AI Strategy Starts With Leadership, Not Technology  🔗 **Source:** [AI Strategy Starts With Leadership, Not Technology](https://share.snipd.com/episode/e2a7364c-5f67-40aa-9fed-3d61c36f7c5a) ## Highlights - **AI As The Operating System Of Business** - Paul Roetzer predicted AI would weave into every part of business and become an underlying operating system rather than a niche tool. - He traced this belief to early Watson experiments (2011–2014) and launched Marketing AI Institute in 2016 to research real use cases before ChatGPT accelerated interest. Transcript: Paul Roetzer Yeah. So I'll give the quick origin story. So actually it goes to the PR 2020 days. In 2011, I wrote the marketing agency blueprints. That was my first book. And at the time we were a few years into being HubSpot's first partner and kind of at the forefront of marketing technology and social media and inbound marketing and content marketing And all of those things. And that was the year IBM Watson won on Jeopardy. And I became obsessed with understanding how that technology worked. And then could it actually be applied, that same idea? It was basically a prediction engine, take data in, understand the language behind it, and then you make predictions about what comes next. And so I started working on this concept of what I was calling a marketing intelligence engine. And this was back in 2012 and 13. And the premise was if we could use Watson-like technology to predict what to do next, what next best action, next strategy, how to spend our marketing dollars, then we could build an Entirely new way of doing marketing. And so that was the original hypothesis. And I shared that idea in my 2014 book. And then that was like out of the 50,000 word manuscript, it was like a thousand words. And the book was not about AI otherwise. And that was all anybody wanted me to talk about. And so fast forward to 2016. And we were like, well, what do we do with this? Like, I'm really intrigued by it. I'm convinced it's going to change marketing and business in the world. But like, I don't really know what's possible. So we created the Marketing Institute to research it ourselves and then tell the story of AI, like what was real, what was happening. And so yeah, we created the Marketing Institute in 2016. And then like I always half joke, like we survived long enough financially for ChatGPT to show up. I sold my agency in 2021, focused exclusively then on AI and the Institute and eventually SmarterX, raised a seed round of funding that kind of got me through the really lean years. And ChatGPT came and all of a sudden the interest in AI exploded. ([Time 0:02:27](https://share.snipd.com/snip/bf5f37ea-3b86-422d-a3f9-4127e81aaa39)) - **Backlash Is Predictable And Requires Empathy** - Public backlash against AI is natural and will grow because technologies have both positive and negative societal effects. - Roetzer compares AI to the internet: inevitable benefits plus dark side consequences that require empathy and mitigation. Transcript: Paul Roetzer Think it's going to grow significantly i think it's to be stoked by interest groups that want it to grow. And then I think it'll naturally grow because people's lives and communities are going to be impacted in negative ways. So I always like the closest thing I can equate to try and make it tangible for people is, you know, if we go back to 1994, 1995, the Internet's like becoming a real thing in society. And at that moment, we said, you know what, there's going to be this thing called the dark web where these like horrible people do horrible things. And it's going to cause like online bullying. And like, you're going to have all these downstream super negative things that happen. But we go back and say, but would we still build the internet? Yeah, like 100 times out of 100, you would probably still build the internet because it has changed society in a bunch of profoundly positive ways. And I think AI is going to be the exact same thing. There's going to be absolutely negative things that happen as a result of it, whether it's building of data centers in communities that don't want them, job loss and displacement, Whatever. Those things are going to happen. They're a byproduct of it. But if all goes well, it's also going to transform health and create growth engines and opportunities we've never had before and solve mysteries in the universe. Like it's going to do all these things, too. ([Time 0:08:05](https://share.snipd.com/snip/34303030-acc8-4d03-b41d-77dfe17327e8)) - **Regulate Harms Not Compute Limits** - Support thoughtful regulation focused on harms rather than naive technical limits like compute caps. - Roetzer notes lab leaders increasingly call for guardrails, but early regulation risks becoming obsolete quickly. Transcript: Paul Roetzer Yeah, there's growing like so recently Demis Asabas posted online about the need for regulation and X presenting a framework. He's the co-founder of Google DeepMind. Anthropic has made proposals around regulation frameworks. Sam Altman has called for regulation on Capitol Hill. Regulation can be done where it actually has the negative effect on society. So there's this like very fine line that I am not the expert in by any means about how to do regulation well. There are very few people that are building the technology who don't think that there needs to be some protections and guardrails in place, that we don't have to stop and say, this is Going to have a serious impact. We should be thinking more deeply about it. The challenge has been the leaders of these labs, they're so focused on just building the technology and competing with each other and competing with China and other countries. John Jantsch China made money. Yeah. Paul Roetzer They don't sit around and think about the writers who are going to lose their jobs. Like it's just not, and they live in a bubble where it's like, they're all just technologists and engineers and like, they're all going to have jobs for the foreseeable future because They're all growing and hiring more of those people. But they don't think about the average knowledge worker and the impact it's going to have. So they're hiring economists and philosophers and like they're trying to now consider it. But for a long time, they were just heads down, accelerated at all costs. John Jantsch Yeah. Well, and I think, unfortunately, when it comes to regulation, you know, you think about the government bodies that are going to decide they need to regulate this. I mean, they can't even line a swimming pool, you know, so the idea that, sorry, that was a cheap one, but the idea that they're going to actually, you know, regulate an industry like this, You know, is probably kind of frightening. Paul Roetzer Well, yeah, and they don't understand the technology and where it's going. Like the idea originally a couple of years ago was to limit it based on how much compute was needed to train a model. Well, that's laughable amounts of compute these days. And then they just find ways around it. ([Time 0:10:25](https://share.snipd.com/snip/a83c7cff-aee5-4d05-9c66-f0de7fa84239)) - **Transformation Fails Without CEO-Led Vision** - Roetzer defines eight pillars for AI transformation: vision, strategy, data, technology, governance, literacy, people, and performance. - He observes most companies treat AI as a tech project, lacking CEO-driven vision and situational awareness needed for full transformation. Transcript: John Jantsch That's something that you have written about. Hopefully you remember writing about that. I'll name them for you. Vision, strategy, data, technology, governance, literacy, people, and performance. Those boxes is you said no company has ever passed this test yet. Where do companies break down in terms of any of those elements when it comes to transformation at an organization? Paul Roetzer Yeah, so this is a relatively new concept that I shared. It's part of a larger transformation system that I'm developing. And it's like the first piece to it because we talk to a lot of companies of all sizes, small, midsize businesses, large enterprises, and everybody's trying to figure, like, what does It actually look like? We throw out this term transformation, but, like, no one really has quantified how do we actually do that. And what we've seen time and time again is, especially in larger enterprises, but it happens in small businesses too, just treat it as this technology problem. Like, oh, we just got to go get some ChadCBT licenses and give them to people, and, like, then we're going to get all these amazing benefits of AI. John Jantsch And the CTO's in charge of it. Paul Roetzer Yeah, and they throw it into the technology pool to do it. What needs to happen, and the fundamental flaw that we see is a lack of situational awareness and vision from leadership. And so my, like, if I boiled this down to one simple thing, the CEO has to drive the transformation. Like, it has to be so important to the organization that the CEO has embedded him or herself in the deep understanding of the moment, of what the technology is capable of, of the impact It's going to have on their organizational structure, their people, their products, their markets. And if the C-suite doesn't have that, then you are not going to see a complete transformation within an organization. So vision and strategy from the leadership on down is what's fundamental. What's driving most of the innovation and transformation companies so far is actually bottom up where people are just like bringing their own devices to work or getting their own personal Accounts and just like doing their own thing. And then sometimes that turns into a collective of people doing their own thing. And then maybe a department's like, oh, let's form around this and let's get a marketing AI counsel or something. But what often lacks is that top leadership that truly understands this needs to be one of like the three biggest priorities we are working on as an organization. ([Time 0:13:11](https://share.snipd.com/snip/44c6290d-d398-4048-ba42-cbd41b946d94)) - **Start With AI Literacy Not Licenses** - Prioritize AI literacy across the organization before buying tools or licenses. - Roetzer says C-suite must gain competency to know when to use tools like ChatGPT and what good outputs look like. Transcript: Paul Roetzer Yeah. So, I mean, literacy is the fundamental thing. So, it's number six on my list, but it's actually probably number one overall because even the C-suite needs AI literacy. They need the knowledge and the understanding and the belief system around AI and its impact before they can prioritize it strategically within a business. So developing understanding of AI capabilities, the comprehension of what it is and what it's capable of, and then the competency to use the tools in an intelligent way. That you know when to go in and ask ChetGPT for help. And then you know what good looks like. So AI literacy is actually the foundation of all the other components. And then if you do that individually and you go through the organization and say, OK, we're going to raise the skill level of everyone, the understanding of AI and the ability to work With it, then you can move the organization forward more, not only efficiency with higher efficiency and productivity, but drive actual innovation and growth as a result of it. ([Time 0:16:20](https://share.snipd.com/snip/7ac7135e-ce49-4731-a791-49a693c5a5b0)) - **Real Time Reporting Replaces Monthly Reports** - Real-time AI dashboards replaced monthly reporting at Roetzer's company, turning 15-day reporting work into live narratives. - Example: COO connects data to tell KPI stories instantly instead of producing delayed PowerPoints. Transcript: Paul Roetzer Like 90% of what we did for clients, AI is capable of doing now. And so like the perfect example we gave was, we used to like give monthly performance reports to clients by the 15th of the following month. So you'd wrap the month up, you'd organize the data, you'd put it into the thing, you do the analysis, you would create the PowerPoint, you'd schedule the meeting. And by the middle of the month, you were talking about what happened the previous month. We now at SmarterX, our COO runs those things in real time. So like at any moment, she has it connected to the data. She can tell the narrative of what is happening across all of our KPIs. And boom, here's the update in Zoom. Stuff that we used to spend dozens of hours creating on a 15-day lag, we now do in real time. And so when you apply that across entire businesses, all different departments, you start to realize how different we can run companies today. ([Time 0:17:33](https://share.snipd.com/snip/529c3928-e4de-4fb9-8afb-9ce352008cdf)) - **Entry Level Work Disappears Creating A Career Gap** - Entry-level work that taught domain expertise is being absorbed by AI, breaking traditional junior-to-senior career paths. - Roetzer proposes apprenticeships where companies reinvest AI-driven productivity gains to train new talent. Transcript: John Jantsch Thing that, and I said it in my beginning kind of question, was that also trained a lot of people, right? A lot of the people that did that work learned a lot about marketing by doing that work. And they're now missing that. How do we fill that gap? Paul Roetzer I don't know. It is the focus of my Macon 2026 keynote. So the name of the keynote is the architect, the orchestrator, and the apprentice. And my basic hypothesis is that we have to redefine entry-level work because the tactical things that all of us did to become experts, to know what good looks like, to have judgment and Taste and to be able to work with these amazing tools in a responsible way, we can do it because we did the data-driven repetitive work all those years and learned right from wrong and Good from bad and things like that. And then we edited other people's work. And it's like, if you remove all of that work from the first three to five years of our careers, how do we get to become the experts we all became and have that domain expertise and institutional Knowledge? And so I don't know the answer, but my current theory is that it looks something like an apprenticeship. That organizations will have an increased revenue per employee number as a benefit of AI. So you use AI to run a more efficient business, thereby generating more revenue per employee. But rather than putting that straight to the bottom line, you reinvest a portion of that increased revenue and profit back into developing entry-level talent through an apprenticeship Program where they don't have a direct impact on revenue. They're actually an expense item for the first maybe two to three years of their career. And so that's a theory, but then you actually have to operationalize. Well, okay, if that actually is a viable idea, how do we do it? How do we train them? How do we use these tools to advance their learning so they still come out after two or three years with not only the level we had after two or three years, but maybe like 2x that. So we actually accelerate their learning, their taste, their judgment, their capabilities by leveraging AI technology to train them in new ways. And I have yet to meet a single leader of any company of any size that has solved for that. ([Time 0:18:34](https://share.snipd.com/snip/8eb7481a-6a78-4802-9366-07c73fb5ae1e)) - **Reinvest Productivity Gains Into Apprenticeships** - Consider formal apprenticeships that treat early hires as training investments for 2–3 years rather than immediate revenue drivers. - Use AI to accelerate learning so apprentices exit with greater judgment and taste than previous generations. Transcript: Paul Roetzer That organizations will have an increased revenue per employee number as a benefit of AI. So you use AI to run a more efficient business, thereby generating more revenue per employee. But rather than putting that straight to the bottom line, you reinvest a portion of that increased revenue and profit back into developing entry-level talent through an apprenticeship Program where they don't have a direct impact on revenue. They're actually an expense item for the first maybe two to three years of their career. And so that's a theory, but then you actually have to operationalize. Well, okay, if that actually is a viable idea, how do we do it? How do we train them? How do we use these tools to advance their learning so they still come out after two or three years with not only the level we had after two or three years, but maybe like 2x that. So we actually accelerate their learning, their taste, their judgment, their capabilities by leveraging AI technology to train them in new ways. ([Time 0:19:44](https://share.snipd.com/snip/e534edd8-d15c-49fe-b19d-89d67bd8af9a)) - **Efficiency Gains Will Pressure Staffing Decisions** - Many companies haven't yet realized the potential to reduce staff because AI can let senior roles absorb entry-level tasks. - Roetzer warns firms growing under 10% face pressure to cut roles as CEOs discover efficiency gains. Transcript: Paul Roetzer No, I don't think we're any, I don't even think we've scratched the surface of people realizing that they can reduce their staff. Like, So my basic premise here is I do think that AI is going to drive a lot of innovation, a lot of new businesses, a lot of growth and jobs through entrepreneurship and creation. But when I talk with leaders at enterprises who are under these very near-term financial requirements to run the company where you have to either be growing, or if you're not growing Fast enough, you have to be cutting expenses to still maintain the profit margins that are required. In those businesses, it's really hard to sit there and say, if you've had 15 marketers for the last 10 years, that you still need 15 marketers. Because if you train someone properly, like a manager director level can do a lot of the entry level work where you maybe just don't need that entry level hire you were going to make this Year. And so in companies that aren't growing, I think it's very hard to make an argument that they will maintain or increase their staffing levels. I think companies that are growing less than 10% will be under tremendous pressure in the very near future. Once their CEOs realize what's possible, I ([Time 0:22:01](https://share.snipd.com/snip/67aecb4a-a342-481c-a1a5-75edd1ff4e6c)) - **MAICON Growth From Niche To Mainstage** - Roetzer described the Marketing AI Conference (MAICON) growing from 300 attendees in 2019 to over 2,000 and expanding to strategic and applied tracks. - He invites leaders to MAICON Oct 13–15 in Cleveland and offers code DUCTTAPE150 for $150 off. Transcript: Paul Roetzer Yeah, So Mekon, this is our seventh year. It's hard to believe. I posted something recently about how crazy it actually seems in retrospect. We started this conference in 2019, three years before ChatGPT. We were running an AI conference for marketers. Sometimes I struggle to think like, what were we teaching at that point? But it was a lot of like, here's what it could become. Here's how to find use cases. Here's companies that are building, you know, like email subject line writing tools and predictive modeling for ad spend and things like that's what we were focused on back in those Days. So it's become something much larger. That first year we had 300 attendees from 12 countries. This year will be well over 2,000. I know, I think we had 19 countries already represented last time I saw it. And we basically break it into applied AI and strategic AI. So now there's like two fundamental tracks, track for leaders that are thinking more big picture about the impact on the organization. And applied AI is all about use cases, technologies, things like that, where you go in and then we have build sessions and workshops. So it's an incredibly immersive environment. It's a great community of other AI forward marketers and business leaders. So if you're trying to find your people, try and find that community of other people who are thinking and trying to work toward like a human centered approach to this, that's what Macon Is all about. So yeah, you can go to Macon.ai. It's M-A in Cleveland, October 13th to the 15th. And then duct tape 150 is the promo code for $150. ([Time 0:23:31](https://share.snipd.com/snip/a6415e7e-9eff-49c6-943d-a4504d151449))