Earley AI Podcast - Episode 32: AI Strategy for CEOs with Glenn Gow

Winning the AI Race: How CEOs Can Harness Predictive AI, Data Strategy, and the Winner-Takes-All Flywheel 

Guest: Glenn Gow, CEO Coach at The Peak Performance CEO Coach 

Hosts: Seth Earley, CEO at Earley Information Science 

             Chris Featherstone, Sr. Director of AI/Data Product/Program Management at Salesforce 

Published on: June 3, 2023

 

 

 

In this episode, Seth Earley speaks with Glenn Gow, known as "the AI guy" and a CEO coach with 25 years of executive experience advising companies including Apple, Google, Microsoft, and Oracle. They explore what CEOs most commonly misunderstand about AI, including the difference between generative and predictive AI, and why companies that move quickly risk being permanently outpaced by competitors. Glenn shares a practical framework for launching bite-sized AI initiatives tied to clear business strategy, building internal centers of excellence, and treating data as the foundational corporate asset that makes AI possible. 

 

Key Takeaways:

  • CEOs often equate AI with ChatGPT alone, overlooking the deeper strategic value of predictive and analytical AI tools.
  • Companies that excel at AI and continuously collect customer data create a compounding winner-takes-all flywheel competitors cannot overcome.
  • Disintermediation is real and fast - Chegg's stock dropped 45% in one day when ChatGPT threatened its core business model.
  • Enterprise AI success depends on treating data as a corporate-wide strategic asset, not a fragmented collection of departmental silos.
  • A strategy-first approach - choosing one focused initiative before selecting AI tools - dramatically improves implementation success rates.
  • Most enterprises cannot build a full AI team and should instead establish a center of excellence supported by consultants and vendors.
  • Cultural change driven from the CEO level is the single biggest determinant of whether AI initiatives gain traction across the organization.

 

Insightful Quotes:

"Become good at all the tools that are being made available to us, because that's going to create opportunity for you." - Glenn Gow

"There's no AI without IA. There's no artificial intelligence without information architecture - that's the foundational piece." - Glenn Gow

"You can't automate a mess. You can't automate what you don't understand. So understanding that process first, then having that very precise intervention - that's what makes AI work." - Seth Earley

Tune in to discover how CEOs can move beyond the ChatGPT hype, build a winning AI strategy grounded in data, and avoid being left behind in a winner-takes-all market.

 

 

Links:



Ways to Tune In:



Thanks to our sponsors:

 

Podcast Transcript: AI Strategy, Data Foundations, and the Winner-Takes-All Race

Transcript introduction

This transcript captures a conversation between Seth Earley and Glenn Gow about what CEOs most commonly misunderstand about AI, how companies can build a durable AI advantage through data strategy and focused initiatives, and why the difference between generative and predictive AI matters far more than most executives realize. Glenn draws on 25 years of CEO experience and his work coaching senior executives to translate AI's strategic implications into practical action.

Transcript

Seth Earley: Terrific. So good morning, good afternoon, good evening! Welcome to today's podcast. I'm super excited about our guest today. He is known as the AI guy and the CEO therapist. I want to hear more about that. His direct experience and background as a CEO makes him uniquely qualified to help CEOs grow their company's valuation. He's an advisor of the Technology Adventures program at Stanford University, a board member at Cetera Intelligence, and a CEO coach at the Peak Performance CEO Coach program. Glenn Gow, welcome to the program.

Glenn Gow: It's great to be here. Thank you for having me.

Seth Earley: I'm really excited. Thank you so much for your time. So I wanted to start out - what are the things that you're seeing in the marketplace that are typical misconceptions or myths? What are CEOs kind of walking around with that you see really needs some dispelling? What kind of issues are they operating under that may not be accurate today?

Glenn Gow: Well, there are quite a few actually that I see. One is that when CEOs enter the world of AI, they start thinking of it as ChatGPT. The first thing I point out is - well, ChatGPT is just a tool. It represents one area, and it only represents an area of generative AI. What may be even more valuable to CEOs is what I call predictive AI, or analytical AI - the more historical world of AI. And what I'm excited about is that at least ChatGPT has brought to the forefront the power of AI. So what I think it's important for a CEO to do is say, help me understand both sides of the equation. One side is generative AI, and the other is all the rest of AI that people have been working on for the last 50 or more years. It opens a door for CEOs to now recognize how this can impact their company and their industry. That's a bit of a myth - that ChatGPT is just how they think about AI alone.

But the other thing it's doing is helping CEOs recognize something that is a theme that, Seth, I'm sure you've said to your own clients - this is happening faster than we expected. The curves are starting to look exponential. For a CEO, what that means is: I really need to pay attention to this, because I don't yet understand the implications in my industry. And there's a concept in AI that I call the winner-takes-all concept - if you become better at AI than your direct competitor, and you continually collect more data about your customers, and you become smarter about those customers, and the AI can learn every single day, it creates a flywheel effect where you have the opportunity to dominate a market. If your competitors aren't on that same flywheel, and if you're left behind, you could reach a point where you're never going to catch up. It's that transformative.

Seth Earley: Yeah, it's a great point. I sometimes tell the story of a customer many years ago - this is related to AI, but it's also foundational to information architecture, which I focus on a lot. What was happening is they were using a traditional method for building K-12 textbooks. There's a lot of variation depending upon state and local requirements and grade, and they found that their competitors were getting to market 6 months faster. It turned out competitors were using componentized content and machine learning to auto-categorize and componentize their content - they were able to draft a textbook 6 months faster. The company had 1 million objects in the repository. The CEO said, well, why aren't we doing that? The staff said, that's the project we've been trying to get off the ground for the last 3 years. He said, great, let's do it now. How long will it take? Well, it's going to take at least a year or 2. And the competition kept moving ahead. They lost that market. It was too late for them. And I think we're going to see the same thing - organizations that just are not going to adapt fast enough are going to lose their market positioning.

Glenn Gow: I'm so interested that you told that story, because an adjacent industry to what you just described is Chegg. Chegg helps students get the information they need to pass their tests more effectively. Their stock dropped 45% in a single day because the CEO said ChatGPT is where all our students are going now - they're being disintermediated. And 45% in a single day is pretty stunning. That company is in trouble because of this new technology that's just shown up.

Seth Earley: Right. And they're going to have to reinvent themselves in some other way. ChatGPT is transforming lots of industries - it's getting all the attention. But there's still a big gap between using ChatGPT for publicly available information and using it internally in the organization. You have to do things behind your firewall. You can't let stuff out into the wild and have it become public. I know Samsung made that mistake. Can you speak to that?

Glenn Gow: There are a few things happening. The big cloud providers - Azure, AWS, GCP, and even IBM and Oracle to some extent - are saying come to us because we're going to create a walled garden for you. We're going to enable a large language model to run, but we're going to protect your data. We're going to let you put your corporate data in this walled garden and augment it with whatever large language model we choose - and by the way, the choices are rapidly expanding - so that you can build your own. Related to that is what's happening with open source. ChatGPT-4, I believe, is not open source, whereas prior versions were. But they're not the only party in town. Because there's open source, there's a tremendous movement to create many variations of large language models. Now it's going to become easier for enterprises to say, I want to take that open source, put it in an AWS framework, load our data in it, and now I'm going to have access and the capability to do all this work inside my organization that I couldn't do before. Bloomberg is an example - they announced they have unique financial data that ChatGPT doesn't have, and they're bringing out their own large language model.

Seth Earley: And the idea is that there will be lots of different models to choose from, not just one size fits all. There's LaMDA, there's BERT, there are a lot of other language models. And as you mentioned, there are specialized language models for particular industries - specialized life sciences models, Med-GPT, Chem-GPT. There's a whole bunch of them for the industries themselves.

Glenn Gow: Exactly. I see this explosion of opportunities for anybody who is looking to take advantage of this and to protect their data. This is very early - we haven't quite seen anybody fully implement this yet. But the implementations are happening, which means the opportunities are massive. Going back to your first question - I'm concerned that CEOs don't yet understand the implications of the speed in this area, and what's going to be required of them to change very, very rapidly so that they are not disintermediated as a company.

Seth Earley: Right. When you think about standardization versus differentiation - standardization gives you efficiency, but differentiation gives you a competitive advantage. So there still has to be a mechanism to control your knowledge and have your own versions of ontologies and language models that are fine-tuned to your organization. We're working with a life sciences firm, and the science model they're using is too big and too bulky. It has terminology that's not related to them. They don't deal with animal diseases - animal diseases are part of this large language model - so you get a lot of terminology that's not relevant. That has to be trimmed down. And then there's specialized language for their drugs, their treatments, mechanisms of action - things that are non-public. Maybe you can talk a little bit about how that differentiation versus efficiency might play out in organizations.

Glenn Gow: I've seen that from a different perspective. Let me share what I've seen, and it's in the area of prompting. The way I describe prompting is: you take a large language model and you shrink it to the area of focus you want the model to focus upon. In the medical technology example, I could literally prompt it to ignore information related to animals and specifically have it focus on cardiology or radiology, or something of that nature. One of the nice things about large language models is that even if you're not good at prompting, you can ask it how you should prompt it - and it will tell you. The second thing I see happening is that this whole prompting area is evolving rapidly. In Stable Diffusion, for example, it can take a simple prompt and suggest a better one - what they call a "prettified" prompt. It suggests many additional ways to be more specific. And the result I received from using that approach was stunning. I looked at this sneaker it created with an amazing logo on it, and I thought - if I'm a designer, I'm going to be able to create fantastic designs with the help of this model. That's where we really want to go with productivity: how do we take anything we're doing and become better by using AI as our coach, our recommender, our intern.

Seth Earley: That's a great point. You can actually ask it to help you with the prompt, and I also find it's a great tool for beginning that research process, building outlines, finding research topics. So what other areas of AI are you seeing get some uptake in the enterprise?

Glenn Gow: I'll tell you a true story. The phrase I use is AI can see patterns that humans can't see - and we want to take advantage of that. Frito-Lay is an example. They're always testing the market with different potato chips, but it's very expensive to say, will this particular flavor work in this neighborhood or location? What they did was layer third-party public data on top of their own data about who's buying what chips at what locations - consumer age, economics, preferences, all available by zip code - and ask, what's likely to happen in this location? Frito-Lay discovered that in Frisco, Texas, there was a large number of ethnic Indians, and it turned out those customers were buying a product that wasn't even available in the US - curry-flavored Cheetos, which was only being sold in India. Without humans having to figure this out, they identified the need to put curry-flavored Cheetos on the shelves there. The AI found a micro market of 200,000 people. It wasn't 200 million people. But Frito-Lay now dominates those shelves, and their competitors don't even understand what's going on. The question I want to ask any CEO is: how many micro markets are out there that you can serve and dominate where your competitor doesn't even know what's happening - by using AI to find patterns that humans can't see?

Seth Earley: I want to step back and understand a little bit more about you and your background. How did you get into this space?

Glenn Gow: I was fortunate enough to be a CEO for 25 years, and most of my work was helping large technology companies with strategy. Apple, Google, Facebook, Microsoft, Oracle - all the big players were clients. Then I got recruited into venture capital, and two things happened. One, I started coaching CEOs and fell in love with that work - that's now my full-time role. But the other thing that happens in venture capital is you can predict the future, because people walk in the door every day telling you what the future is going to look like. When you start to watch the patterns - what I call waves of technology - you start to see them all happening at the same time. And what I noticed was this tsunami on the horizon called AI. When you start listening to very intelligent, cutting-edge people talking about what they're going to do with AI, you want to make sense of it. I'm not a deep technologist - you know a lot more about technology than I do, Seth. My expertise is at the board and senior executive level, translating what's happening in AI so that senior executives and board members make the right decisions. Now I coach CEOs who run AI companies, I'm on the board of an AI company, and I write a column in Forbes on AI.

Seth Earley: What do you see as the biggest stumbling blocks? My sense is there's a lot of assumptions being made - what vendors call "assume a knowledge base." Many organizations don't have a good handle on that. Their knowledge is fragmented, distributed, not well organized. There's a prevailing notion that you can simply point AI technology to bad or incomplete data and it's somehow going to produce a result. We're still doing basic blocking and tackling for a lot of enterprises - getting their knowledge house in order - and then building POCs that leverage large language models to access that content. But there are a lot of assumptions about that content being ready. Do you run into that? And when it comes to obstacles, do you see it as more cultural or technical?

Glenn Gow: The vast majority of organizations have large to significant issues with their data. More often than not, data systems are not designed to be treated as a corporate-wide asset. They're usually functional. We know a lot about our customers in the sales organization, or in client success, or in support. Marketing did some research over here. Engineering talks to them occasionally. But none of that is brought together in a way designed from the corporate level. So what I say to CEOs is: if you are interested in doing significant work in AI, the long pole in the tent is getting your data together - and that can take a year or more before you feel like you have a true asset. Before you go playing more with ChatGPT this evening, focus on what your organization looks like. What is the team that's managing your data, and how do you treat that as a deep corporate asset?

Seth Earley: That's a really great point. There are so many organizations chasing the Holy Grail of the 360-degree view of the customer, and yet there may be 50 or 100 applications that touch that customer, and all of those representations of the customer are different across those different systems. Pulling those together is extremely challenging. There are organizational silos, political silos, responsibility silos, governance silos, funding silos. It takes that C-level view to say: we need to stop the turf wars and bring people together to look for the common good.

Glenn Gow: So I know of a Fortune 500 company that said: we want to do the most effective marketing campaign, which is marketing to our current customers to enable them to upgrade what they've purchased from us. The problem is, we don't know who our customers are. Nobody drove the desire to ensure we know who our customers are. They had lots of data points about customers, but they didn't know how to differentiate between the user, the buyer, the purchasing agent, and the economic decision maker. And this company didn't know any of that.

Seth Earley: We just went through that process with a software subscription company. They're trying to optimize their customer lifecycle and experience, and because they didn't have the correct customer data model, they weren't describing customers in terms of those attributes. They weren't able to identify what stage customers were in. They weren't able to impact conversions because they didn't know how to target. Who's in the middle of the funnel? What descriptors can we identify to give them the right content to move them to the next stage? How well is our program working? Do we have the metrics to determine what content is most effective? Personalization requires those customer attribute models, product data models, and content models to all be aligned so we can do multivariate testing. That was a huge problem - many organizations don't even know how to describe the problem. What are you advising organizations when it comes to beginning this process and getting their data house in order?

Glenn Gow: I think we start with the CEO. Let's pretend I'm in charge at a big company, and I've got to wrangle data. I may not own all the data - it might be sitting in different functions, different divisions, different geographies. Unless I have the CEO saying to the organization: this is one of the cornerstones of success for this company, and we as an organization as a whole need to bring together mechanisms to create assets out of that data - unless I have that coverage from the top, it may never happen. And the same thing is true of AI. It isn't AI-first, it's strategy-first. Here's a strategic opportunity for us to take a dominant market lead. And oh, by the way, AI is going to be one way we do that - and the long pole in the tent is data, at least on the predictive analytics side. When the CEO deems this critical, that's when the organization begins to shift. One of the biggest obstacles to adopting AI in a company is culture. It's a cultural question: who are we as a company? Are we going to be fiefdoms saying, no, this is my data, you can't have it? Or are we going to be a company that says, if we can leverage all the rich data we have, AI can give us incredible insights into what we're going to be doing going forward - what suppliers we work with, what products we build, what consumers we go after?

Seth Earley: That's so true. The vision does have to come from the top. We have to have leadership that not only understands this but is ready to allocate the resources and organizational time. And leaders have been burned before - on big master data initiatives, big digital transformations. I like to say a digital transformation is a data transformation. But many of those top-down initiatives fail. Large consulting firms don't necessarily have the incentive to solve the problem at the root because these are annuities for them - evergreen problems. What are your thoughts on getting past that? Maintaining organizational traction, retaining funding, ensuring return on investment?

Glenn Gow: Let me speak to the AI-specific part. The best place to start is strategy. Let's say the strategy is: we're a B2B company selling to the enterprise and we want to expand into the mid-market. That's the strategy. Now the question is: how can AI help us do that? Notice I'm not saying we have to clean up all the data in the company in the next year. I'm saying we're choosing one initiative - move to the mid-market. AI is a tool to help us get there. I've shrunk the focus down to just the right size. There's a sizing question: what are we going to invest in time, energy, training, and change management? That's what I would call a bite-sized initiative. Next, we want to narrow the scope, establish baselines, experiment, and see what we can impact in terms of metrics. We're taking a bite of the elephant rather than tackling the whole thing. Then comes build or buy. In the world of AI, most companies - even Fortune 500 companies - can't afford to hire all the AI people they need. There are very few of them, and you're competing with Google, Amazon, Apple, Meta, and Microsoft for those people. What I recommend is building a center of excellence for AI - talent who can run that center - and then relying on two parties: external consultants with industry experience who can help with the data problem, and vendors who are incorporating AI into their products.

Seth Earley: That's a really great point. It really is dependent on where they're trying to create value or competitive advantage. It's less about building the foundational elements of deep learning models and more about applying them to the problem. Many organizations have challenges going from pilot or POC to production - not because of the AI model itself, but because of their environment, data, and processes. If they get that right, it's much easier to leverage a vendor offering who has already done the heavy lifting. I totally agree. It's really about understanding your business process, where you want to have that intervention, and then finding the tools and technologies and vendor applications that will help you with that precise intervention. You can't automate a mess. You can't automate what you don't understand.

Glenn Gow: There's another element - change management. In the example of moving from enterprise to mid-market, we have to be very careful to choose something we can succeed at. We're actually rolling out a strategy, and it's important to be able to show some success with what we're doing with AI, because the rest of the organization is watching. If this fails in the context that we can't get AI to work correctly, that's going to slow down everything else when we try to roll it out elsewhere. Two important parts: one, it has to have support from the most senior management - that's how change management gets rolled out. And second, we need to manage expectations. If we enter into this effort and discover we have a terrible mess in areas related to our data that will take a long time to clean up, we need to be very clear about that. It has nothing to do with AI in this case - it has to do with how we've managed data. There's no AI without IA - no artificial intelligence without information architecture. As a CEO, if I hear that it's going to take a year longer than I thought, maybe I put that on the back burner, lower its profile, call it a long-term project, and go find another place where we can find a near-term success.

Seth Earley: Quick wins are very important. Whatever you're doing when building a transformation roadmap, we frequently create roadmaps with organizations with long-term initiatives but also clear quarterly wins along the way - because you can't promise something 3 years down the road without showing progress. I also like to talk about the "excuse case" - the use case that's really the excuse to do something bigger. We did something very similar to what you're describing with Silicon Labs - a content and knowledge management initiative that started with a very narrow focus on their training content. It was a huge success, and then the rest of the organization said, wow, what did you do over there? Can I have some of that? And now it's spreading from one part of the organization to the next. There can be a top-down mandate, but then a bottom-up execution. For large global enterprises, that's often the process: get those wins at the functional level, department level, business unit level - and then rinse and repeat. You eat the elephant one bite at a time. You show that you're able to do this, it builds on itself, and that creates that flywheel and that virtuous cycle of success.

Seth Earley: I know we're coming up to the end of our time, and I wanted to give you an opportunity to finish with your final thoughts. If you could go back in time - you just graduated from college or graduate school - what kind of advice would you give yourself back then?

Glenn Gow: I would say: be more innovative. Don't be afraid of making mistakes, and grab on to the changes you see happening. One of my favorite quotes is from Neal Stephenson, a science fiction author. He said, the future is here - it's just unevenly distributed. What I think about when I hear that quote is: I can lean in to learning the latest things that are changing the world, or I can wait to see what happens. Right now, at this moment in time, we're seeing some massive change that we don't fully understand the implications of. It's like 1994 when this little company called Netscape popped up on the radar screen and said, we're taking the Internet from just a communication mechanism in academia and putting a consumer layer on top of it. And then the world exploded with everything that we live in today. This moment - or the moment in November when ChatGPT came out - is a similar moment. What's important is that we've now put a consumer layer on AI, and now every single person on the planet can be impacted in a positive way. There will be negative impacts too, because this is just a tool, and a tool can be used for good or bad. I want to tell you a story - one of my CEOs sold his company a little while ago, and the other day he was advising the company he sold on technical issues and said, hey, would you like me to write an algorithm to solve that problem? He got a note back from a junior engineer that said: yeah, no, we don't need you to write an algorithm - I asked ChatGPT and it did it for me. And my CEO was devastated. He said, I've always been known as the person who could solve problems no one else could solve. Suddenly I'm not that person anymore. I'm encouraging everybody - whether you're a CEO, a board member, or a first-line worker - to lean in to what's happening now. Become good at all the tools that are being made available to us, because that's going to create opportunity for you. People who don't lean in are going to be losing a lot of opportunity, maybe tasks in their job, maybe satisfaction in their jobs, and for some of them - their jobs themselves.

Seth Earley: Absolutely. Well, I really want to thank you. We've been with Glenn Gow, who is known as the AI Guy. Glenn, where can people find you?

Glenn Gow: It's easy. You just need to know how to spell my name, and that's the name of my website. It's Glenn, G-L-E-N-N, G-O-W, dot com.

Seth Earley: Thank you so much. That's been wonderful. Thank you again. Thank you to our audience. This has been another episode of the Earley AI Podcast, and we will see you all next time. Thank you again.
Meet the Author
Earley Information Science Team

We're passionate about managing data, content, and organizational knowledge. For 25 years, we've supported business outcomes by making information findable, usable, and valuable.