33 | AI Unleashed - Achieving Huge Results with a Small Team, Detailed Use Cases from Mike Kaput, AI Marketing Institute

0:00Hello and welcome to Leveraging AI. This is Isar Meytis your host and I have a great show for you today. My guest today, Mike Kaput, has been in the forefront of helping businesses integrate AI in generally and specifically in the marketing world. He is part of the founding team of the Marketing AI Institute, and he's been doing this for seven years, so way before ChatGPT ever started. And we are going to talk about how to dramatically amplify the results of yourself and your team using AI. We're going to talk about specific use cases in marketing as well as in general business. This episode is brought to you by the Ultimate AI Course.

0:41You can find that course on Multiply. That is spelled M-U-L-T-I-P-L-A-I. So like multiply, but instead of Y in the end, A-I in the end. Multiplyai.ai forward slash A-I dash course. There's going to be a link in the show notes. This course can take you from a complete beginner or wherever you are on your AI journey to understanding how to implement AI in a business context and how to build your business strategy around it. These courses are highly sought after and they fill out fast. So go and check it out. And if you do decide to take the course, you can use promo code leveraging AI, all lower case, leveraging AI to get$100 off the price of the course.

1:29At the end of this episode, I am going to share some news from this week, as every week in the AI world, there's some pretty big news happening. But now let's dive into how to significantly increase your output and the output of your business using AI. In the next few years, AI technology will change our world dramatically. Whether you are a business executive trying to catapult your business forward, or just somebody who refuses to be left behind and want to advance your career, this is the show for you. I'm your host, Isar Maitis, a serial entrepreneur and an AI enthusiast. You'll hear invaluable practical tips from innovative business leaders, AI practitioners, and some of the brightest AI minds in our world today on how you can leverage AI in ethical ways to advance your career and grow your business.

2:29Hello and welcome to Leveraging AI, the podcast that shares practical, ethical ways to leverage AI to improve efficiency, grow your business, and advance your career. I am really excited about today's show, and I'm excited about today's shows for several different reasons. The first reason is that my guest today, Mike Kaput, is a brilliant AI person. He has been writing, sharing content, and investigating how to use AI for marketing and in business in general for six or seven years. So he did not start recently when the big craze around ChatGPT came out, but he's been doing this for a very long time.

3:07He is in his profession, he's the chief content officer of the AI Marketing Institute, and he's the co-author of the book called Marketing artificial intelligence, which makes him an expert on how to use AI in businesses and especially in marketing. And that is going to be our topic today on how to use AI and AI tools in order to improve the efficiency and the results of various marketing aspects. And probably we'll dive into some general business aspects as well. In addition, I'm excited because Mike and I are both talking a lot of conferences about AI. And so our schedules weren't very easily set up.

3:47And we probably rescheduled this interview four or five times in the past three or four months, most likely because of me more than Mike, but we finally are here together. And I know we're going to have an awesome conversation and share a lot of value. Mike, welcome to Leveraging AI. Thanks for having me. I'm glad we were able to set this up. As I'm sure you can attest to, It's been a really banner year for AI speaking. We've always done a bit of speaking, but man, it has accelerated dramatically in the last 12 months. Yeah, absolutely. We were laughing about this. As soon as this interview is over, I'm actually driving to the airport to fly to talk in another conference.

4:24And I'm sure Mike has maybe a few days, but not much more than that for his next speaking gigs. But let's really dive in, Mike. Let's go and talk and define the setup, right? Like you're in a business, what are the things you're responsible for? And from there, let's start diving into how did you evaluate where does AI fits in? And then from there, they'll dive into how you're actually doing it. Absolutely. Yeah. So a few things that are important to note up front is first off, Marketing AI Institute is somewhat intentionally a very small lean company. We just brought on two more people. They actually start today, which is cool.

5:03Congratulations. The day of recording at least. But we're now up to only about seven people. So we really do quite a bit for a really small team. On paper, I am the chief content officer, which means I'm in charge of growing our audience. But there's also a dozen, two dozen other hats I wear on any given day, depending how the business goes. So I will say for us, like given our size, this is a very iterative experimental process. We help larger organizations more formally define AI roadmaps, and we've got V1 of our own, but we're really moving very fast and doing a little kind of quick and dirty experimentation.

5:44So I don't know if there's some necessarily grandmaster AI five-year strategy here. It's more, let's actually start eating our own cooking. We advise people to start with pretty narrowly defined use cases that have high probabilities of success. That's not the only way to start applying AI to what you do, but it's a really good place to start. So in the day-to-day, outside of our bigger strategic conversations, that's a lot of what I am working on and others are working on. So really looking at what am I spending the most time on in a given day or week, and how can we be applying increasingly some of these really powerful AI tools to dramatically shorten the amount of time it takes or get a better result for the same amount of time.

6:28I love that. I'll say one thing that is critical, and I share that with my clients. We both do very similar things, right? We create content, we share it, and we help in consulting to businesses. And what you just said is very critical because the biggest trick in applying AI is finding things that has high value, but also low risk. Doing internal projects. If it's an internal project, it could completely fall on its face. And yes, it will give you some weekend work to come up with the outcome that you couldn't come up with your experiment, but it doesn't damage the business. It doesn't hinder any future strategy.

7:03It doesn't lose you clients. And it allows you to experiment more freely with different strategies and AI tools and so on, really reducing the risk. And if it works, now you can go and take it to your clients and apply it at a bigger scale and so on. So even from that concept, I think it's very important. So let's dive a layer deeper. You started saying you look for things that you spend a lot of time on and looking for ways to improve that with AI. So let's share some of those things and how you have been applying or things you're going to try in the near future. Yeah. So a big thing that we have applied AI to in various aspects is our podcast.

7:42And so I'm going to start with a little context into what the podcast used to be, because it's important for where we're able to get to actually start using AI to actually capture more efficiency. So there's a piece of this that's actually before you even get to the AI piece. That's a strategy piece, a human piece to this. Our podcast used to be something where the format was our CEO, Paul, would interview AI experts, marketing AI people. They were awesome interviews, really valuable content. But we pretty quickly realized they were pretty unsustainable because Paul wasn't just a podcaster. Paul is the CEO of our company doing all this other really important strategic and business development work.

8:27I'm somewhat aware of that problem. So we aspirationally, it was like, this format's great. We'd love to hear these conversations every week with Paul and these leaders. But realistically, we just couldn't be spending like 10 plus hours of Paul's time on, and this was a couple of years ago too. So we weren't capturing as many efficiencies with AI. But at the time, it was probably about 10 or so hours for him to say, here's the people I want. I have personal connections to them. So I'm going to get them on the show, schedule with them, et cetera, et cetera. So at the time, the podcast itself, we probably had put out about 20-ish episodes.

9:07And they were really infrequent. They were inconsistent. They were great. Barely built any type of audience. Now, so before we were even saying, how can AI solve our problem here? we were like, are we even thinking about this strategy here the right way? So first, before we even got to applying AI to the workflow, which I'll talk about in a second, we actually had to take a step back and reevaluate this strategy. And we say, okay, look, even if we get AI helping us do this really well, Paul is still going to be spending an outsized amount of time on this format. So just sucks, but here's where we're at.

9:45We can't do this format. So we switched to a format where me and Paul every week cover top AI news. Now, that still takes a lot of work. And that workflow, which I'll get into in a second, still requires a lot of manual steps. But just changing the strategy alone suddenly freed up Paul's time. And we said, okay, this might be a little more sustainable. And so we started doing the podcast. We also got very lucky that we changed the format right before, I think like a month or two before ChatGPT came out. So there's this huge explosion in AI news at the time that really benefited the format. So a huge amount of this was just luck in terms of the podcast actually getting really popular, I would say.

10:29But at the time, basically what we're doing is curating all the AI news for the week, determining which things are most important, building out our commentary on it, and me essentially conversing with slash interviewing Paul about what is really important to know here. How can you sort out the signal from the noise for us? Format worked great. Podcast downloads took off, but this workflow still took a ton of time. We are still doing all of this manually, and at least Paul wasn't doing it, but our team, our small team, including myself, were spending a large amount of time on the podcast. It was super important to do, but we were like, okay, now that we've got a strategy dialed in.

11:10I want to pause you just for one second, because I want to highlight two very important things on the subtext of what you said. One is that, again, we both do this for a living, but we are huge believers in a business is a business first. Any AI pixie dust you're going to throw on it is not going to make a major difference. It starts with strategy, and then you have to figure out the business processes that will support that strategy. And then you can try to find the AI tools to solve some of those steps within the new process that you've defined. So that's number one that is clearly important.

11:48The other thing that I want to say, for those of you who do not have a podcast and are thinking like, I don't want to listen to this anymore, This applies to anything that you're going to do in marketing and probably in business because the process that Mike is talking about now generalizing this is how do I collect a large amount of information that is relevant to my industry on a weekly basis and analyze it in a way that will allow me to have a business benefit. In Mike's case, is producing a podcast that drives leads and people to the conference and so on. But the same process can be applied to any business intelligence collection and analysis that you just cannot do today because it's too time consuming.

12:32Sorry for stopping you. No, that's so important. That's such a critical point to mention is, yeah, this is not about on the surface. It sounds like it's about how to do better podcasting. It's an analogy for, no, here's how you theoretically could go about and transform any business or marketing process that you might be struggling to do efficiently. So keep that in mind. Yeah, I'm sharing some of these steps because, yeah, the first step was, are we even doing the right thing? Is this even sustainable? A second step is, okay, now that we know we have some confidence we're doing the right thing, what does the process look like?

13:08What are all the steps that go into it? So this type of process mapping is so important. And honestly, it's a great exercise, even if you're like, we can't use AI for any of this. It's still really helpful to say, okay, there's some redundant steps in here. There's some things we could do better. So that was our kind of next step of saying, okay, now that we have this big manual process that we do to produce something that we know is working, what is every single step in that process? So literally a whole checklist of here's every piece of what happens to get the podcast out the door. And after, here's how the content is created based on that podcast.

13:40Or we produce several articles from each episode. Here's how the promotion is created. And then from there, it became iteratively applying artificial intelligence tools to certain parts of it. And there's plenty of parts of the process today that still are manual. But even by applying just a little bit of AI to some key areas, we have reduced the time it takes to put this podcast out by 75%. So that's huge for us. That's just a game changer. And honestly, it's still such early days, I would say, with it. As much as we've captured all these efficiencies, it's, wow, okay, how can we even do this better?

14:20Or how can we apply what we've learned here to other workflows? So as an example, one big thing that we're doing with AI is using AI to professionally edit like video and audio. So we used to have to hire a production company to do this because we don't have very strong video and audio editing skills in-house. Using for one tool in particular, Descript, which is a great podcast editing tool or audio editing tool. We have someone who's able to do that very quickly internally. And then we're able to actually take the transcript from that tool. And then we run it through Pick Your Poison, ChatGPT, Claude2, things like that.

14:58And we're actually able to turn that transcript very quickly into summaries that are then used in blog posts. We can have it actually break down takeaways for us just to make give a bunch of raw material to a human writer, often myself, and say, hey, can you actually layer in your unique expertise, creativity and insight? knowing what you know from, as you said, doing this for six or seven years at this point, what comes out of that is a blog post that is far better than anything I would write from scratch. It is very quick to write, and it has a unique point of view that we wouldn't get just by clicking a button saying, Chad GPT, write something interesting for me, which I wouldn't really counsel people to.

15:43So I want to jump on a few points that you said, because again, they're pure gold. The first one, or if you want the last one that you just said, is that the real trick is finding ways to have this integrated approach of human and AI. Let AI do the tedious, mundane, annoying part of the work and bring your point of view, your unique thinking, your background, your experience, or human connections, whatever it is, into it in order to make it really a great piece of content. And that applies to anything that you do with AI. And by the way, even just the AI stuff, you can train it over time to do more of the work for you.

16:27So in the beginning, you can use it just to write bullet points. Later on, you can train it how to write in your voice. And then maybe the bullet points have some more meat on them and they have, and you need to do less. You still want to add your point of view because that's what's going to make it unique. That's what I make it you versus, oh, anybody else can hit the button and write the same blog post. So what am I really bringing to the table? And if I want to go a step back to what you said, going back to how you've applied this is find the steps in the process where these tools shine the most.

17:05In the case that you mentioned, it's okay. We still do the podcast, meaning there's two people talking to one another, exploring an idea, there's still content being produced. But in the middle, there's this process of how do I take the content from the podcast, it was two people talking to one another, it's an audio recording and turn it into valuable key points that I can then expand on when I create content, whether it's long form or short form or whatever content you want to create. That is really tedious and annoying. It's literally listening to the whole thing and taking notes and writing them aside and making sure that they're cohesive and all of that, Claude2 or ChatGPT or Llama or whichever tool you use, I use Claude for that because for long form content and summarization, just amazing.

17:51It takes about five seconds. And if you are smart enough, and I'll add a small little tip for those of you who are not doing it. If you create a prompt library for your company and save the prompts that do these little steps, you don't have to reinvent the wheel. So I use a tool called Magical Chrome Extension, which is just a Chrome extension that you can save really short snippets and it will write down long paragraphs for you. And I use that in my organization across other people and teams sharing through folders, different prompts. So when I want to do a step in a process that requires two or three different prompts.

18:32There is, let's take in your case, summarize a podcast. I will write a shortcut called some pod, which is not a word. So I will never randomly type it. And it will then prompt the full three paragraphs that I want. And then I have some pod two after it has done the first step to get the second step and so on. So in, I don't know, three words that I write and a few seconds of the large language model doing its thing, I've got an outcome that I can use for whatever I need to use. And like I said, generalizing it, you can use this for any business process. Let's go back to the first step of what you said, because I think that has a lot of value to businesses in general.

19:15You said content collection. You're somehow screening. And those of you, again, shameless plug for my, what's the name of the podcast? The Marketing Artificial Intelligence Show. Okay. So the show is amazing. Those of you don't listen and you're listening to this, go sign up for that as well. It's another great show. And it's probably the best show out there to stay up to date with what's going on in the AI world while analyzing the news. So it's just a news stream, but Mike and Paul literally go in and say, okay, here's why this is important. Why should you care? And so on. But to do that, you have to know what happened each week and a lot happens each week in AI now.

19:58Going back to generalization, you can do the same thing for any industry on any topic. So what steps are you taking today and what AI tools are using in the process or not necessarily AI tools in order to collect and then summarize the data in a way that allows you to even define what goes into the show? Yeah, it's a great question. So first off, I'll say in terms of actually getting links to possibly interesting news, that is an aspirational use case for us today for AI, I would say, because that's largely manual still today. We're lucky in the sense it's not a huge burden because we're already reading and looking for these resources as part of our other work anyway.

20:40So that's not a huge lift for us to be doing. It's something that happens. It would be happening if we didn't have a podcast. So So really, you're starting with a ton of different, honestly, each week, it's probably 30, 40 different links, at least to different news items we've collected in a central kind of chat throughout the week. Then from there, AI can be really helpful in summarizing the articles, in unpacking for me with some of the longer ones, like what are the takeaways here? I will still today, as of right now, read every article, no matter how long it takes. Again, that helps me do my job better generally, even if we didn't have a podcast.

21:20There's certainly an argument to be made if you were doing this in another marketing process or industry. You could absolutely get away with not having to read every single thing. But basically, I'll have certain tools. Claude 2 is, like you mentioned, one of the ones I find increasingly a go-to to just quickly summarize for me news items. I think it's also really helpful in terms of describing or simplifying certain topics to me. Occasionally, there'll be like a paper that comes out or especially meta, I feel like is guilty of this. They'll release these really cool like multimodal models and I'm like reading it and I'm like, I think I need a PhD to understand all of this.

21:59So it actually really helped me say, okay, what are the points I can extract from my audience to really simplify this? Because you can lose people in jargon, in really technical stuff. So that's been really helpful is almost, hey, explain it to me like I'm five because A, I need that. B, that's honestly how the audience will get more value out of it given our audience. Not that they're all beginners, but they're often really hoping to get to the practical, actionable advice on this stuff. So summarization and explaining concepts are really important to how we curate. Now, there is a piece here that honestly, I'd certainly be open to having AI curate this for me.

22:39I would love to explore that. We haven't really done that yet. I think there's a secret sauce here today, at least, having myself and Paul, our CEO, curate this. because we do have a sense of, here's what I think will actually be important because of all the context and background we have built up this kind of hard one experience, which is like the one thing that AI probably won't be able to do in a year or two. So there is a really interesting mix here of humans working with machines to get to this essentially research brief that informs the entire episode. And from there, once we've got, okay.

23:15Again, I just want to pause you for a second because again, if I want to generalize it for people, but let's say you want to do research about stuff that's happening in your industry or clients or and so on, you have some kind of a centralized location. I don't know if it's a Slack channel or a Google Sheets or whatever you're using, but that everybody involved just drops links of interesting stuff that they're seeing. I assume that's step one. Step two, you're taking those links at the end of each week, dropping them into Claude 2 and getting a summarization. And as I said, in some cases, a simplification of what the stuff says.

23:50Now you have, I assume, a document of some sort with all the main things and all the main topics. Do you then yourselves prioritize them or you still let the AI tool now prioritize what that is more important, more in depth and so on? Today, it's 100 % us prioritizing. Again, I would be open to seeing what could be done. I'm not going to be skeptical that AI will be able to do this, but we have found it really useful to say, okay, we know this is going to be important upcoming. Or one example, we are ramping up here in the next few months to host a virtual AI for agency summit in early November.

24:29So anything that is really interesting from an agency perspective, for instance, we might actually elevate to a main topic in the podcast because it's a natural piece of content to start nurturing more of that audience. So it's 100 % manual today, but I wanted to maybe double click on something you said and yeah, zoom out and say, okay, forget the podcast. Like what I just described is literally to your point of process, you could use for a blog post, a long form report or asset, your newsletter, any type of piece of content you are creating comes from some other research generally, whether it's your own subject matter experts, external research, especially as things like Google docs, Google workspace, incorporate more AI features.

25:08I still use Claude too, but you might get away with in the near future, drop all this stuff into a Google doc and you can use AI right in there to start summarizing and simplifying. So yeah, just like a quicker, I call it internally and in my head, like a quicker time to knowledge. Like instead of me having to sit here and like sift through all of this stuff, I can very quickly get a sense of, oh, I just consumed a huge amount of information. And that skill, that's the transferable skills. You can consume and synthesize a vast amount of information very quickly by using these types of workflows.

25:41Yep. And the same thing goes, by the way, for those of you who don't know, the same thing goes for video. So there are plugins within ChatGPT as well as external tools, as well as plugins for YouTube on Chrome that will literally summarize what the video is about. So if it's a 40-minute video, you can read five bullet points with timestamps and decide whether you want to watch the full video, whether you want to jump to minute 13 and 53 seconds to watch this segment rather than watching 40 minutes of video. So regardless of what means of consumption you have for the content, there are ways today to really understand what the content is about, to decide for yourself, like Mike said, do you want to consume the deeper, broader content and go and read, watch, listen to the rest of it.

26:33So this is how you guys curate the content. You talked a little bit about the post-production using Descript, which is, I agree, an incredible tool. We'll put a link to that in the show notes for those of you who won't remember the name or know what to use. What are the things in the process? You talked initially about content distribution and some other stuff that you use AI for in your marketing process. Yeah. So right now, when it comes to distribution, it's incredibly helpful as well. And again, this is where strategy meets AI and together they're far more powerful than they are individually.