top of page

Speaker Details

Paul Roetzer

Founder and CEO
Marketing AI Institute

Session Transcription

Now, I'm really excited for this next 30-minute session. And I know this event is dedicated to sharing best practices regarding how B2B SaaS companies use metrics and benchmarks to help inform decision-making and leading to success. But in 2023, no event would be complete without including some sessions on the impact of artificial intelligence and how it will change not only the B2B SaaS industry, but business as we know it. When I first heard Paul Roetzer speak, I was blown away by the foresight and vision that he had. And it was really developed over these last seven years since he founded the Marketing AI Institute. And then all of us kind of had to try to catch up starting in November when ChatGBT was first released to the general public. So with that, I am so pleased and proud to introduce you to Paul Roetzer, the founder and CEO of the Marketing AI Institute to the SaaS Metrics Palooza. Welcome, Paul. It's great to be with you. It is certainly an interesting time, isn't it? It is. And I tell you what, when I first met you, I really was blown away. And then I've been reading and following your post and a lot of your content. And it's like, oh my gosh. Paul has been out there in 2023. I think you've had over 60 different presentations on AI, dozens if not hundreds of conversations with companies about AI, including investors and educators. So let me ask the first big macro-level question. What is the reality of the state of generative AI adoption here in October 2023? Yeah, it's something I think a lot about because we do talk to a lot of organizations and especially a lot of marketers and up to CEOs. And I think everyone is fearful that they're so far behind. Like they're just trying to figure this out themselves and they think everyone else has solved it. And based on my experience, having been with thousands of people this year answering hundreds of questions for our Intro to AI class that we teach, the vast majority of people are just at the starting gate too. And I think it's just really important for people to realize that most organizations are just getting started. And that's small businesses, medium-sized businesses, large businesses, B2B, B2C, it doesn't matter. Even SaaS companies, like even some of the companies building the technology, like we've gone in and done educational sessions for the marketers, sales, and service people at the companies that are building AI tech. So it's just so early in this. And to me, I think that's where the opportunity lies, is that once people get past that feeling that they're behind and there's just no way to catch up and they realize, no, you can catch up very quickly and even get ahead, then people start to get excited about the potential of AI. Since our audience is almost exclusively B2B, SaaS, and cloud executives from CFOs, CEOs, and go-to-market leaders, I was really interested in something I heard you say once. That is, a lot of people are really investing the majority of their time in understanding and deploying AI as a product extension or as a technology. And that's so true for SaaS companies, right? But yet, at the same time, you have a very unique insight that companies are almost under-investing, if not investing at all, in the change management that's going to be required to truly harness the power of AI And truly. Can you share a little bit more of your insights on that one? Yeah. So, I mean, so many people come to Marketing Institute. So many people reach out to me for speaking or workshops. And they're just trying to find the use cases to pilot within their organization. They're trying to find the technology they should start with. I get asked that all the time. Which tech should we be using? If you could just pick one, what should we be using? And so the thing I advise them and the thing I think about a lot is, let's say my answer to you is GPT-4 or that it's Anthropic Cloud or it's some AI-assistive tool for language. Well, if you think about all the ways within an organization that people write, it's not just marketing and sales and service professionals. It's HR and legal and finance and accounting and C-suite and operations. Like, everyone writes all the time, from articles and emails to proposals. And so if you inject just an AI writing tool, and let's say it's GPT-4. You get the chat GPT-Plus for $20 a month. You get everybody on your team access to that. Who is training them how to use it, how to prompt it, how to improve those prompts, how to save and share their best prompts across the organization, how to make sure no private or confidential information is getting put into the system that goes into future training models? If we unlock efficiencies from AI, which is the promise of it, that a writer maybe saves 20%, 30% of their time or an email marketer, an SEO professional, who is teaching them what to do with that additional time? Do they just do more of what they did? Do they work fewer hours? So when you start to really think about this, this isn't a technology thing. The technology is the enabler. But what it does to a corporation when you put these tools in that actually work and actually do save meaningful time or increase productivity in meaningful ways, I have yet to meet an organizational leader who is thinking deeply about this and building their teams around it and rethinking what it means to their hiring and training and reskilling and all of these things. So that's, to me, the opportunity ahead, again, is to think about this more deeply within your organization. It's not just about finding the tools and technologies. It is about how is this going to transform our organization. And there's pros and cons to that. And again, you have to be thinking about it at a deeper level. Yeah, I'd like to get your insights on this because I see so much of a groundswell and a bottoms up kind of model. That is early career professionals or someone in the company said, I really see how this could improve the productivity of my job or my performance. And it doesn't seem like a lot of the leadership actually has a clear understanding of how AI can be leveraged. And that's part of the challenge of I can't lead initiatives if I don't really understand the technology. Yeah, we always start when we teach like, what do you do about AI? Education and training is always the key. And it needs to start at the top. Like the leadership needs to deeply understand what it actually is. And for me, that comes from comprehension where you actually take the initiative to learn what it is and what it's capable of doing, get rid of the sci-fi and the fears and all these things and really learn at a fundamental level what it enables. Then there's the competency piece, which comes from experimenting it with yourself. So what happens a lot of times is that the lower level, the younger employees or even like the managers, directors, they're playing with the technology. They're seeing what GPT-4 can do. They're experimenting with Claw. They're playing with Google Bard. Like they're out trying things, Mid Journey, Dolly, like they're experimenting with these generative AI tools. And so they're learning through experimentation. They're building these competencies because they can actually talk about what it is good at, what it's not good at, where its flaws are. So for the leadership to really be able to steer their team and their organization, they have to invest the time to understand it. And they have to invest the time to become competent at using the tools. Because you and I can sit here all day long and go through a bunch of use cases and sample tech, but until you get in and try them, you really can't comprehend fully what they're going to do to your organization. Paul, do you have any stories or best practices that you can share of a company or two that have really done a good job of galvanizing and organizing around introducing, or at least evaluating how AI could impact and how they're introducing it to different functions across the organization? Is it working communities by department or what are you seeing out there as best practices early on? I don't know that there really are yet. I mean, the way we teach it is you should form an AI council. So if you do just within a marketing council within your department, but find the people in the organization who are curious about this stuff and willing to be kind of on the frontier experimenting with the technology and put some formal structure and resources to them. So VMware is an example here. Like we started talking with VMware back in around February, 2023. They started with two people and like a very informal AI council. And today I think they have over 30 people. They have regular meetings. I think they have a Slack channel where they're sharing resources. They take our Piloting AI for Marketers course. So they, like everyone on their team is trained within specific learning journey of AI. They do intro then to piloting. So they've actually taken the initiatives to do that. And then we train this idea of you have to have very structured pilot projects. So you can't just flip a switch and become all AI all the time. You have to pick the things where there's going to be immediate value and you have a high probability of it working. So what we do is like kind of walk through a framework of trying to identify what are the best ways to start testing AI in a more structured model. And so you pick some pilot projects like AI writing tools, podcasting, whatever the things your team already does and you find ways to use AI to assist in that process. And then you take the learnings from that and you try and say, okay, what would it look like if we now scale this project? So let's say we're doing it for one podcast and we have five podcasts. Let's test it with the one, let's prove out what's possible and then let's take that learning and apply it to the other five. So VMware is a good example., we had Dan Slagan, the CMO of did a talk at Maco this year and theirs was more just kind of organic. They decided that they could be more efficient as a marketing team. They went and hired an AI marketer. They trained from the beginning for her to have kind of this cross-discipline capability to use these AI tools. They encouraged that. So that's kind of almost like a different approach where they just said, let's just do it. And they just went and did it, which is going to be more in like the SaaS world. A lot of times that's what happens. It's just like iterative testing, get in, try things. You don't have three months to plan it all out. So we're seeing a lot of different approaches but the general way we teach it is you have to go education and training, build an AI council, develop some generative AI principles that guide for your team how they're allowed to use these tools. Are we or are we not allowed to use AI generated images in our blog posts? Are we or are we not allowed to use GPT-4 to help write articles, to outline articles? Can we put transcripts into GPT-4? Your team doesn't know the answers to these things. And so that's really where we talk about this need for really understanding the bigger impact this has on an organization and starting to look for it. So we're just starting to see the kind of best practices emerging but there's no one right way to do it. It's dependent upon your industry, the structure of your organization and what you know is gonna be possible within that company because there's a lot of barriers to doing this stuff right. Paul, do you think it makes sense because a lot of our audience here will kind of be, I would say smaller, not classic enterprise, maybe in that $10 billion to $100 million range. I know it's a broad range. Should they go to the extent of maybe having an AI policy and guidelines about what is and is not acceptable or is that overkill for earlier experimentation? No, I mean, I think every single organization, if it's a two-person organization, like any size should have that because even for me, so market answer is seven people, to have the team not know what they're allowed to do is a misguided approach. Because again, people are testing it already. Like they're already probably using it to write stuff and you may not even know it as a leader that it's happening and maybe they're not doing it in the most secure way. And again, whether that's two people or 2000 people, it's the same basic concept. It's about the integrity of your brand and your organization. It's about the brand voice. It's about all of these things that your organization stands for. And you have to make sure that there's uniformity in that approach. And then there's the privacy and security and all these other concerns. So yeah, I think generative AI policy, like you can get those in place in an afternoon. It doesn't have to be anything crazy and it can evolve, but it's literally just saying, if we think about generative AI, like we think about it in five buckets, you have language or text, you have images, video, audio, and code. Kind of these five major categories that AI is able to assist in creating. You should have just very basic policies that say how and when you're allowed to use AI in those different buckets. Because you have to consider copyright issues and things like that. Like the example you're asking, like a two person shop, would it be applicable? Think about marketing agencies or outside agencies. They work under work for hire agreements. So let's say a B2B SaaS company is hiring an outside agency and they have a work for hire agreement for creative and ads and things like that. If that agency is using generative AI to create it, passing it along to the client as work for hire work, they can't own the copyright to it because you can't copyright something AI created. And so you have to address this no matter the size of the company. That's a really important point is, if your vendors are leveraging generative AI technology, you should have some terms and conditions of that also when you're agreeing with that. You have to, and it's not happening. Like people don't know to do that. It's crazy. Well, about 50% of the audience here today are finance leaders, VPs of finance, CFOs. And they're gonna be asking their marketing leaders, et cetera, oh, you wanna invest in generative AI? How are you gonna measure that return on investment? Is it productivity improvement? Is it better demand gen performance? How would you respond to CFOs? How do I measure the returns here, Paul? So I've said for years, there's two reasons you use AI, reduce costs and accelerate revenue. Like there's a bunch of other KPIs you can look at, but at the end of the day, if you're doing AI, it's just technology, it's just smarter tech. And it's like any other tech. I'm buying the tech to either grow my business or reduce the cost. So the main thing, what ends up happening, especially in bigger companies, is they look at it as an efficiency and or productivity lever. So I can do more in less time and I can increase my productivity. So I can just like create more output or I'm gonna keep the output the same and I'm gonna spend less time creating that output. And so from a CFO perspective, from a finance perspective, they're going to look at it as an efficiency slash cost reduction activity. And so that may mean, this is why I'm concerned about a very significant impact on the knowledge work labor force, is I think that a lot of organizations are gonna look at this and say, okay, so let's say we do coding, let's say we do writing or SEO or email or whatever activity, even in accounting, like auditing, like anything. Like let's say we're doing this repetitive data-driven task over and over again, whatever the function of the business is. If we can use tools to help us do those things, the same 10 things each week, 20% more efficiently, 30% more efficiently, 50% more efficiently, that is a likely cost savings. And the question becomes, this is the question every organization is gonna have to deal with, do we just need fewer people doing the same work? Like we're not replacing humans, the AI is not at the point where it's gonna replace the need for these professionals. We just aren't sure yet if we're going to need as many of them doing the same amount of work. So if you're in an industry where demand doesn't scale up if you're able to produce more, that there's kind of a fixed demand and you can meet that demand in less time, then inherently you need less people. And so this is, it's the challenge we're all going to face is like brands are gonna have to make these choices about what are we going to do if we actually start saving this time? And this isn't a three years from now conversation, this is a now to three months from now conversation with a lot of industries. Yeah, and as head of finance, I'm like, wow, I can reduce my operating expenses, I can increase profitability. The short-term benefits on an Excel model seem great. 100%. The longer term social impact and even culture impact of your organization may be larger than we even contemplate. Do you have a lot of organizations really trying to try to understand what that cultural impact is gonna be yet? Or is it too early? No, not nearly enough. It is honestly still at the point where you get in front of a room of 500 CEOs and you show them this stuff and you have to pick their jaws up off the floor. Like they're not having these conversations, they're not considering like the real impact this could have. And again, so I look at in the United States that 132 million full-time employees, 100 million of them are knowledge workers. People who think and create for a living. Accountants, marketers, lawyers, doctors, like all these people. When you look out and you project in each of those industries, the impact AI could have in one to two years, you have to ask that same question in every single one of these industries. Do we need as many people doing what they're doing? If AI comes in and can write briefings and analyze data and things like that at scale, do we need as many people? And then the debate is, is AI is also gonna create jobs, which it will. We just don't know. And the best economists in the world don't know how long it's gonna take for those new jobs to be created and what those jobs are going to be. It's a very gray area right now. And there's just no clear guidance. There was a study that just came out from Harvard Business School where they teamed up with BCG and they analyzed 758 consultants with and without GPT-4. And they looked at the lift in productivity and creativity of the workers. And it was significant, the impact that the AI cohorts had, the people who used the AI had over the people that didn't. Significant. And they looked at it against benchmarks, so. I'm so glad you mentioned that because the partner at BCG in charge of that research is actually gonna be here at SaaS Metrics. No, nice. Palooza, his name is Pranay Arwal. So he's gonna be sharing all the findings they had from that research. Yeah, go deep. It's fascinating stuff. We've been needing that kind of empirical data in the industry because it's a lot of assumptions right now and it's one of the better studies I've seen that puts real numbers behind it. Yeah, so recommend everyone to follow Pranay here at SaaS Metrics Palooza. I'm trying to think which way to take this, but I was at an event last week and there was a lot of all the big foundational platforms for LLMs were there. You know, AWS, Google, Microsoft, and of course OpenAI. And there was just discussion, Paul, about on a product expansion and integrating large language models into your platform that you have to go evaluate and experiment with it, but you don't know your costs because there's such a lack of capacity of infrastructure to really process all this capacity required for LLMs. Are you involved in that part of it all? And what would you recommend to our audience out there? It's like, you may not know the cost, but you have to make the investment and deploying LLMs into your platforms over time. Yeah, it's definitely an evolving area. We do have some knowledge of kind of what's going on in the space. We talked to a lot of larger enterprises that are looking at different ways to infuse this technology. The way to think about it is there's basically two approaches here. So you have your foundation model companies and I would throw like Cohere and Anthropic and Inflection into that bucket. Stability, AI, those are the companies that are building these foundational models. And then you have the application layer or the SaaS companies that are building user interfaces on top of these models. So in like Writer and Jasper, Writer just raised $100 million, Series B, Jasper raised I think 125 million last year. So these companies are building on top of foundational models to build like enterprise level solutions, like platforms for language, where you can train it on your style guides and templates specific to your organization, fine-tune it on your data. So there are ways to do this with companies that are already building these enterprise-friendly solutions. And you can get value faster. It's probably more affordable. You're paying in a traditional SaaS model where it's a license fee per user. If you go work with a foundation model directly like an Anthropic or Cohere, it's probably a more expensive build. You're doing a lot more tuning, a lot more custom work. And so that's the challenge a lot of organizations are facing right now is like, what do we do? Like, how do we actually get started here? How do we not make the wrong choice? Like build on open AI and then realize, oh, we should have built with Cohere. Like we would have had a better choice. So that's the debate a lot of CIOs are having. Like you're having these kind of challenging conversations, certainly at the CFO level, these aren't gonna be insignificant investments. Like you're talking about multimillion dollars. I think I saw some data that McKinsey was charging like five to 10 million for an engagement to like custom train a language model. So in the big enterprise, you're looking at massive investments. But again, at the small to mid-size level, there's more affordable ways. Like we get 20 bucks a month, go get JetD plus. Like you can prove out the value of this. And so I often advise people like start small. There's enough tools available for free or very minimal where you can pick three to five use cases in your organization and say, let's just pay the 20 bucks a month for three months. Let's run a test. Let's properly train our team, onboard them on the technology, give them sample prompts, like make sure they know what they're doing. And let's look at a user story today. And then let's look at one three months from now. So today we do these 10 steps. It takes us this long to do it. And here's the quality of the output. If we three months from now look out, we do the same 10 steps, is the AI doing? How much is the human doing? How much time is it now taking? And is the quality of the output increased? And so every instance where you're looking at AI in your business, you could follow that same basic framework. Look at what it looks like today and what would it look like tomorrow? And then is the efficiency gain 10% in the first three months? And then does it cap at 50% after six months? And then it kind of plateaus. So you want to really think about this technology and the impact over a period of time. It's not like you turn it on tomorrow and you're 50% efficiency gains and that's it. Yeah, and we're very fortunate. Also OpenAI is also going to be here at the Palooza. The head of enterprise is going to be talking about some of their customer success stories where they can now provide some of that security that a CIO is looking for. At the same time, tipping your toes and getting some of the benefit. One last thing, I'm going to zoom back out. You mentioned something to me in a previous conversation that I thought was real enlightening. AI is not new, Ray. We've been doing machine learning and predictive AI for the last 10 years in marketing. It's generative AI, which is new. Can you just spend a couple of minutes kind of sharing what you mean by that and what's the difference? Yeah, so I mean, AI as a discipline goes back to the 1950s. This theory that we could build these intelligent machines with human-like capabilities of reasoning, of language. Like this isn't new concepts, but we went through kind of these AI winters in the 70s and 80s and even in the 90s where it just didn't reach its promise. And then what happened is around 2011, there was a breakthrough in computer vision with Jeff Hinton and his team. And what they did was they proved that there was this concept of deep learning that we could actually build these neural nets that could function like a brain. And we could really make leaps forward in vision and language and understanding and reasoning. And that started this massive like shift in how we thought about AI and the ability to commercialize it and really build valuable tools with it. Then in 2017, there was a breakthrough from the Google Brain team. They released a paper called Attention is All You Need that invented the transformer architecture, which is the basis for GPT, generative pre-trained transformer. That was a language model breakthrough. And that is what then created this massive sea change in what AI was able to do with translation and transcription and speech to text and text to speech and all these things that we're now seeing come through. Chat GPT was the moment where we all all of a sudden got access to it. So Google was building this technology, Microsoft, Amazon, like everyone was building it. Open AI released it to the world. And that started this kind of like ultra competitive phase we now find ourselves in where every month or three months, each of these major tech companies is trying to kind of one up the other one and what they're releasing and the power of these models. And now everyone's trying to build more powerful models. So now the next version like GPT 4.5 or GPT 5, Google's working on Gemini as the model there. Apple supposedly we're going to call Ajax, like inflections building when it's going to be a hundred times more powerful than what they have with Pi. That's the world we're in now is we had machine learning for decades. We could do predictions, recommendations, personalization, but we couldn't generate content at scale. Now we're into the world where all of us can generate anything we can imagine basically. And we're only at the beginning. Like we like to say, this is the least capable AI you will ever use that we will ever have in human history. It only gets more powerful from here. And this is going to be a softball, but I was involved with e-commerce in the early nineties when it first was a concept, right? And it changed my entire career trajectory. So what would you say to those early career professionals who are out there right now, even mid-career, how would you advise them on how they should invest learning and leveraging AI to chart the rest of their career journey? Yeah, so I like to say, don't wait for the world to get smarter around you. Like this stuff is coming. It is not if, it's when and how impactful it will be in every industry, every business, every profession in the knowledge work world. And I think the people who seek like a broad understanding of this and then connect their domain expertise to how to apply it. So if you're an insurance or if you're in law or if you're in accounting or if you're in marketing or whatever you're in, you're an expert in that. And maybe you're an expert within an industry. If you can then go learn AI and apply that domain knowledge to it, you're going to be the one that connects the dots and can be a change agent within your company, within your industry. And so again, as we started the conversation, don't feel like everyone is ahead of you. Like you can learn this stuff relatively quickly. I teach an intro to that class, it's 30 minutes. Like you can get a baseline understanding in 30 minutes of what it is. Then you figure out what is the best way for you to learn. Is it books? Is it podcasts? Is it online courses? You don't have to go back to school for this stuff. They're not even teaching this stuff in school. So like, I wouldn't advise that as like the thing you think you need to do. You can self-teach this stuff. There's so much resource out there to do that. I just encourage people to take the next step. Like if you're curious about it, go find the next thing to do, read a book, listen to a podcast, whatever that is for you. Well, I know that as part of your mission is to help inform and educate industry, right? On how to take advantage of AI. Tell me a little bit about how Marketing AI Institute could help people in that journey and that self-directed learning journey. Yeah, so we think about building learning journeys for individuals and for companies. But I mean, we offer a whole host of free solutions. There's blueprints you can download related to industries and categories. There's free courses, our podcast, the Marketing AI Show is a weekly show. It's free, we have a free newsletter. So like start there with our stuff if you're interested in it. And I would say like our podcast is very much broader business. It's not really marketing focused. And then from there we have events and online courses and things that are more in the paid realm. But start with the free stuff. Like just go absorb as much as you can. And when you're ready, then find your learning journey. Like find where you really wanna invest in it. If you're a leader in an organization, I would seriously think about trying to infuse AI education into whatever your professional development programs are. They really, it really needs to be a part of it or you're gonna be at a competitive disadvantage in the months and years ahead. Build it into every employee's 2024 objectives. I love it. So we're gonna put your contact information up here on your screen, Paul. We have your email. I encourage people to follow you on LinkedIn. You just have some of the most insightful posts out there. And I just thank you for investing 30 minutes of your time. You're one of the most popular speakers out of the circuit today. And the fact that you allocated time to the SaaS Metrics Palooza audience, it means the world to us. Thank you so much. Paul Roetzer, founder and CEO of the Marketing AI Institute. Thank you. It was great to be with you.

bottom of page