George Jagodzinski (00:00):

Today we learn how the speed of human decision making, especially board level decisions, can be a huge obstacle to AI adoption at scale and that humans are not in fact products. I'm joined by Lynda Pak. Lynda brings a fantastic perspective from a career span in consulting at PwC, Deloitte, to VP of commerce marketing at Bed Bath and Beyond, to serving as the global CIO at Estee Lauder Companies. We discuss how to keep humans in focus when everything around them's changing. And from there, we get into why organizations moved fast on AI pilots, but then they hit a wall the moment they need the rest of the company to make decisions. We also talk about how you need to put up another duck when you shoot a duck. I know it sounds strange, but you have to listen to find out about that one. Please welcome Lynda.

(00:43):

Welcome to Evolving Industry, a no BS podcast about business leaders who are successfully weaving technology into their company's DNA to forge a better path forward. If you're looking to actually move the ball forward rather than spinning around in a tornado of buzzwords, you're in the right place. I'm your host, George Jagodzinski.

(01:21):

Lynda, thanks so much for being here.

Lynda Pak (01:22):

My pleasure. I'm really excited about this conversation.

George Jagodzinski (01:26):

Me too. Something that you mentioned last time that really stuck with me is just how do you keep the humans in focus? How do you think about the people? And especially the fact that you've played these global CIO roles, technical roles, I'm curious if you walk me through, is there a philosophy that you hold or an approach that you take on how you keep them front and center when you're talking about massive technology investments?

Lynda Pak (01:52):

Yes. Well, there's several facets of humanity and AI that I look at all the time, I consider all the time. One is when you're working for a company and then I'll delve into a little bit, and then there is the accessible side of AI, especially in markets or in communities that don't even have internet and then to compound it by, well, can you access AI and further disparity of capabilities?

(02:32):

And so initially when I look at the industry where I've focused, consumer goods, consumer products and retail, for companies with whom I've worked and have been employed by, when you consider AI, there has to be very thoughtful decisions made around what truly are we using AI for? Is it to be in the service of humans of employees or is it mainly a cost-cutting type of activity? And it could be a mix of many different things. When you decide to make these changes or any changes, you need a very thorough and thoughtful, again, I use the word thoughtful, change management plan. And you're not just talking about an email or a town hall. You have to really think about the organization, how are roles changing.

(03:38):

I mean, it's such a rich topic. I feel like we can talk about this topic in its entirety, but I think you have to consider there's a lot of fear around AI and jobs and the type of narrative that you want to convey and communicate around why AI? Is it just a shiny tool, we're just going to give it a try, or is there really a reason, a business reason as to why we need to take some of these steps?

(04:09):

I've spoken to other peers in tech and a lot fall under, well, it's an automation, it's an efficiency play and there's some cost-cutting plays. Maybe we're dabbling with some growth. And some of the growth side is so exciting when you start to think about how are we going to evolve our employees, our humans within our organizations? That has to be very thoughtfully laid out as well. And I think it's the methodically thinking about how you're going to roll out AI or activate AI, really make it a true collaborative capability with your employees.

George Jagodzinski (04:54):

It's hard for people to collaborate when they're in fear, right? There is a lot that we can talk about. Let me just pull on the one, you talked about roles, because what I find is if people don't know what their new role is going to be, then there's a lot of fear. And then combine that with the fact that within different cultures, a lot of people, they have something set in their mind that certain ladders in certain roles mean success. And you've dealt with these global organizations where across the different cultures that must be so different. So maybe just expand a little bit on how you look at roles.

Lynda Pak (05:25):

Those roles, at least those traditional roles, they've evolved, but there's still roles where there have been career paths laid out. And it's taken a bit of time to allow those career paths to evolve and to be able to communicate to employees saying, "Hey, this is what I'm expecting to see from you, and these are the things that you are going to be measured by. These are the goals that you need to establish in order to demonstrate that you are evolving and getting to that next level."

(05:58):

With AI, there is this, one, there's an existential uncertainty around, well, what is AI going to do? Because it is changing so fast. What AI was able to do or what companies were able to do six months ago around generative AI is different from where agentic is taking companies as well. Again, that existential uncertainty plus the lack of true, "This is my growth path." You can't just sit there and go, "Well, I'm just going to iterate as we go along." I mean, humans are not products.

George Jagodzinski (06:38):

What? That's Crazy, Lynda.

Lynda Pak (06:42):

I know, right? We're evolving. But I think humans really like to have that sense of safety. A former boss of mine, and I love this concept that he introduced to the company, is the culture of joy. And joy is not just about your psychological safety, you have a steady paycheck. There's this confidence that you know where you're going and you know what you need to do. There's a path ahead of you. And a lot of employees are fearful of, well, if that is constantly changing, what does that mean for me? Employees who have spent decades establishing themselves as experts in a certain area, certain domain or broader domains and now it's upended. All of the new job descriptions, they have to be pressure tested as well to allow for the growth and yet they're still evolving at the same time.

(07:43):

And so I think there is some fear around that certainly. And if you don't employ the transparency and care for your employees where it's just things just come out in an email or in a town hall and there aren't feedback loops or open conversation loops, that's when it starts to be more chaotic. And as you said, rightfully so, to collaborate with a machine essentially is going to be harder when you're in fear. Now you say, "I've got to collaborate with something that is not, it's a different type of colleague."

George Jagodzinski (08:27):

Yeah, and people also have different thresholds of change that they want. I feel like some of the existential fear right now I think comes from this idea, which I don't think is true, that every single company out there has to change or is changing very, very fast and it's going to be in a constant state of change. At smaller companies, probably, not all of them, but at larger companies, I would imagine there's going to be pockets that can't change that fast and pockets that do. And this has happened before AI also. I'm curious, maybe some stories or perspective you've had on that, the different pockets of change.

Lynda Pak (09:05):

If we go back to ERP, which was over 20 years ago, that was a massive change to enterprises and within the tech industry as well. When ERP first came on, it was like, "Well, what do we do with this?" And sometimes you got to a place where there were all of these offerings and you were basically buying this massive Cadillac and all you were using was the fender. And so you had to figure out a way to best integrate or incorporate those types of platforms into your corporate ecosystem, your enterprise ecosystem.

(09:47):

And so there was a massive amount of change required. Now the organization needed to change in terms of the roles that they started to. Things started to get automated. Let's not forget that things got automated and processes started to get stitched end to end and upskilling needed, upskilling was required. It was a brand new technology and all of the complexities and integration processes had to get re-engineered.

(10:16):

Now I feel like we're going through that all over again, but things are also moving a lot faster than it did when ERP came into the picture. And you'd like to think that, well, many of us have gone through that process before so we could be ready for some of it, but there's a lot that we're not ready for. There were already experts in place like the SAPs and the Oracles, the giants, they're the ones who actually built these systems and continue to work with enterprises to further evolve them, but AI just isn't happening that way. It's happening on a more granular scale, individuals as well as these large partners. And then there's the physical AI component to it as well as of course agentic and all the flavors of AI that the normal person gets introduced or hears the buzzwords every single day.

(11:22):

When you think about the magnitude of change, it was pretty big back in the ERP days. But again, that was really focused around enterprises and companies. It was not as broadly disseminated or distributed or accessible by everyone else. I mean, if you're vibe coding, now anyone, you don't need to be certified in SAP. You can just start vibe coding and you can figure that out. You can watch a YouTube video to learn how to vibe code if you need to. And so the scale is much bigger.

(12:03):

And so I can't say that I know exactly how this is going to turn out or that we all know how to manage it. I think we have to use some of the best practices that we had around thinking, around problem solving, yet at the same time you need to be able to turn everything that you've learned upside down too.

George Jagodzinski (12:28):

Yeah, that makes a lot of sense. The ERP example's a funny one because having gone through them myself quite a bit, it's like, okay, we're going to go through a bunch of change. It's going to take three years. There's going to be an army of people over in this side of the building. And most of the employees are just thinking to themselves, "Eh, this will probably never even be done anyway, so I don't need to worry about it too much." So they don't feel too threatened.

(12:53):

And then even when it does, it's not revolutionary to the organization. It's like slightly better of what they were doing before in the grand scheme of things from what they were doing prior to ERP implementation, where now it's like, "Hey, everything's going to change maybe next month and I don't know how you're going to fit into it and good luck, everybody." And it's very different in how you manage that.

Lynda Pak (13:15):

Absolutely. Absolutely. And anyone who says that they know exactly what's around the corner, I mean, you're looking around many, many, many corners. Things are changing every week.

George Jagodzinski (13:27):

Crap, that was my next question, Lynda. I wanted to know what's around the corner.

Lynda Pak (13:31):

Well, I think that you have to be ready for anything. I mean, you have to be in the mode of anticipating that something new may be around the corner. And it's how you synthesize this information, because remember the old term storming and norming?

George Jagodzinski (13:50):

Oh, yeah.

Lynda Pak (13:51):

Do you remember that term? I think there's a lot of storming still happening. And I think there's going to be a lot of storming that is going to continue to happen, but there is also some norming happening too.

George Jagodzinski (14:05):

Yes. Storming, forming, norming.

Lynda Pak (14:06):

Token economics, you're starting to... It sits between the two, right? The token economics is something that FinOps will need to get their arms around. Not just FinOps, but leaders will need to get their arms around, but it's not something that you can really pin down at the moment. But you know that that's something that you're going to have to tackle.

George Jagodzinski (14:29):

Curious, with your experience, I'm sure you've had to deal with, I'm assuming, similar things with just license management and just all these other global scale things at your organizations, what's the same and what's different from the governance of these things? Because the prices are very volatile, people have different thresholds of what they could do. You hear these horror stories of people just all of a sudden spending a quarter of a million dollars in tokens in one day by accident. As an organization, what can we learn from the past to put in place better governance and measurement of that?

Lynda Pak (15:02):

I think an approach from a governance perspective that one can take or a company can take is I think a lot of tokens are expended due to duplicate requests. You and I ask the same question and 500 other people, you're expending tokens, as opposed to a more structured way of, okay, let's make sure that if there are similarities, let's make sure that there's a layer, that metadata layer that pulls that together and starts executing your input prompts most efficiently as well so that the output prompts can also be distributed efficiently and effectively.

(15:59):

And I think it may be a, hey, no duh reaction to, well, if you're basically paying for the same thing 1,000 times or 500 times, of course you're going to spend it faster. If you have a budget, you're going to spend it faster. Whereas if you're, again, thoughtful about how it's going to get spent, then you get a little bit more out of what you've been allocated, because there's constraints around resources and TPU, GPUs out there right now. I mean, companies are clamoring for it.

(16:34):

From a token economics perspective and from a governance perspective, I think that that is an area that needs few more guardrails to manage the financial aspects of using AI.

George Jagodzinski (16:52):

Yeah, it's interesting. It's almost like the best practices of good architecture and good performance are going to directly tie into token economics. It reminds me of the early days when rapid development platforms came out and data access platforms and someone would feel really happy that they built the CEO some nice, sexy new report. And you look behind the scenes and it's doing a database call for every single row of data. You're like, "What are you doing?"

Lynda Pak (17:20):

You're spot on. I mean, again, going back to some more traditional tech lingo is, remember before middleware came into the picture, you were building point to point and they were making the same calls? It's highly inefficient and very expensive. The overhead is very expensive.

(17:40):

And consumption models are not new. I mean, cloud, you pay based on consumption, but it's also predictable. You know what your seasonality trends look like. You know what your volume ranges are going to be for sales, thereby you know what the backend processing requirements are going to be. But that is not the case with AI, with agentic, because you have agents collaborating with other agents when it makes sense.

(18:13):

And so those are areas that are going to be very difficult to get our arms around, but it's an imperative that your FinOps process determines how to at least measure what that looks like so that you can measure your cost better. I called it the prompt architecture, the metadata of prompts for lack of better term. I think you have to think about that kind of architecture too, not just API calls here and there, but true architecting. How are even AI prompts, maybe there's a prompt operator or model and it's all going into how you're going to teach a model. So I think that's going to be very tightly integrated.

George Jagodzinski (19:00):

Now on the organization front, we poked at this a little bit, but how fast can organizations truly change and what's limiting them? You've been in some very large organizations. I'm curious what you think.

Lynda Pak (19:12):

The speed of adoption or activation is going to be as fast as a company can make decisions. Let's take pilots for a moment, AI pilots. You can make a very quick decision to have a subset of data that you have full control around, you have full visibility, the volume is manageable, you run a pilot, you're teaching your model and it's giving you the results that you need and it looks like things are looking really great and look how fast we're able to build this pilot in a matter of weeks. And then now you start to bring the broader organization into the fold because you start to need process experts, domain experts, you need to have access across multiple systems and there are business owners of those systems as well as technology owners of those systems. And now there's budget that you need to start to prioritize.

(20:18):

If those decisions in terms of what the priority is for the organization and which areas it will have the greatest impact through AI, if some of those decisions aren't already made, everyone knows how long their own organization takes to make those types of decisions, bring all the leaders on the same page. I mean, in my previous world, it was, I needed to have one-on-ones to make sure that I convinced the individual leader, and then I'd have to bring the individual leaders for the particular brand regional function, and then a broader consensus, yes, we're all thumbs up and then it goes up to the executive team. And that could take weeks and weeks to do.

(21:04):

So it is as fast as the ability of an organization to make decisions and commit to those decisions. It's one thing to make the decision at the executive level, but then those at the -1s and then -2s, et cetera, if they're not on board, then it takes another round of convincing and influencing and getting on the same page in order to do that.

(21:33):

Now that's the culture of some companies where you need to have alignment and it's best to have alignment versus, all right, this is the way we're going and this is a decision. It's arbitrarily made by one or two individuals and then that's it. That can happen too. And so in that case, decisions can be made very quickly. But it goes back to, all right, now we're activating AI and we don't have the rest of the organization on board.

George Jagodzinski (22:00):

I know a lot of organizations that I've been in, it's like getting a bill passed through Congress. It's brutal, and something that should take one week to make a decision on takes now six months to actually make a decision. And that's the difference between someone vibe coding within a department somewhere versus true organizational change.

(22:19):

I'm curious if you can anonymously share an example of an actual decision that maybe should have taken very short amount of time, but took an extremely long amount of time.

Lynda Pak (22:29):

Well, it's not specific to AI. It's around the just decision making, which seemed at the time very straightforward. We're looking at replacing a point of sale platform in one of our markets. The current system has probably been in place for, I don't know, at this point, close to 30 years and it went down a lot. And when it went down, you've got opportunity lost there.

George Jagodzinski (22:58):

You can see the lost revenue just cranking up, right?

Lynda Pak (23:00):

Exactly. There's opportunity loss there, there's frustration, there's a store associate frustration. What we presented was here is the new, and it was so obvious. And yet the decision could not be made because there was alignment and support from one of the regional leaders, but they weren't paying for it because it was coming out of a technology budget. There was contention. Of course, there are constraints on your budgets. It needed several rounds of convincing and we never got it off the ground, because new leaders came in, they wanted to change. When you have new leaders come in, they want to look at other options that are available in the marketplace. It became a prolonged process when it should have been something that, I mean we quantified everything, it should have been a very quick decision and it was not.

George Jagodzinski (24:04):

Oh, those are brutal. Yeah, and a new leader comes in, they say, "Well, I used so-and-so at my last company. Why don't we do that?"

Lynda Pak (24:10):

Exactly.

George Jagodzinski (24:11):

"We just figured this out. We don't need to figure it out again."

Lynda Pak (24:14):

And between calendars and pushback, I mean, they take a very long time.

George Jagodzinski (24:21):

Well, I think that's a perfect example because it's not AI, because that same thing is going to play out with AI. And it ties back to your economics of tokens.

Lynda Pak (24:29):

There you go.

George Jagodzinski (24:30):

It's like, who's paying for these tokens. What's the value?

Lynda Pak (24:32):

And there's so many more players in the AI space and the different offerings. I mean, point of sale is pretty straightforward, right? It is a process that has been around for decades. And so you would think, here's a need, here's the new platform, here's the new stuff that they're going to offer. It should be pretty simple. And then you look at AI where it's like, "Well, we don't know what we need." And there are 50 different companies that are offering a sliver with some overlap, a little bit of overlap, and so it becomes this very really convoluted process.

(25:15):

And then you have to go through, well, do we build versus buy? And that's another thing, platforms that were a massive market player who had a lot of market share two plus years ago and now they've lost that market share to others. And so instant change very quickly.

George Jagodzinski (25:34):

And as an organization, you probably took a while to make a gamble on that.

Lynda Pak (25:37):

To make a gamble, make an enterprise decision and then to switch to a new model, no one really talks about that, that is a significant cost there as well. Think about the tentacles that you have embedded in your systems and within your organization. To unravel that is, I think, the next phase that companies are going to have to start looking at more seriously, because enough companies have already started to integrate some of these models in their ecosystem and then to think, "Oh, well, things are changing and so now we're going to move to this new or add." There's going to have to be a lot of analysis or you can have AI analyze where all the work we'll probably need will require.

George Jagodzinski (26:30):

That's actually a great call back to a prior guest, Steve Wunker, on the podcast. He wrote a book called The Octopus Organization, which really is great, talking about all the tentacles in the organization. I mean, those are great examples of the true deep challenges behind organizations, why change can be slower than everyone thinks. I think there's still going to be quick change happening out there in pockets and in certain places and for the right organizations that they can align the budgets the right way, they can measure value the right way, they can make decisions the right way. I think that they'll fit in, right?

Lynda Pak (27:00):

And talent, get your data prepared. I mean, all of those things, right? Your infrastructure, your foundational aspects, right? Your infrastructure, you have to have your governance in place, you got to have your talent in place, you got to have your data in place and that is massive. That is massive.

(27:18):

I mean, you can try to do things in smaller pockets and many companies have as opposed to something like a big bang happening, but you still have to get some of the lessons learned while you're building that. And I'm not going to say pilot, while you are scaling it in production, right? Once you've got a successful activation in production and you're starting to see some of the results that you're expecting, that's when you have to methodically look at, all right, how do we do this a couple of more times where we start to feel a bit more confident that then we can scale more broadly within the entire organization. Because now we have in-house experts, because initially a lot of companies are going to have to buy some of that talent or borrow some of that talent, pay for the talent while you're building that in-house expertise as well.

(28:18):

And that's going back to the human side is really a commitment to invest in the talent that you have to say we want to have in-house expertise versus that you have consultants knocking on your door constantly who have that expertise and you think, "I'm going to get it quickly." But then you have to go through the process of having them understand your culture, your environment, et cetera. And so I'm not yet convinced that there is a lot of time saved compared to building some of those capabilities in-house. And I know from a governance standpoint, a lot of companies are deciding we're going to build that talent internally or we're going to figure out what that right balance is.

George Jagodzinski (29:08):

Yeah. I think there's a lot of value, selfishly speaking as a consultant, in getting help to build it in-house, get in some outside coaches to say, "Here's how we're going to build it. Here's what good looks like. Let's build this bionics within the organization."

Lynda Pak (29:26):

Yeah. I think the best collaboration that you typically see with consultants, and I was a consultant for almost 20 years, I think the best collaborations have been when I've been on longer term implementation projects. And you build that knowledge of the organization's processes, the culture, the organization itself, while you are also doing that knowledge transition to your workforce. I think that's always been the best collaboration because then you're leveraging your consultants not as operations people, not your operations workforce, but they're the ones who are actually consulting and advising and getting you past that initial skills gap.

(30:22):

And then once it's there, then they can help in a different way. More strategic decisions of now that we've put the foundation in place, how do we now start enabling growth initiatives? Now that we've put some of the efficiencies in place, how do we start thinking about... Because that's the way consultant brains and consultants are trained to see beyond the operational side of things.

George Jagodzinski (30:51):

Yeah. I always think good consultants, it should be hip to hip and you should be working yourself out of a job. And having that mentality of working yourself out of a job actually just ends up getting you other jobs, which was fun.

(31:02):

But Lynda, you're truly insightful and very pragmatic. I love it. I like to finish with a fun question, which is in your life and in your career, what's the best advice you've ever received?

Lynda Pak (31:13):

I use it to this day, but the way it was said to me, it may be a little generational translation required. So if you shoot the duck, put up another duck. What that conjures up for me, and this was a senior partner when I was growing up in consulting, he said to me, "If you're going to put down an idea or put down something, you got to put something else up that is even better." And the reason why there may be a little bit of a gap there, a generational gap is because it makes me think about carnivals. Remember the duck that you had to shoot?

George Jagodzinski (31:55):

Oh yeah.

Lynda Pak (31:55):

Yeah? And then when you shoot that one, another one pops up. And he was kind of trying to explain it in that way in terms that he was familiar with. And of course, I got the reference at the time. To this day, even with my kids, my friends, I always say, if you're going to put down the idea, put up another idea that's better. That's where you show value. Otherwise, you're just kind of sitting back going, "Wow, that's terrible."

George Jagodzinski (32:26):

That's great. That's a much more enjoyable way of being too. So I love it. Thanks, Lynda, so much. I appreciate it.

Lynda Pak (32:33):

Oh, it's my pleasure. It's an absolute pleasure speaking with you.

George Jagodzinski (32:37):

Thanks for listening to Evolving Industry. For more, subscribe and follow us on your favorite podcast platform, and pretty please drop us a review. We'd really appreciate it. If you're watching or listening on YouTube, hit that subscribe button and smash the bell button for notifications.

(32:51):

If you know someone who's pushing the limits to evolve their business, reach out to the show at evolvingindustry@intevity.com. Reach out to me, George Jagodzinski on LinkedIn. I love speaking with people getting the hard work done. The business environment's always changing and you're either keeping up or going extinct. We'll catch you next time and until then, keep evolving.