I just put picked out two random flavors I was like ice cream Mediterranean I went google searched it um could not find it and then search and said hey make me a recipe for Mediterranean ice cream and it punched out a bunch of ingredients I was like you know what this would actually work.

5:50And that's when I realized, I'm like, it has the ability to come up with new things if you're willing to guide it, direct it. And that changed everything for me because that was the missing piece. That's what to me made it artificial intelligence is it was able to actually think through and come up with a very kind of elaborate thing because making recipes is kind of hard if you're not pretty familiar with it. And that was a big unlock for me. Did you try the recipe? Did you actually taste the ice cream? No, no, no. I did not make the ice cream. But I remember looking at the ingredient list and thinking like, that was workable.

6:21I wonder what else this can do. And then I moved on and started knocking out other ones. But that was the first big one. I was like, ah, this isn't just delivering something back. This isn't just summarizing what it's found. You can mash up new things with this together. Yeah, that's great. Recently, we have these big bushes in our backyard and they've been bothering us for years, but not enough to actually do something about it. And we finally decided to rip them out. And before, like I would have had to like go to gardening stores and figure out what to plant and like talk to a number of people and spend all this time like designing.

6:53But instead, I took a picture and I put it into ChatGPT. And I asked like, you know, we're in Boise, Idaho, and this is full sun and all the things that I needed to know. And I ended up like designing this whole space of plants to put in that place. And while I was going through that experience, I had an aha moment of like my role prop like problem solving is completely changed like how I go about problem solving is different because now what I need to get really good at is prompting AI to help me solve problems and to push it for like further and to redirect things versus before you know I was going out and doing all that stuff manually and so that was an example of just like an aha moment of like how I solve problems is completely different than it used to be you you actually don't need to learn how to prompt ai as as much as you'd think anymore the ai models before you did because it was a little squirrely kind of like if you've gone if you've done ai video right now currently that is very squirrely right you try to prompt it and it's like it's all over the place and characters are disappearing and reappearing you're like my gosh i gotta really hone this thing and get what i want but it was like that in the beginning like it couldn't go that far without going off the rails in some way back in like 3.5 and early for chat gpt4 but nowadays it's gotten so good at anticipating what you want that yeah i i just talk to it like it's a person i'm like hey chatty buty interesting question for you my dishwasher is not working and i've already tried to troubleshoot it through some youtube videos but it's just not working but here's here's what i'm seeing and here's what's happening it's turning on but it's flooding with water but things aren't getting cleaned i don't hear it running and it'll just start asking you questions and you just have dialogue with it almost like it's an expert in your pocket you can call up anytime.

8:33And I was using the voice model as talking to it. But I find I'm doing it with like that all the time, whether I'm assessing my own strategic position in the market, whether I'm just asking to come up with a LinkedIn post. I'm just talking to it like it's an assistant that I just need to give it some enough context in order to carry out the task. Yeah, that's a great point. I've definitely seen it's improved drastically over the last year, I would say, in terms of like not having to engineer as much with the prompting. I'm curious, So we haven't shifted too much into the conversation of marketers.

9:01And so there's a lot of unknown in a lot of industries, but marketing is obviously being deeply impacted. And you had a great LinkedIn post that went out this week that I saw about, you know, AI tools are potentially going to replace humans in the future. And so I would just love to hear your take on to the marketers that are listening to this. What do they need to be thinking of moving forward in their roles as marketers? Is there an opportunity to evolve and shift? or what do you recommend for those that are feeling a little bit uncertain about the industry that we're in right now? There is a lot of uncertainty.

9:33And I try to think about the uncertainty in scenario planning methods where it's like, okay, let's say it is like, we're going to lose 90 % of marketing jobs. You're like, well, who are the 10 % that do have jobs? And what do I need to be in that 10 %? So I think about it like that. But I think about on the other side, like let's say this is going to be like every other technology revolution. Well, there's going to be a whole ton of new jobs that exist. What's going to be in those new jobs? Well, they're probably all going to be AI driven. So in either scenario, it's probably going to be who's me to become AI driven, right?

10:08And it's probably going to land somewhere in between. It's probably not going to be like this glorious thing. It's probably going to be good. There's going to be bad. There's going to be some loss on some side. I did recently post because a lot of people, there's been this, this trending topic on LinkedIn that I really had an epiphany. I'm like, you know, it's not going to be all that kind of like this idea that like human first is going to be the one that powers it. Like AI frees us up to do the more human things. And I'm like, that will, that will be true. There will certainly be a place where a lot of companies lean into being more human, more service oriented, and those will be great.

10:41And they'll win. There will be a whole nother set of businesses that win from just being more fully automated because somehow through AI, they create systems that deliver more value at just a much lower price. And you know what? People, a lot of people will do that. Like it used to be that you'd have a tax filer help you file your taxes and almost everybody's using TurboTax now, right? Unless you have a company in some kind of more complicated tax situation. You are hiring a CPA, but I'd still be even a little nervous to be a CPA right now. Unless you're like a really good one, you know? So I think a lot of businesses will be automated and there'll be people that go into the whole all human thing.

11:17And the cost difference between the two will probably be pretty dramatic, but there'll be a lot of ways to win. But I think what will help the most is trying to figure out what different paths will happen in the future and then finding like the common denominator around them. The common denominator I'm seeing is that AI skills are going to be a big piece of it. I heard someone talk recently about if you're a marketer, really anyone in a role is basically writing down everything that you're working on on a regular basis and then doing a bit of audit on that work saying, like, is this increasing in value because of AI?

11:51Is this decreasing in value because of AI? Or is this staying consistent moving forward? So have you thought about anything like that? Auditing your time and your skill sets to see what you should continue to leverage and grow on versus maybe start delegating the different tools and solutions? For me, it's probably a little harder because I'm an AI educator. So like, for my job, I literally get to waste some time experimenting and using these things so that I can report on whether it's actually helpful or not. I find the process of auditing on a regular basis to be pretty burdensome. I'm like, I wish I would just audit my days more.

12:29In fact, I've even thought about going to making a project in ChatGPT to be like, here's what I thought I would get done. Here's what I didn't get done. Here's some extra stuff and just dictate into it real quick, kind of keep like a daily journal and kind of a little bit of a coach. It's gotten way better at that recently. But it's, I don't know, I don't think I would, I would do that. I think on larger projects, I think it's really helpful, especially if you can bring some of that data back into AI, because it's learning now and can remember things across different chats now, and it will get better over time.

12:55I think that will become a strategic advantage. But yeah, I, for anything new in businesses, you do have to start small and kind of test your way there. Yeah, I will say it is probably like, there's enough effort and a momentum and society going towards AI, especially with businesses right now that I promise it's worth at least just going deep into chat GPT. It's the main one. And I heard somebody say even recently, like, oh, like I know chat GPT, but like I want to go beyond beginner. And I'm like, no, like, trust me, all the pros are using chat GPT to like, like, if anything, they're only spending more time in that tool because they're finding it more and increasingly valuable.

13:31Just don't waste time learning all the tools, like literally learn that one. And then if you have time and you have a need, start learning some of the other ones but time spent learning and how to leverage chat gpt specifically and if you like claude go with claude or gemini like pick one of the main ones and then just hone in that one craft in order to make the most of it yeah that's how my brain works with it all as well like i've gone all in on chat gpt and that's where i typically live every day but i know other people out there they're like i use this for this this for this and i'm like how do you have time for all of those things?

14:03Like I have to go deep in one to actually be able to use it to some of its potential versus, you know, going through all the different tools. So I think that's great advice. I'm curious. I'm part of a mastermind, an AI mastermind. And we were talking last week about how websites and marketing in general is going to have to shift because of the AI agents world. And we're right now we're designing websites for humans and ads for humans. And eventually, you know, it might be agents going to these different websites to make buying decisions for their people. And so have you thought or talked much about how marketing might shift in the next, I would say, six months to a year with the idea of agents becoming more of a thing or more of a focus?

14:44Yeah, I've thought about it a lot. I don't think it's going to change much in the next six months. AI agents, in my opinion, they're just not a thing right now, for the most part. What we do have, what we're calling, what most agents are, or what are labeled as agents aren't, they're not agents. There are, there's a few, there's a few exceptions and I'll talk about those in a second, but most agents are what I'm calling intelligent automation. They are just automated sequences like we had before with marketing automation, you know, like the little drag and drop builders. They're just that with one of the modules being chat GPT.

15:15That's it. Some of them are slightly more sophisticated because you're giving a little bit of autonomy to AI to choose between a few different tools. And maybe it's not injecting a prompt, it's actually got access to a database. Oh, that's starting to feel more agentic, but it's not like this fully autonomous thing that can go out shopping for us. It's just not. Now, there's some precursor tools out there that you're like, oh, that's definitely agent-ish, but they're not good yet. OpenAI has operator baked into chat GPT. You got to pay the$200 a month license to get access to it. It doesn't work well.

15:50Manus is the big one people are talking about from China, it also doesn't work well. It's just too many holes in the system. It maxes out too often because the server space isn't ready. The memory isn't ready. We have all the ingredients to make agents right now, but we're still... The cost of compute needs to come down a little bit. The context window needs to go up a little bit. We need to be able to give it more access to more things. There's a lot of talk right now about giving it access to... Google just launched its agent to agent framework so that it can interface, different tools can interface, agents can work with other agents from other tools.

16:25These kinds of standards and models have to be developed to create the infrastructure for it to happen. Right now, it's not happening. The one agent that I've seen that is actually good, it's agentic and it's worth, it's like one of the most underutilized AI features out there right now is deep research. Yes. It is going and doing a lot of work. And I love it. The more I use it, the more I fall in love with it. If you're a ChatGPT user now, you're paying for plus and you're not using all 10 of the instances of that you get every single month, you haven't figured it out yet. I promise. The best advice I give is like upgrade just for one month, upgrade to the$200 a month one.

17:01So you can get 120 instances of it and just throw everything you can at it, practice at it. You get 120 of them, like throw away things at it and just try it. It's different than using ChatGPT because it's going and doing like 20 to 40 hours worth of human work for you, which means you kind of, like I said, prompting wasn't good a minute ago, but for deep research, prompting actually is more strategic because it's less of a prompt and more of like a mini project charter, if you think about it. You kind of need to put some barriers on like where you want it to go, what you want it to do, what you want it to accomplish, where you want it to not go before you give it 40 hours of work.

17:39Even though it's doing it in 20 minutes, you got to remember these reports are so sophisticated. kid. You're like, that would have taken a human a long time. But that's the most agentic thing that I've seen out there. That's actually remarkably good right now. Yeah. I use deep research a lot for things like competitive research. If we're working with a client or if I'm present, like creating a new presentation and I want some data to like back it up, I'll have it create initial research to put together that. I'm curious, do you have any other examples of how people might start wrapping their head around using deep research?

18:09I got a few. There's one prompt that I fell in love with and it went like super viral on LinkedIn. It's like my most viral post to date was a deep research prompt. And it's really useful. So it is, I will give it to you just gripped out. I'm not going to read the whole thing because it's kind of long, but I'll give it to you. You can put it in the show notes. Okay. But it's essentially a prompt that goes and collects all the questions your audience is asking about your expertise. Okay. And it goes and searches Reddit, Quora, forums, and social to go and find them all and then organize them into categories and then rank stack them so that you can get at a glance, what are the most frequently asked questions your audience is asking about the thing that you do or the thing that you sell, whatever category that is.

18:56And that's just so helpful to see. And it actually not only rank stacks it, but actually gives you a header for each one and then puts bullet points of the exact how they're wording the questions with the link to the source so you can spot check it. It's so helpful because as a content marketer, a lot of things are still done by content, right? That's my planning path. I used to have to just have a lot of conversations on social or put out polls or just talk to a lot of customers. Now I can just extract it from the internet in 15 minutes and have a pretty dang good path of what I need to be talking about on social or on podcasts or blogs.

19:32That's amazing. That's a great example. We've, this isn't a deep research thing, but it's chat GPT thing. We've started recording a lot of our sales calls and that's just been really great content to put into GPT as well to analyze not only from a, what can our sales team be doing better, but also a marketing content perspective, because now we're capturing exactly, as you said, exactly how prospects phrase certain pain points and things they're struggling with. And then we're able to create marketing content that speaks directly to them moving forward. So I love that example. I got one more for deep research.

20:04Yeah, give me one more. Unless you want to wrap up. Give me one more. Because the deep research prompts are a little bit more sophisticated, something I've started doing is if I want to use a deep research prompt to dive deep, and maybe I'm thinking about launching a new product or I'm about to do something big, and I don't want to just do it willy-nilly. I want to have a substantial conversation with AI about it. I will start it off in 4.0, just talk about like, hey, this is what I want to accomplish. Help me build a prompt that would do really well and chat GPT's deep research. Ask me some questions.

20:33You know, this takes time. So like I'll tell it generally what I'm going after. It'll ask me some questions, get clarity. It'll craft the prompt. Then I'll switch it to the O1 Pro model within that same window or the O1 thinking model. And then on deep research, let it, I'm like, hey, that prompt above, go and do your thing. You know, it's like, because it's already crafted the prompt for me. Then you'll go do the deep research, come back with the refindings. I'll read it and switch it back to 4.0 or maybe even a different thinking model, depending on what you're going after, and then have a conversation about the research and pick it apart.

21:06But now it's got this like big research report in there that then you can have a conversation with AI to be like, okay, well, it looks like this. Like, what do you think? And then you can have a conversation and dialogue about the research, which is kind of a fun way to do it is chat GPT, deep research, and then going back to talking to chat GPT about it after the deep research report. Yeah, that's really interesting because you're using it in that sense in that example as a research assistant with the deep research. And then you're going into more of a thought partner co-pilot mode when you're going into conversation.

21:37Very cool. Well, thanks, Dan. Is there anything else that you want to share before we part ways today in terms of anything on the topic of AI overwhelm and strategic thinking? For anybody that's listening to this and thinking they're behind on the AI train, you're not behind. It's still very, very early. I promise. I've just got back from a conference just two weeks ago. People were asking all kinds of questions and I could tell just from the types of questions and their hunger they had that this is still extremely early. It is not too late. I know the hype has been crazy over the last two years, but as far as marketers actually using it in a meaningful way on a weekly or daily basis, very few.

22:15So it generally pays to be early on these trends, but don't be overwhelmed with trying everything. just taking some of the things we've talked about in this episode and practicing it and finding use cases that are meaningful for you. Again, look for those daily or weekly things you use all the time and start experimenting with AI and count it and write it off as like education time rather than, oh, I wasn't as productive as I was hoping it would be. Your first couple of swings at it are just going to take time. It took us all time to learn how to Google. It took us all time to learn how to actually write our first blog post.

22:44It'll take you time with AI, but it's early and putting in the reps now will pay dividends later. Yeah, that's great advice. I always like to think we're all learning together right now on this. We're all learning together. Well, where can people connect with you online, Dan? You can find my podcast wherever podcasts are at anywhere you search AI Driven Market on any podcast app. It's also on YouTube. It's a video podcast. And LinkedIn at linkedin.com slash IN slash digital marketing Dan is my most active social network. Awesome. Thank you so much for being here, Dan. And thank you all for listening to another episode of the Duct Tape Marketing Podcast.

23:17We will see you next time.