<img height="1" width="1" style="display:none" src="https://www.facebook.com/tr?id=2006193252832260&amp;ev=PageView&amp;noscript=1">

34 Min Read

[The Marketing AI Show Episode 29]: Microsoft May Invest $10B in OpenAI, AI Clones Voices, and AI Disrupts Marketing Agencies

Featured Image

Wondering how to get started with AI? Take our on-demand Piloting AI for Marketers Series.

Learn More

It’s been another exciting week in the world of artificial intelligence. What does it all mean for marketers? The guys break it down on The Marketing AI Show.

First up, Paul and Mike have hypothesized about this, and now the speculation has come true.

According to a scoop from The Information, Microsoft plans to incorporate OpenAI’s ChatGPT into its Bing search engine. Today, Google handles more than 80% of search traffic. In comparison, Bing processed about 9% of worldwide searches in 2022. Rumor has it Microsoft is making a major play into using ChatGPT-like AI answers in search. Not to mention, this could open the door to using ChatGPT in Word and other Office products.

Just days after the scoop, Semafor reported that Microsoft has been in talks to invest up to $10 billion in OpenAI. The funding would eventually give Microsoft a 49% stake in the company, valuing it at $29 billion total.

Next, Microsoft just announced a major milestone development in voice AI.

The company revealed an AI model called VALL-E that can synthesize anyone’s voice using just 3 seconds of audio. Microsoft shared some audio samples, and simply put, Mike Kaput said they are “pretty stunning.” Based on just a few seconds of a person saying a random sentence, VALL-E can create completely new speech from scratch in their exact voice. The implications of hyper-realistic voice AI technology are widespread and mind-blowing.

Finally, Paul’s recent LinkedIn post on AI disrupting marketing agencies got a lot of attention.

In it, Paul said, “Artificial intelligence is going to disrupt a lot of marketing agencies in 2023. Some will seize the opportunity and create enormous value for themselves and their clients, while many will be obsoleted. These laggards won’t all die off overnight, but their relevance and value will quickly fade, and their clients (and people) will move on.” Hear Paul’s motivation for writing the post and the conversation that ensues.

As advancements become more prevalent, conversations have become more important for marketers and business leaders. Tune in to see how you can act on these developments. Listen to the podcast below or in your favorite podcast player.

Timestamps

00:03:28        Microsoft potentially investing $10 billion in OpenAI

00:12:52        Meet VALL-E, which can clone your voice using just three seconds of audio

00:21:14        Why AI will disrupt agencies

Links referenced in the show

Watch the Video

Read the Interview Transcription

Disclaimer: This transcription was written by AI, thanks to Descript, and has not been edited for content.

[00:00:00] Paul Roetzer: You're usually not paid to create a bunch of outputs of blog posts. You're paid to make an impact on a business to create value for a company to drive their growth, to build their brand. What is the thing you are being paid to do? And then I would look very seriously at like, how do we do that moving forward?

[00:00:15] Paul Roetzer: Like what are the ways to enhance what we already do? And what is AI opening the possibility for us to get involved in that maybe we didn't even do before.

[00:00:24] Paul Roetzer: Welcome to the Marketing AI Show, the podcast that helps your business grow smarter by making artificial intelligence approachable and actionable. You'll hear from top authors, entrepreneurs, researchers, and executives as they share case studies, strategies, and technologies that have the power to transform your business and your career.

[00:00:44] Paul Roetzer: My name is Paul Roetzer. I'm the founder of Marketing AI Institute, and I'm your host.

[00:00:53] Paul Roetzer: welcome to episode 29 of the Marketing AI Show. I am your host, Paul Roetzer, along with my ho co-host as always, and fresh back from dodging the massive outage of the airlines, San Diego, Mike Kaput. Welcome back home, Mike, and congrats on. Avoiding a massive delay.

[00:01:13] Mike Kaput: Yeah, it's good to be back. It's good to not be stuck in . I mean, there are worse places to be stuck than San Diego, but I'd prefer not to be stuck

[00:01:21] Paul Roetzer: at all. I felt slightly bad because like this was a panel that I was originally thinking about doing and then I just couldn't make it. And you know, Mike stepped in and you know, went out to San Diego and I was just thinking.

[00:01:33] Paul Roetzer: Man, I'm going to feel really bad if Mike ends up stuck in San Diego for two days. Like I'm watching the f AA alerts that morning. So, yeah, if you, if you don't know what we're talking about, there was like a major outage this week for, it seemed like about 45 minutes or an hour, maybe like a ground stop of all flights in the us and Mike was heading in the airport around that time.

[00:01:51] Paul Roetzer: So anyway, onto other topics they need, maybe they need some artificial intelligence in that f FAA alert system or something was

[00:01:59] Mike Kaput: I'm shocked. I'd be, I'm too scared to even look at the details of it cuz that's probably like running on technology from Don't,

[00:02:06] Paul Roetzer: I didn't want. Don't even want to know. Okay.

[00:02:10] Paul Roetzer: So today's episode is brought to you by the new piloting AI for Marketers online course series@pilotingai.com. This is a 17 course series that Mike and I researched, planned and recorded. We, we released it in mid-December 2022, so it just came out after chat, G P T and a lot of the generative AI stuff.

[00:02:27] Paul Roetzer: So content's super fresh. 79 demand courses, dozens of use cases and texts, templates, quizzes. There's a final exam with professional certification at the end. I think we're approaching like a hundred people have already signed up for this in the soft rollout period. We just sort of put it out in December, right during the holidays and haven't made a big push on it yet.

[00:02:45] Paul Roetzer: I know a few people listen to this podcast have already registered, so, we'd love to have you take part in that. If you're looking for, again, that step-by-step learning journey to really figure out how to apply AI to your career and your company. Piloting AI for marketers is the way to go. It is piloting ai.com.

[00:03:01] Paul Roetzer: There's a promo code AI POD 50 that will get you $50 off the registration. With that, I'm going to turn it over to Mike, who has our three hot topics in AI for the. All right,

[00:03:14] Mike Kaput: so first up is a big one because we've hypothesized about this a bit. People have speculated about this, but it's finally seems to be coming true, which is according to a couple scoops that have come out.

[00:03:28] Mike Kaput: First, we found out that Microsoft does have firm plans to incorporate. Open AI's Chat, g p t, into its Bing search engine. Just for some context there, Google handles about 80% of the searches today, whereas Bing is only at about 9% as of 2022. But now it seems like Microsoft is making a pretty big play into doing chat, G P T, like AI powered answers from your search results.

[00:03:57] Mike Kaput: And not to mention, Microsoft has a ton of other products that chat G P T could. Make a huge impact in in Word and other office products. Now, we were originally going to just talk about that release, but actually a few days later, Sefo reported that Microsoft has been in talks to invest up to 10 billion in open ai, and that funding would give Microsoft eventually.

[00:04:22] Mike Kaput: The terms are a little complicated. A 49% . Bizarre. They're

[00:04:28] Paul Roetzer: really strange. Yeah. read the article. It's, it's very what? And again, they don't know for sure, but yeah, the terms are not

[00:04:33] Mike Kaput: weird. And so from my understanding of the reporting, the funding would give them eventually a 49% stake in OpenAI. And it values open AI at about 29 billion total.

[00:04:45] Mike Kaput: And this is just amidst tons of other rumors and reports that Open AI is raising. Tons of capital as it kind of tries to capitalize on the, you know, the, all the rage of chat, G p t being, highly buzzworthy and popular and trying to reach this path of profitability. I just, this feels like a huge announcement or a huge, development from Microsoft.

[00:05:09] Mike Kaput: So I wanted to kind of get your initial read on this. What do you, what do you make of Microsoft move?

[00:05:14] Paul Roetzer: Yeah, I mean I think it was obvious this was coming, not, not the financial terms and all that stuff, but we talked about, like you said on past episodes, that it was obvious they put a billion in for a reason.

[00:05:25] Paul Roetzer: They have, we assumed some leverage here that, you know, they don't want this technology getting into other players' hands, the competitors. So makes total sense. You make a much bigger. We're just guessing what that would be. I mean, even the sourced articles, I mean, I saw one that said like, someone actually read the terms and here was the deal.

[00:05:44] Paul Roetzer: But reality is nobody knows. They haven't. Microsoft hasn't said anything as of recording of this podcast On January 2nd at 2:00 PM Eastern, open AI hasn't said anything, so. We don't really know, but it does seem like it is full steam ahead on infusion of this technology into Microsoft products, Excel office outlook.

[00:06:03] Paul Roetzer: Like anywhere where language happens, you're probably going to see this technology we've talked about before. It'll probably trigger the need for Google and some other players to make moves maybe they weren't ready to make initially. You can't sit back like this is it, it's happening like, so I. It's, I was saying earlier on something I was doing recording of something or other that I think the, the big story of 2023, yes, there's going to be thousands of generat AI tools built.

[00:06:31] Paul Roetzer: Yes, open AI is going to do insane stuff and runway and all these other companies are going to create stuff. What? Amazon, apple, Google Meta, Microsoft, what they choose to do this year with tech, they likely have been sitting on in some capacity. What they choose to do and what they, when and how they choose to bring it to market, I think is the story of 2023.

[00:06:53] Paul Roetzer: Like, I keep going back to like, what if Siri actually becomes useful? Like what if Apple is the dark course here and they've been building this stuff forever, but all we keep hearing from Apple is the AR glasses. And maybe that's like, it's a diversion. Like yeah, they're probably doing AR glasses, but maybe that's not actually the.

[00:07:11] Paul Roetzer: Maybe it's the infusion of language in, into their technology at a, a much more advanced level than what we're seeing. And maybe meta does spin out the AI or maybe they spin out the metaverse and and focus on ai. And so I just think that what those big companies do, they cannot sit back and they can't go earnings report to earnings report doing nothing.

[00:07:28] Paul Roetzer: They're going to get asked about this on the reports. I'll be fascinated to read into the earnings reports in the next few months of what these companies are saying and what they're being asked by the analysts. It's going to be in insane. Yeah, I mean it's, it seems like an obvious major play for Microsoft.

[00:07:44] Paul Roetzer: The financials. I think we're just going to have to wait and see what the end, what the terms end up really

[00:07:48] Mike Kaput: being. Right. Yeah, and I think, I don't have the exact numbers here, but you know, Office and teams alone are doing tens of billions of dollars as product lines. So, you know, it does seem like an enormous number in terms of valuation, but honestly, for how much money Microsoft is printing.

[00:08:06] Mike Kaput: I

[00:08:06] Paul Roetzer: don't know the average company and average person, it sounds like a lot, but yeah, we're talking about trillions of dollars in market value in the next 10 years. So it's, it's nothing.

[00:08:16] Mike Kaput: Yeah, and I think to your point about people having to like, it's here, people have to figure it out. That's going to be such a sea change for all the people who have been listening to us and maybe have been on the fence.

[00:08:27] Mike Kaput: I mean, if you use a Microsoft product, you're probably going to start seeing this functionality rollout in the next year or two. Suddenly you are very. Office technology is going to all be AI powered. You're going to have to understand how to get the most out of

[00:08:42] Paul Roetzer: this technology when your tech stack's going to get complicated.

[00:08:45] Paul Roetzer: Because even if they infuse this right away, it's not like if Microsoft Word all of a sudden has like G P T like capabilities within it. It's not going to necessarily obsolete the need to have a third party AI writing tool, because there's going to be a really good chance companies like Jasper and writer and hyper writing word Tune and go, Charlie and all these other players, they're going to build complimentary capabilities that aren't going to be living native within it.

[00:09:11] Paul Roetzer: And so it's just going to, it's going to get really confusing really fast. And I think the thing we always say is you have to take the steps to understand the technology. Because you're not going to be able to solve your tech stack, or do I need this tool plus this tool, or just this tool if you don't understand what's actually going on with the tech and how it works.

[00:09:28] Paul Roetzer: So I just, I think the faster that marketers and, and business leaders move to develop comprehension of this stuff mm-hmm. , the easier your life is going to be. Because it's going to move really fast.

[00:09:41] Mike Kaput: Yeah, that's a great point. And kind of to that point, when I was reading through this news and thinking about the implications of it, I just kept coming back to that hypothesis you made in a blog post, a few months ago, that there are two types of businesses moving forward, basically, ones that.

[00:10:00] Mike Kaput: Use AI and ones that become obsolete and within the ones that use ai. There's a couple different categories and you know, I know we have referenced it before, but I wanted to maybe see if you could connect a couple dots for us and talk through your hypothesis. With that, that was in that post and Yeah, because I think you really see it starting to play out.

[00:10:18] Paul Roetzer: Yeah. So what he is referring to is, there's a blog post, we'll put in the show notes, but the Future of Businesses AI or Obsolete, was the post I wrote it, I think it was like May 16th, 2022. And it was basically just looking ahead and realizing how this was all going to kind of play out and that you had a choice.

[00:10:33] Paul Roetzer: You could either be AI native, You build a smarter company from the ground up that has AI infused not only into the product, but eventually the, the marketing, sales, service operations, r and d, HR, finance, like you build a smarter company. So you just look at any industry, any business model and say, what's a smarter version look like?

[00:10:50] Paul Roetzer: And then you build it. The second is AI emergent, which is you are an existing company, which is probably most people listening to the podcast. You work at an existing company, maybe you own it, maybe you're a director, maybe you're just a practitioner. But you look at that company and say, oh my gosh, what is, what's a smarter version of our company look like?

[00:11:05] Paul Roetzer: And what if somebody else builds it before we do? And that, I mean, in some ways is, is crazy. It is an analogy. Would. Google to a degree, which is an AI native company, like they've been built since 99 as an AI company, and yet here comes along. Someone who, who released, I'm not saying, built a smarter.

[00:11:23] Paul Roetzer: Product, they Google probably has something like it. But they released it. They forced your hand like they did something to obsolete potentially, or at least disrupt something that's critical to your business. They didn't, either they didn't see it coming or they didn't move fast enough. And so if you're in insurance or legal or manufacturing or SaaS or any industry e.

[00:11:45] Paul Roetzer: you have to look at your business and say, how could someone disrupt us? How could they build a smarter version of what we have? And like we're building this consulting practice around this idea of working with middle market and enterprise companies to ask this exact question. Like, what does the smarter version of your business look like?

[00:12:02] Paul Roetzer: And how do we get there first? And then building these like AI roadmaps. You know, we're at three to five years. What does the smarter version of our company look like? So I think AI native, AI emergent is the only two type of companies that will matter. And I think, you know, marketing agencies we've talked about, and we'll talk more about them in a minute, is an example of like, you're not going to be obsolete overnight, like people are.

[00:12:23] Paul Roetzer: Your clients are just going to fire you in March cuz you're not doing AI services. But if by like end of this year, early next year you haven't moved and started building more efficient solutions and building more creative solutions. They look around and think, what do I need to pay you to be a writer for?

[00:12:38] Paul Roetzer: Like, I can use chat sheet PT to be my writer, or whatever the next version is. I just, I think that that, yeah, the thesis of AI native, or AI emergent or obsolete is, is starting to come to life very quickly. Yes,

[00:12:52] Mike Kaput: and I for our second topic today, there are two more developments that when I read about them, I just, my jaw dropped, and this happens every week now in in ai.

[00:13:02] Mike Kaput: But basically what happened is Microsoft also announced a major milestone development in. Voice AI and what they did is they rele revealed this AI model called Valley, v a l l dash E, clearly a play on Dolly that can synthesize anyone's voice using just a three second audio sample. So that includes preserving the person's voice style, their intonation and their emotion.

[00:13:32] Mike Kaput: And they actually on their site, and we've got the link in the show notes, where they publish the research and it's still just in the research phase, they shared some audio samples that are pretty stunning. I mean, based on just a few seconds, they have a person saying just a random sentence, and then Valley is creating.

[00:13:49] Mike Kaput: Completely new speech from scratch in their exact voice. And to me, I mean just that alone, the implications of having Hyperrealistic voice technology are just kind of widespread and pretty mind blowing, creating these like digital clones of someone's voice. And I thought another really interesting example that came out around the same time not using the Microsoft model is from Apple.

[00:14:12] Mike Kaput: And Apple is now rolling out AI for select audiobook narration. The way they described this is AI narrated audiobooks as quote, narrated by a digital voice based on a human narrator, and the company says it's starting to roll this out for fiction and romance books to start, and then they're moving on to non-fiction as well.

[00:14:36] Mike Kaput: And just these two things kind of hitting at the same time really made me kind of do a deep dive thinking about what happens when we can actually start using AI to realistically clone our voices. I mean, in marketing and business alone, I feel like there's hundreds of use cases for that technology, so I just wanted to get your take.

[00:14:55] Mike Kaput: What do you think of these developments? Do any use cases stand out here?

[00:15:00] Paul Roetzer: The first thing that came to mind for me was, I mean, how quick this happened because we looked at doing this for our book. Yes. So if anybody did, I mean, we did a, a podcast episode on the book earlier. I don't remember what episode it was, but, I think I told some of this story then.

[00:15:15] Paul Roetzer: But our book's like 60,000 Words, a marketing artificial intelligence book. It. I spent 18 hours in a studio recording that book. So I read the entire thing. I had two producers in my ear, so one onsite, one by a Zoom and I would read, we did I think they were five hour sittings. And then I did three hours of like, you know, finishing it up and stuff.

[00:15:40] Paul Roetzer: And we approached the publisher months before that and said, well, what if we got an AI to read this thing? What if we trained an AI on my voice? And we did. And so we started digging in like, is that even possible? So again, we're talking, this is like almost a year ago. It was like February, March of 22. We were looking into this and I knew that there was major advancements that have been made on this.

[00:16:03] Paul Roetzer: So I had cited a couple years earlier, someone from the Department of Defense that was talking about the Pentagon's capability to do this with under a minute of training. Hmm. So I knew that there was. Living within governments, that was far beyond what we were seeing as commercial users. But usually what happens is the government makes these advances and then they trickle into to business and get commercialized.

[00:16:27] Paul Roetzer: It's how sur came to be. So, We looked into it. And Google Wave. So Google has similar capabilities. Google Wave needed, I believe, as of early 2000, 22, 40 hours of training data Oh wow. Of, of my voice. So if we wanted to train my voice to synthesize it through Google Wave, it was going to need 40 hours of training data.

[00:16:50] Paul Roetzer: I believe that was best in class around that time. I think that might even been a technology that they talked about, Sundar talked about at a Google event early last year. So again, I knew that the, what was available through Google and I knew what the government was doing at least a couple years ago,

[00:17:07] Paul Roetzer: So I also knew, I guess, that just in general, this was a major area of focus. When I saw three seconds, my, my mind was like, I don't understand how it's possible. Like, I haven't had time to really think about this. But if you, if you contemplate the training data that would be needed to. That it's actually creating a model of someone's voice with three seconds mm-hmm.

[00:17:29] Paul Roetzer: because they're not going to train it on three seconds. They're going to, they're probably going to have to synthesize Yeah. What your voice sounds like in all these different scenarios through some training data source. So, I don't know. I haven't, I would love to read the research paper and find out how in the world did they actually pull this off.

[00:17:45] Paul Roetzer: Yeah, they do. Yeah. I mean, it's major implic. It's

[00:17:47] Mike Kaput: quite, it's a, I mean, it's still quite technical, but they do a good job of laying out the information in it. Okay. And providing examples. So anyone should certainly, even if you don't understand all of it, just take a look at what goes into this because, and a lot of it's predicated on.

[00:18:04] Mike Kaput: New, newer areas of AI that we've been seeing in a lot of different tools where it's like, like one shot learning or zero shot learning, where you don't near need nearly as much information or data to actually train the machine. And that's a really. Big development overall in AI as well. We're going to su soon be here in a, in a world where you don't need millions and millions of data points to train some of these models in certain area.

[00:18:29] Paul Roetzer: Yeah. I'll give you two quick thoughts. So one, a pro, like a really interesting area is accessibility of content online. Mm-hmm. . So if someone's coming to the site, they can't see the content and, and let's say I'm the author and we train a synthesized version of my voice. Every piece of content that we put.

[00:18:47] Paul Roetzer: We could, we could use that to, to read the content in theory. So you could click on the button and, and it could, it could just talk to you and explain it. So like it could improve accessibility in different ways. My other thought is like, well, could someone just take our podcast and train a version of my voice and just have me say whatever the hell they want?

[00:19:03] Paul Roetzer: Like, right. Is there any restrictions? Is it going to be like when I upload a list to my c r m? Yes. The list is legit. No, I didn't buy the list. It's like, am I going to, is someone going to upload a voice and be like, yeah, it's my voice. Or , I have permission to use this person's v I don't know. Like it's, and I'm not sure that that governance exists anywhere, like.

[00:19:23] Paul Roetzer: The tech is moving faster than the bylaws and the guidelines and like the rules for sure.

[00:19:29] Mike Kaput: You have to wonder too, with some of the regulation potentially coming down the pipeline, which we've talked about probably on the, in the European Union, you have

[00:19:38] Paul Roetzer: to believe, and China's got some, some advanced stuff.

[00:19:40] Paul Roetzer: Karen Howell, wall Street Journal is writing some cool stuff to speak about what's going on in China with them. You have to

[00:19:45] Mike Kaput: wonder if someone like the EU will. Look at the rapid pace of change and just bring the hammer down because what else are you going? They're, they're just not going to be an evolving set of regulations.

[00:19:57] Mike Kaput: They can't keep up

[00:19:59] Paul Roetzer: well, in the us I have to think it's going to be a state level thing, cuz Yeah. God knows if our, you know, government can get their act together and solve anything at this point. So I, yeah, I don't know. I mean it's, I was actually getting asked about that earlier today, about like, regulations in the US and what the rules are going to be, and I'm just, Yeah, I mean, I have to think states are going to move faster than federal at this point.

[00:20:20] Paul Roetzer: Yeah, well, but they better be having think things like, I don't, I don't know what's going on at the government level, but they better be doing more than the, the AI Act draft that they put out. I dunno who the committee is, but there better be one. Yeah, I mean,

[00:20:32] Mike Kaput: it will probably accelerate their timeline the moment we hit a really robust election season and we're hearing made up digital voice clones of candidates.

[00:20:42] Mike Kaput: I mean, and videos.

[00:20:43] Paul Roetzer: You've already seen you, stuff like that. Yeah. Yeah. You're going to have the deep fakes that are hard to distinguish. They're already out there. All right, so our third time Elon Musk, figure out how to get it off of Twitter. I'm sure. I'm sure they're working intently on keeping that stuff off on Twitter or,

[00:20:57] Mike Kaput: or just claim every time he says something.

[00:21:00] Mike Kaput: He didn't really think through that. It was a deep

[00:21:03] Paul Roetzer: fake. I love Twitter. I love Elon Musk, but. My confidence in Twitter becoming the, the, the gatekeeper of all of this is at, at, at a, a low, very low level at the moment. .

[00:21:13] Mike Kaput: All right. Now switching gears here to our third topic. This is something that we've referenced before on a couple episodes and.

[00:21:22] Mike Kaput: But I wanted to do a real deep dive on because you got a lot of attention from a recent post on LinkedIn and on the blog that you wrote, and it's about why AI will disrupt marketing agencies and really in it you said, quote, artificial intelligence is going to disrupt a lot of marketing agencies in 2023.

[00:21:41] Mike Kaput: Some will seize the opportunity and create enormous value for themselves and their clients. While many will be obsoleted, these laggards won't die off overnight, but their relevance and value will quickly fade and their clients and people will move on. Now, I saw tons and tons of comments that were, you know, discussing various aspects of the full post, which we'll get into.

[00:22:04] Mike Kaput: But I wanted to start off by asking you kind of what was your motivation for writing a post? That kind of detail. How AI is going to disrupt

[00:22:12] Paul Roetzer: the agency model? Yeah, I mean, I think anybody who knows my background knows I ran an agency, my first book was the Marketing Agency Blueprint. We were HubSpot's first partner.

[00:22:21] Paul Roetzer: I sold the agency in 2021, but I spent 16 years of my life. Building my own agency, and I spent five years before that working at an agency. It's all I knew. So I have 21 years in the agency world. So all of this comes from a place of love and of respect because many of my friends still run agencies.

[00:22:37] Paul Roetzer: Like I'm, I'm still involved to a degree in our agency, you know, as an advisor to the agency. I think with chat, G p t and, and just the advancements in generative AI and the move into the creative realm, I know that most agencies aren't ready, because we know them. And, and I think the, the horizon for their ability to figure this stuff out and shift their services, their business models has dramatically shortened in the last six weeks.

[00:23:11] Paul Roetzer: Hmm. So while they may have gotten away with not being AI emergent, to start 2020, I don't see that as an option anymore. I think AI emergent is an imperative. I, and, and I don't know that they're, they're taking it seriously enough. Like I don't, and so you and I have talked about this before, like I always, always, always with AI, have tried to avoid fear as a motivating factor to get anyone to do anything.

[00:23:38] Paul Roetzer: Even hand of free class, like read the book. I don't, whatever. We didn't want to use fear. What I came to believe was the urgency was critical. And there's people I care about that run companies. There's a lot of people I care about that work companies, that agencies, and I think they're going to be in a lot of trouble a lot faster than they believe, if they don't figure this stuff out very quickly.

[00:24:02] Paul Roetzer: And that's why in the post I just have like three things. Like, I mean, I could have made a list of 20 things, but I think I said creative billing and innovation maybe were the three key areas. Yes. So,

[00:24:11] Mike Kaput: yep. Yeah. Creativity, billing models and innovation are the, you said you would focus on these few key areas to get started understanding how this is going to change.

[00:24:21] Paul Roetzer: Mm-hmm. . Yeah, and I think it's, I, again, It's a logical place and I believe the, the, kind of the ex example I gave was like, if you're charging hourly rates, which a lot of these agencies still do, which blows my mind, but if you're charging hourly rates for creative work, How, like what is, what is the financial model that works now?

[00:24:41] Paul Roetzer: Like the, the, you just get it got obvious deleted in like a day, like Chet changed it. Mm-hmm. because now everyone was aware they could get content at a reasonable price. Even you not saying it writes it for you and you're done like human in the loops don't need to be there. Still need writers, editors need to fine tune whatever the machine outputs.

[00:24:56] Paul Roetzer: Like I'm not saying it's replaces writers. But you and I do this, like every podcast episode, we turn into three blog posts and how, how long? Okay, so walk people out through our workflow. Let's say we were paying an agency to do this. You and I get on a podcast, we talk for an hour. You spend an hour probably prepping for it.

[00:25:12] Paul Roetzer: So there's two hours combined our time, three hours in to record the podcast. Then they get to transcribed by the ai. Then, You take it, if I'm not mistaken, and we turn it into three blog posts. We split it up into three different posts, which we use the transcriptions for. And then I think if, I don't know what we talked about this, but you started actually using chat g p T to do summarization of the transcriptions to help with the creation of the blog posts.

[00:25:39] Paul Roetzer: So podcast three blog posts, which as an agency, we used to charge like 700 to a thousand dollars per blog post to create. How long do you think it takes us to, to create the, the podcast that gets published? The three blog posts, and I guess we could throw in social shit. We could throw in all the other things we use AI for, but like, let's just say those three, three components, three blog posts and a podcast.

[00:26:03] Paul Roetzer: I'd

[00:26:03] Mike Kaput: probably say overall, yeah,

[00:26:04] Paul Roetzer: 10 hours, 10 hours total now,

[00:26:06] Mike Kaput: but with recording production and then the writing of the posts

[00:26:09] Paul Roetzer: and things like that. Yeah, yeah. What would an agency charge us do you think? If we, if we had them same level of output. Honestly, descriptions three blog posts. My

[00:26:18] Mike Kaput: conservative thousands estimate would be, yeah, probably like two grand, 2,500 minimum all in

[00:26:27] Paul Roetzer: for, well, we would've charged if a client came to us and said, can you turn this into three blog posts?

[00:26:31] Paul Roetzer: And before we had these writing tools. So Mom, my question is like, if, if you were the agency that you were, we were, we were paying an agency $3,000, let's say, to turn each podcast we record into three pieces of content, get it on online, distribute through the channels, whatever. And we knew what's possible.

[00:26:47] Paul Roetzer: Now, would we still pay the agency $3,000? Hell no. Not, I'd be like, you could do this job in like an hour, whatever it is, like a thousand max. Like, right. So that's my, my debate right now is like, as soon as the market figures all this out, as soon as your clients like the directors, VPs, CMOs, like the people who hire you, Once they understand what AI is capable of doing, there is no way in hell they're paying your traditional rates to be inefficient.

[00:27:13] Paul Roetzer: To do this stuff with all humans. Not going to happen. And that's where I think like in design, in video, in language generation, content creation, all of it. So anything that creative touches, advertising, social media, we as business leaders, markets, we're going to know you can do it more efficiently than you were doing it before.

[00:27:33] Paul Roetzer: You cannot charge us what you were charging us. You can create three times the output at the same cost if you'd like. But I'm not paying you the same fees cuz I know the tools got smarter and they're, boy, they're far, you should have gotten more efficient. If you didn't get more efficient, that's your fault.

[00:27:46] Paul Roetzer: I'll go hire my own writer or find an agency that did. So that's where I realize like, oh, they're screwed, like it happened. So, Again, o over time and then all of a sudden, or whatever the saying is. Yeah, that's what happened to agencies. And so that's why I said like, you have to figure out your billing model fast, your financial model.

[00:28:03] Paul Roetzer: You have to figure out how to infuse this in your creativity, and then you have to get ahead of this, like, what's the innovative thing you can do? How can you use these tools in ways other agencies are thinking about and seize the opportunity to build massive value for yourself and your clients, rather than just being scared and thinking it's just another disruption.

[00:28:20] Paul Roetzer: What? Pick, pick whatever people use as an excuse for like, we've been through this before. No, you haven't, right? We've been through nothing like this before. Unless she ran an agency when the internet came around. You have, you've been through nothing like this before, so don't use that as an excuse that it's just another disruption and we deal with this all the time.

[00:28:36] Paul Roetzer: No, you don't. Yeah,

[00:28:38] Mike Kaput: it's, it really does seem like the fundamental economics of the business have now changed and that I, so I don't, you know, you mentioned one thing just here that I wanted to maybe dig in on. I don't want to put you on the spot, but you said, Hey, how do agencies get ahead of this? And that's where I keep coming back to that question of, If all this stuff we just mentioned is no longer worth paying, the rate you expect for, or even hiring an agency in the first place, what are some areas where agencies might start to think about creating new types of

[00:29:08] Paul Roetzer: value?

[00:29:10] Paul Roetzer: I mean, I think about what are you paid for? You're usually not paid to create a bunch of outputs of blog posts. You're, you're paid to make an impact on a business to create value for a company to drive their growth, to build their brand. Like what is, what is the thing you are being paid to do? And then I would look very seriously at like, how do we do that moving forward?

[00:29:28] Paul Roetzer: Like what are the ways to enhance what we already do? And what is AI opening the possibility for us to get involved in that maybe we didn't even do before. Where can we move into, can we, can we use our knowledge of language models and how they work in marketing to advise their sales team? Maybe you were never in on the sales side.

[00:29:46] Paul Roetzer: Maybe you can go in and teach HR some of this stuff. Like you might be able to expand if you think of yourself as a growth firm or like, you know, somebody's there to really build value for that company. I would just take a different look at what is it that we can do that can create value for them knowing the economy we're in and that many businesses are struggling and having to lay off workforces and knowledge work.

[00:30:07] Paul Roetzer: And there's lots of different variables right now to running an agency with, without even factoring in the AI side. But I would just look at that and say, how do we do this better? How do we create value in a very unique way? Maybe it's vertical specific, like maybe you're just really good at applying these models.

[00:30:24] Paul Roetzer: Like financial services or manufacturing, whatever it is you do is fine. Go find the smarter tools being built. Like that's one thing I'd say, like if you're a verticalized company, let's say we do manufacturing, I would go right now and find the AI tool is building, being built specifically for manufacturing companies.

[00:30:39] Paul Roetzer: Yep. Go. Like that's how we built PR 2020. We were HubSpot's first partner. We just saw the vision of HubSpot, believed in what Darsh and Brian were doing, and we said, well, let's build an inbound marketing agency. Let's build around HubSpot software and we can help drive growth for for clients. I would do that right now.

[00:30:54] Paul Roetzer: Like, who's the next HubSpot in your space? Like what is the AI native company being built that you could potentially become a value added reseller for that you could introduce to your clients and say, Hey, you didn't even have this ability before today. We now have this AI tool that does X, Y, and Z. We can do this for.

[00:31:12] Paul Roetzer: It's something maybe they haven't even thought

[00:31:14] Mike Kaput: about doing. . Yeah. And it also strikes me that there's probably, it's a less sexy opportunity, but every agency should be looking to use these tools to cut costs internally, for admin or thing, not get rid of people, but for all the stuff that's distracting you from figuring these problems out.

[00:31:34] Mike Kaput: Like there's probably very few, you know, admin aspects at a smaller agency at least that can't at least be augmented in part and dramatically sped. By these tools. I mean, we just even talked about the podcast. If you're doing that for your own marketing at your agency, there's ways to dramatically simplify and streamline the process with ai.

[00:31:54] Paul Roetzer: Yeah, and I mentioned earlier, like the consulting stuff we're doing and the whole premises to like build these AI roadmaps. Well, I'm actually working on building one for the institute. I'll be doing the same thing if I was running an agency and I, what I would be looking at is like, how do we make every component of the business smarter marketing, sales service.

[00:32:10] Paul Roetzer: I would look at my operations, I would look at my client's services. I would look, so if I was running an agency, I would, I would sit down and say, okay, what's the next three to five years look like? And how do I start project by project to infuse AI into this business? And I'm probably going to start in an agency with my services.

[00:32:24] Paul Roetzer: What are the core services we offer? How is AI affecting those today? and I would then work backwards from those services because that's, that's where the money's coming from. Simultaneously, you're looking at the billing model, you're looking at creativity, you're looking at talent. You're looking at up-skilling, re-skilling your team because of AI is going to be a key component of the services.

[00:32:41] Paul Roetzer: You have to think about professional development and hiring. There's a lot of moving pieces. I don't, I don't like, it's not going to be easy. I'm not saying like you just AI's going to fix all of this, but you have no other choice. Like you just have to build a smarter business.

[00:32:57] Mike Kaput: Yeah, I think the professional development piece stuck out to me.

[00:33:01] Mike Kaput: Obviously we have talked about how we sell our own course that is piloting AI for marketers. But really what strikes me from our journey over the last five, six years trying to figure this stuff out. I was really proud of us at the beginning. You know, on the agency side, having like regular book club and stuff for this top of these topics, that's all great and there's nothing wrong with that.

[00:33:21] Mike Kaput: But given how breathtakingly fast things are moving, you need to hit the gas. This probably isn't like a, Hey, we're all going to read an AI book every three months anymore to get to solving

[00:33:32] Paul Roetzer: this now. Yeah, and I think that's, you know, with the pilot AI series, like that's what we designed it as. You can do that entire class in eight hours.

[00:33:39] Paul Roetzer: Yeah. You can speed it up and do it in six. And if I was running the agency, I would, I would probably look at the next month and say, okay, as a team, everyone is going to go through and complete this series. They have to complete it by X date. And then we are going to have a half day all hands meeting, and we're going to go through the use case model together.

[00:33:59] Paul Roetzer: We're going to do the problem-based model together, and we are going to come out of January. With a preliminary idea of what we're doing in 2023 for this. Hmm. And I would, but I would involve everyone in that this is not a, the c e o of a company agency brand, whatever, just dictates to everyone, here's what you're going to do.

[00:34:17] Paul Roetzer: You have no idea what they do for their job. Like you don't know all the tactical things these people are doing every day to create the output. So the people who are actually doing the work need to be part of this process. They need to think about, well, I'm the one in charge of the podcast and I spend 20 hours a.

[00:34:31] Paul Roetzer: Doing these 12 things, like I think what they're saying with AI after taking these courses or whatever is that I can find some tools to cut this in half. Like that'd be huge. I should recommend that to my team. Mm-hmm. So that can surface in that like run an internal AI summit, like bring everybody together, get ideas from all across the table, say, okay, let's start here, here and here.

[00:34:48] Paul Roetzer: That's so we can save the most money for most time next month. So yeah, I think the thing we always. Hammer home to everyone is, you can do this pretty quickly. Like you don't need to go back to school to learn this. Yeah. You don't need to be a data scientist. You don't need to be tech savvy even. You just have to understand what it's capable of doing and what you do for a living, like what your tactical work is every day, and then find smarter ways to do it.

[00:35:12] Paul Roetzer: You can, you can make progress in a week. Like it's, it is not a quarter a year. Like you can start right.

[00:35:20] Mike Kaput: That's awesome. And yeah, I think especially too with some of the comments on your initial LinkedIn post, I was curious to maybe end our discussion of this topic with one final question, which is there are still some agency owners and and agency executives out there who it seems like are.

[00:35:37] Mike Kaput: Skeptical still that AI is going to be more than a flash in the pan, be more than just buzz or hype. What do you say to them when you hear that kind of

[00:35:50] Paul Roetzer: skepticism? I mean, I understood it. We've been at this for a long time talking about ai, and you know, I've heard a lot of that stuff through the years and I think I get it.

[00:36:01] Paul Roetzer: Like I could understand it previously because, Prior to the middle of 2022, there weren't really tools where if someone said, you're just so full, like it's not going to be that big of an impact, I think you're overhyping it. I could say, listen, just go try the tool. I'd be like, no, like here, here's what's going to happen.

[00:36:18] Paul Roetzer: Here's what these people are saying and these people are saying, and this is what, what we're seeing. Like it's developing now. I just don't think there's any excuse for it. Like I think if someone's making that case either haven't tested the tools yourself and seen. Or you just don't want to believe it. No matter what you do, you have locked into your position that it's just not that big of a deal.

[00:36:43] Paul Roetzer: And we've seen this all before and, you know, we're over, exaggerating what's going to happen. So I think you either, I don't, naive might, might be the best word I can use, just kind of naive to what's about to happen. Maybe because you, you didn't try it yourself and see mm-hmm. or. You're never going to be convinced.

[00:37:05] Paul Roetzer: I, I really don't know how you could look at image generation, look at language generation tools, play with them yourself. Create these outputs on single prompts for free, or $15 a month, whatever it is. Mm-hmm. and not see a complete disruption of creativity, every aspect of creativity and communi. Being reinvented before our eyes.

[00:37:25] Paul Roetzer: I really don't understand how you couldn't arrive at that perspective. And I'm a very open-minded person. Like I'm the first to admit if I have no clue about something or like aren't that confident in it, I just don't know how you arrive at that perspective. I don't know what arguments you could stand on to believe that.

[00:37:43] Paul Roetzer: Yep. But people believe a lot of things and they have, they have their reasons for it. And I, my position, I was like, I totally respect that. I just hope people don't take your advice , like, it's cool, but, I, I would, I would just, I always, what I always days take, experience it for yourself and come to your own conclusion.

[00:38:04] Paul Roetzer: Don't listen to me, don't listen to some other thought leader, author, speaker, whatever. Synthesize knowledge. Listen to multiple sources, multiple people experience things for yourselves, arrive at your own conclusion. It's the best advice I can give about life in general. So certainly do that with ai. And if you come to the conclusion it's not a big deal and it's not going to affect your career and your business, great.

[00:38:27] Paul Roetzer: Like that's awesome. I don't know what else to do. . Yeah, and we've done, we've done the best We.

[00:38:33] Mike Kaput: We've talked about this a little too. It's unfortunately because of how popular, a traffic topic chat g p t is, we're now seeing 10 times the amount of people suddenly have an opinion on this. The overnight

[00:38:47] Paul Roetzer: AI experts, I like to call 'em.

[00:38:49] Paul Roetzer: Yeah, yeah,

[00:38:50] Mike Kaput: exactly. The overnight AI experts have definitely sprung up. So if you are going. Experiment with the technology and try to learn more about it. Just beware of who you're listening to, I would say, and just make sure they have the

[00:39:03] Paul Roetzer: credibility. , it's an important point, like don't confuse AI with chat G P T as like that's what AI is.

[00:39:12] Paul Roetzer: Right. And I think that's what I've seen more than anything is a lot of people talking very authoritatively about ai, who played around with chat, G P T mm-hmm. , and don't even really understand how the language model un underlying chat G P T works. They just did it. They created some prompts, they created some cool things, and now they think they understand AI and can advise everyone else about ai.

[00:39:33] Paul Roetzer: Hmm. That's what I would be very conscious of is that someone can appear to have a lot of expertise in a very specific area of AI that doesn't cut it. You, you need a depth of understanding of this technology and what it's capable of doing, which again, you can learn in like eight hours, like you can learn it fast, but a lot of the people spouting off about chat G P T right.

[00:39:55] Paul Roetzer: Haven't even taken the time to understand what AI really is. And I just, I worry that too many people are going to listen to and follow those people and we're going to get a bunch of misinformed people at what is the optimal moment. Like this is what we've needed is that one watershed inflection moment where everyone all of a sudden was paying attention to ai.

[00:40:16] Paul Roetzer: We have to seize that and create people who truly comprehend it, and, and I think it's just going to take a lot. I dunno, a lot more work to make sure that that's where the end, that's where we end for

[00:40:28] Mike Kaput: sure. Yeah, that's very, very well said. Well, Paul, as always, thank you again for your time. I really appreciate the insights.

[00:40:36] Mike Kaput: I think our audience is really going to appreciate too, based on our podcast numbers. I think that's true. We've, we've almost doubled our amount of downloads and stuff, so I thank you for resonating.

[00:40:47] Paul Roetzer: So, oh, and I will say, little insider information. We, we just, signed the deal for Marketing AI Conference to come back to Cleveland July 26th of the 28th.

[00:40:58] Paul Roetzer: So I think we, we just shared that with like our Introed AI class today. So we, we haven't announced this officially yet, but if you're listening to this podcast, a little insider information, if you stuck around at the end. go to, Mayon, M a I C O n, ai, and the registration is now open, to join us in Cleveland.

[00:41:14] Paul Roetzer: We'd love to see you. It's going to be incredible. This is our fourth edition of it. Third time we're in person. But yeah. So Macon Do AI marketing conference is coming back to Cleveland July 26th to the 28th. Great. Okay. That's all I got. Fantastic. All right. See what next week brings us, man, .

[00:41:33] Mike Kaput: I'm excited.

[00:41:34] Paul Roetzer: All right. We'll talk to everybody, so thanks so much.

[00:41:36] Paul Roetzer: Thanks for listening to the Marketing AI Show. If you like what you heard, you can subscribe on your favorite podcast app, and if you're ready to continue your learning, head over to marketing ai institute.com. Be sure to subscribe to our weekly newsletter, check out our free monthly webinars, and explore dozens of online courses and professional certifications.

[00:41:58] Paul Roetzer: Until next time, stay curious and explore ai.

Related Posts

[The Marketing AI Show: Episode 12] AI for CMOs: John Dougherty, Brighton Jones

Cathy McPhillips | March 22, 2022

John Dougherty, CMO of Brighton Jones, is a guest on the Marketing AI Show podcast with Paul Roetzer.

[The Marketing AI Show: Episode 11] Author Talks Genius Makers and AI

Cathy McPhillips | March 15, 2022

Cade Metz, author of Genius Makers, is guest on the Marketing AI Show podcast with Paul Roetzer.

[The Marketing AI Show: Episode 15] AI for Event Marketing—Finding Use Cases in Real-Time

Cathy McPhillips | May 16, 2022

How can you use AI to transform your event marketing? In this episode of the Marketing AI Show, we brainstorm AI use cases for events in real-time.