Episode 153 · Season 7 · ·
The Refounding
Benjamin Herndon, Chief Strategy Officer at KUNGFU.AI, on why the race to automate everything misses the real prize, and why AI transformation is a people problem before it is a technology one.
Benjamin Herndon, Chief Strategy Officer, KUNGFU.AI. In conversation with John Horsley.
The Refounding
0:0033:00A strategist working where AI meets organisational change, from messy enterprise data to the premise of enterprise software itself, Benjamin Herndon argues the rush to automate everything is the wrong first move. The real prize, he says, is refounding the business, and the hard part is always people.
Herndon is Chief Strategy Officer at KUNGFU.AI, a company whose name, he explains, nods to the idea that kung fu means anything that takes discipline to master. His own space is strategy and organisational change, and he works alongside the firm's engineering staff to turn fast-moving technical change into organisational strategy. By his account the firm works across regulated sectors, including healthcare and life sciences, finance and fintech, and government, and is structured to bring clients something net new rather than staff augmentation or substitutes for their employees.
In this conversation with host John Horsley, Herndon argues that the current trend towards automating everything to cut costs is misplaced, because it ignores the full opportunity and treats AI as a substitute for people. He makes the case for refounding, returning to the principles that made a company great and shedding what no longer serves it, and warns that AI is already threatening the premise of enterprise software and the services around it. He is blunt about the risks: messy, uncurated data that amplifies past mistakes, false attribution across silos, and governance that controls without enabling, which drives shadow adoption and leaks proprietary data onto public tools. His through-line is that this is a people problem before it is a technology one. Do what only humans can do, he says, and make one plus one equal three.
Many companies have jumped at efficiency and cost cutting. I think you need to look at the business model first. Would you agree, and what is your approach?
“This is an opportunity to think, where should the road go? Not, where have the cows been for 100 years?”
I wholeheartedly agree. The current trend towards automating everything to reduce costs is misplaced, because it ignores the full extent of the opportunity and treats substituting AI for people as the best path to value. Automation has its place, but it is a tiny piece of the potential. Michael Hammer said, do not pave the cow paths. I use a term called refounding: go back to what made your company great, refocus on those principles, and let go of what is no longer necessary. It is a renaissance for leaders to re-envision what their company can look like in this new era.
Do you walk in and find companies have made a mess by putting AI in too fast?
“Coworkers are exercising judgment about, and they're leveraging prior experience and shared history”
A great example is thinking all you need to do is put a foundation model in front of your data and value falls from the sky. Most data is messy and not designed for that purpose, so without curation you get out what you put in, and you can inadvertently perpetuate a failed project or a loss. A lot of people say you can just talk to it like a coworker, which I think is a mistake, because coworkers exercise judgment and leverage prior experience and shared history. The AI is not going to do that out of the gate.
What does catastrophic look like?
“So you basically created a Wild West.”
Giving everyone unfettered access to foundation models and hoping value falls from the sky rarely works, and when it does it is in pockets you cannot attribute spend to. You create a Wild West. More problematic is failing to set up governance, and by governance I do not mean just controls, I mean training people on what good looks like. Without it, companies make catastrophic mistakes, like agents surfacing salary data or exposing personal information. Nobody sets out to release that data, but the AI does not know it is not supposed to take advantage of it.
Put yourself forward or recommend a guest for us to consider.
Request an interview →Most companies try to reach senior decision makers. We give them a reason to engage.
Partner with Us →Have your say.Add your voice to the conversation.
Voices is where senior practitioners share what they're seeing, what they believe and what they'd change.
Become a Voice →There is also shadow adoption, employees using their own AI. How big a problem is that?
“We've had senior C-suite leaders who were putting proprietary company data on a public Claude license just because they didn't have the most basic understanding of those privacy requirements.”
It is especially a problem if you put too much governance in place, or governance focused only on preventing mistakes, because people work around you to get the tools or bring their own licence. In prior epochs of software you had little access to anything the company had not procured. Now you can get a Claude licence for a few dollars a month. We have had senior C-suite leaders putting proprietary company data on a public Claude licence simply because they did not have the most basic understanding of those privacy requirements. Across a whole company with no oversight, that exposes you massively.
What advice would you give a business early in a transformation?
“All of those gains will be lost.”
This is not an easy thing you are setting out to do. It is centred in changing the basis of your organisational structure and how people think about how value is created. It is not about the technology, it is about the people. Executing change across a large base with long tenure and a strong culture is challenging. You get early momentum because the CEO is on board, but if the CEO loses focus or declares victory at a milestone, things regress and all those gains will be lost.
Never miss an episode. Follow Business of Marketing.
Read the full transcript
Lightly edited for readability
Speaker 1: My name's John Horsley. I'm the host of the show today. I have the absolute pleasure and delight of being joined by Benjamin Herndon, who is Chief Strategy Officer at KUNGFU.AI. Welcome to the show.
Speaker 2: Thank you. Very happy to be here.
Speaker 1: I've gotta say, firstly, what a cool company name.
Speaker 1: What was the origin? How come?
Speaker 2: We get that a lot. We get that a lot. Um, our CEO will tell you there are two answers. Uh, one is that he does not like naming companies, so he kind of w- uh, left it to one of our other co-founders. Uh, the other is that, uh, kung fu actually means anything that takes discipline to master.
Speaker 1: Mm.
Speaker 2: So it's not just the martial arts, but it can be anything. So we thought kung fu, uh, around AI reflects, you know, how hard it is and how important it is to do it right.
Speaker 1: Yeah, I love that. And the discipline.
Speaker 2: Yes.
Speaker 1: Yeah.
Speaker 2: Exactly.
Speaker 1: Um, it'd be really great for the, the listeners for the show to, um, just briefly explain the, I guess, the raison d'être, the purpose, your reason for being.
Speaker 2: Uh, as a company?
Speaker 1: Yeah, as a company.
Speaker 2: Really it's about helping bring our clients into the AI age and helping them succeed. So this is a massive change point for every, you know, company, industry, uh, market. You know, I'm sure this is not lost on your listeners. But AI is changing everything about the way that we operate. Um, and our job is really w- you know, across our vast client base to try to help them reinvent themselves and develop and sustain, uh, unique advantage in their futures from the application of AI and from being an organization that is able to effectively leverage AI for value.
Speaker 1: Sure. My assumption is that many companies are looking in the wrong place at this point in time, and what they've done is jumped for efficiency and cost saving-
Speaker 2: Mm-hmm
Speaker 1: ... and perhaps reduce headcounts, et cetera. Um, perhaps crudely using tools like GPTs or, um, various others within the business-
Speaker 2: Mm-hmm
Speaker 1: ... um, to craft content or to build d- develop workflows and so on. But what I fundamentally believe is that this is perhaps the wrong way around-
Speaker 2: Yeah
Speaker 1: ... and that you need to look at the business model first.
Speaker 2: Mm-hmm.
Speaker 1: Um, and then start thinking about how are we gonna go through the transformation. Uh, would that be correct? And, and what's, what's your practice? What's your approach? How do you consult with your clients?
Speaker 2: Yeah. Uh, that's a great question. Uh, first of all, I wholeheartedly agree. I mean, the current trend towards automate everything to try to reduce costs and create efficiencies, um, i- is, I think, misplaced, right? It... And it is misplaced for two reasons. One, it, uh, m- it does... it fails to take the full extent of the opportunity that AI presents into account, right? Um, the other thing that it does is it looks at substituting AI for people as the best path to value. Um, and I don't agree with either of those premises.
Speaker 1: Yeah.
Speaker 2: I mean, um, there are places, times and places where automation makes sense. But it is, uh, just a tiny piece of the potential. Uh, you know, uh, Michael Hammer, uh, back in the reengineering days said, you know, "Don't, uh, pave the cow paths," which was a way of saying, you know, you should... This is an opportunity to think, where should the road go? Not, where have the cows been for 100 years?
Speaker 1: Sure.
Speaker 2: And I think organizations are at the same point in time. The challenge is that, you know, doing this well requires... And this, I think, speaks to your point, like, what is our approach? Uh, really going back to the fundamentals, you know. And I like to use a term called refounding, which is like, like, you know, what is it that made your company or your organization great and successful and got you to this point in time?
Speaker 1: Mm.
Speaker 2: Um, refocus on those principles and try to let go of the things that are not necessary anymore. So it might be assumptions about how value is created that are not core to those initial value hypotheses. Uh, it might be, um, you know, functional design of an organization that's no longer necessary or can be-
Speaker 1: Yeah
Speaker 2: ... uh, modified in this new world. So it really is kind of a renaissance for organizational leaders to go back to those founding principles and re-envision what their company can look like in this new era.
Speaker 1: Mm. Yeah. That's, yeah. That, I whole- whole- heartedly agree with your approach. Um, the other thing I find interesting about AI at the moment, and perhaps some businesses are prepared for it, others maybe not, is that it also removes parts of the, the value that a business may have previously provided-
Speaker 2: Mm-hmm
Speaker 1: ... or the experience.
Speaker 2: Yeah.
Speaker 1: So you could look at retail, for instance, and clearly that's gonna fundamentally transform.
Speaker 2: Yeah. No, absolutely.
Speaker 1: Um, and, you know, e-commerce now is just a rep- a replication of a department store.
Speaker 2: Mm-hmm.
Speaker 1: But we're not gonna need to enter that department store any longer-
Speaker 2: Right
Speaker 1: ... because I'll be sending my agent there shopping.
Speaker 2: Yeah.
Speaker 1: Um, to a certain extent, or actually the customer journey on my side's changed.
Speaker 2: Mm-hmm.
Speaker 1: So I might be using, uh, an AI to discover a product-
Speaker 2: Mm-hmm
Speaker 1: ... um, source a supplier and so on. Um, and then go through to, you know-
Speaker 2: Mm-hmm
Speaker 1: ... X, Y or Z website to complete the purchase myself, but it's done the task of all the filtration for me-
Speaker 2: Mm-hmm
Speaker 1: ... which is the hard work, right?
Speaker 2: Yeah, exactly.
Speaker 1: Uh, and being confronted sometimes by choice and perhaps too much choice. Um, do you work with clients within, um, industries that are potentially under threat? So, okay, um, or that are gonna go through a heavy transformation. So retail's gonna go through an enormous transformation. I don't believe that all retailers are gonna make it out the other side. Consultancy businesses-
Speaker 2: Mm-hmm
Speaker 1: ... sell people.
Speaker 2: Yeah.
Speaker 1: Yeah, at scale into, to companies. Um, so your Accenture's, your Deloitte's-
Speaker 2: Yeah
Speaker 1: ... your Bains, et cetera, McKinsey's of the world. Um, okay, at the moment, they're putting AI into their clients.
Speaker 2: Mm-hmm.
Speaker 1: Great. Um, but- They're gonna need to transform-
Speaker 2: Mm
Speaker 1: ... enormously to, um, provide value in the, the new world-
Speaker 2: Yes
Speaker 1: ... as it were. Um, are you working with clients in these types of categories that, uh, uh, the models, uh, the, the, you know, the, the model of the business is gonna be enormously transformed-
Speaker 2: Yes
Speaker 1: ... to survive?
Speaker 2: Yeah, 100%. Um, you know, we don't ad- we don't consult to other consultancies in that regard, but you know, we work, we work with them, we partner with them occasionally, and they're all acutely feeling the pain that you're talking about.
Speaker 1: Sure.
Speaker 2: Um, a lot of it because so many of them, you know, went to kind of where's the high margin, uh, opportunity in consulting, which for years was enterprise software, uh, and all the services required to go around it. Um, AI is probably going to threaten, if it's not already directly threatening the premise-
Speaker 1: Yeah
Speaker 2: ... uh, of enterprise software. And so then you have to ask, do we really just need, you know, uh, people in seats to help facilitate our enterprise software installation or can that be done by agents?
Speaker 1: Yeah.
Speaker 2: So it's changing the economics of a lot of these consulting industries in ways that I think they're just beginning to feel.
Speaker 1: Yeah.
Speaker 2: Um, we are structured fundamentally differently, so we don't operate as, you know, staff aug or, you know, substitutions for your employees. We're bringing something fundamentally net new, and that could-
Speaker 1: Mm
Speaker 2: ... you know, again, be on the technical side where we're helping the client build something or get performance out of something that they've been unable to without someone like us. Um, or we're bringing them a roadmap to, you know, true organizational transformation that most organizations can't generate on their own. You have to have an outside trusted partner to have the objectivity, uh, to look at your organization and be a, a, you know, a true hand-in-glove, uh, partner, partner in that journey.
Speaker 1: Yeah. Yeah, I mean, you touched on software.
Speaker 2: Mm-hmm.
Speaker 1: Enterprise software, you know, f- frankly, most tech out there-
Speaker 2: Mm-hmm
Speaker 1: ... is completely rep- uh, you can replicate it in days now.
Speaker 2: Mm-hmm. Yeah.
Speaker 1: Um, last week, and actually you may have even got, um, uh, an e-signature thing sent out to yourself, I built an e-signing platform in four days.
Speaker 2: Mm-hmm.
Speaker 1: Um, and put it through a lot of stress testing, security-
Speaker 2: Yeah
Speaker 1: ... various sort of things, and that's it.
Speaker 2: Yeah. It's amazing.
Speaker 1: Um, I now no longer need the service that I was paying probably 100 bucks a month for.
Speaker 2: Right. Exactly.
Speaker 1: That's now completely redundant. And more importantly, I've actually got something that works better for me 'cause it suits the way I want to use it and the workflow that I needed, and has additional things-
Speaker 2: Mm-hmm
Speaker 1: ... that I wouldn't be able to get in someone else's software.
Speaker 2: Yeah.
Speaker 1: So it's better custom- customized to my use case.
Speaker 2: Mm-hmm. And that's ... If you think about so much of the services that originally were wrapped around enterprise software implementations, it had to do with that customization. It had to do with those data linkages. Um, you know, a lot of those things that are now relatively self-service. You know, you can set up the data linkages on your own, you know, with, you know, an agent or with, you know, a, a customization engine. You can tell it exactly what your use case is, and you don't have to go through this painful process of-
Speaker 1: Yeah
Speaker 2: ... you know, turning off half the functionality in the product you just bought-
Speaker 1: Yeah
Speaker 2: ... you know, so that it can work in your context.
Speaker 1: Are you helping your clients develop technology or better connect it so that ... I do, I'm thinking about data-
Speaker 2: Mm-hmm
Speaker 1: ... frankly-
Speaker 2: Yeah
Speaker 1: ... rather than necessarily the tech.
Speaker 2: Mm-hmm.
Speaker 1: Are you helping them better, um, connect to the data that would be sat in disparate as- applications?
Speaker 2: Yes, absolutely. Now, you know, we will try not to do that if all they're trying to do is bring together data into like a warehouse or a lake or something like that.
Speaker 1: Sure.
Speaker 2: Um, we will do it when it is critical to delivering the value of the AI.
Speaker 1: Yeah.
Speaker 2: So, um, and a great example was a company that we worked with that had proprietary data in 20 different locations across different continents, uh, and the data could never leave the location. So we had to create a federated learning, uh, system re- engineering across all those different, uh, uh, well, repositories, and train the model in a way that basically was as if you were doing it, you know, in one common place. So-
Speaker 1: Mm
Speaker 2: ... um, but we couldn't have delivered the value of that model without tackling that-
Speaker 1: Yeah
Speaker 2: ... engineering challenge.
Speaker 1: Were they in a, um, restricted industry of sorts, or regulated?
Speaker 2: Yes, they were in a healthcare diagnostic-
Speaker 1: Okay
Speaker 2: ... space. So, you know, with ... Most of the data was resident in hospitals-
Speaker 1: Yeah
Speaker 2: ... and that data could not leave those confines. And obviously, the PII associated with that data was highly, um, protected.
Speaker 1: Mm.
Speaker 2: So, you know, we had to go through quite a bit to be able to leverage what we needed to to train the model.
Speaker 1: Yeah. Is that one of your specialisms? Do you work with other companies in highly regulated categories and industries?
Speaker 2: We do, yeah. We work quite a bit in healthcare-
Speaker 1: Mm
Speaker 2: ... um, and, and life sciences. We do quite a bit in finance industry, fintech as well.
Speaker 1: Yeah.
Speaker 2: Um, uh, we work in government, so we've done quite a bit of work, uh, in this government, you know, across sectors, security space, IRS.
Speaker 1: Mm.
Speaker 2: Um, and, you know, you obviously have to be very mindful of the regulations-
Speaker 1: Yeah
Speaker 2: ... and unique requirements.
Speaker 1: Yeah.
Speaker 2: Yeah.
Speaker 1: Um, I'm thinking, let's be honest, you know, business isn't ever clean. Most businesses are messy, right?
Speaker 2: Mm-hmm.
Speaker 1: Um, do you find yourself walking into a company and thinking, "Actually, these guys haven't thought about this. They've, they've, they've actually caused a bit of a mess-
Speaker 2: Mm-hmm
Speaker 1: ... by trying to put the AI in perhaps in a hurry in their approach"?
Speaker 2: Yeah.
Speaker 1: What, what might that look like sometimes?
Speaker 2: Uh, well, a great example is like, you know, thinking that all we need to do is get all of our data into one place, and then we can put, you know, a foundation model in front of it, and then, you know, value is going to fall from the sky. Um, and most organizations' data is messy, um, and it's not designed for that purpose. So if without any kind of curation of the data asset, you're gonna get out what you put into it, and sometimes that could be misleading, right? You could have a huge corpus of data, uh, in your repository that came from a failed project or from a, you know, something that ended up in, you know, a loss of revenue. So-
Speaker 1: Yeah
Speaker 2: ... you know, you're inadvertently perpetuating a lot of that by telling the AI that it's something good to access. Now, it all depends on, you know, again, wh- how you prompt it but, you know, a lot of people are trying to get in front of, you know, that by saying, "Well, you don't need. You can just, you know, talk to it like you would a coworker."
Speaker 1: Mm.
Speaker 2: Which I think is a mistake because coworkers are exercising judgment about, and they're leveraging prior experience and shared history and-
Speaker 1: Sure
Speaker 2: ... the AI isn't necessarily gonna do that out of the gate.
Speaker 1: Yeah. And do you find that people also ... I mean, hey, look, businesses have obviously got a lot of pockets of or silos of data-
Speaker 2: Mm-hmm
Speaker 1: ... and information.
Speaker 2: Yeah.
Speaker 1: That sometimes that when they're putting the AI in, all they're doing is amplifying the bad, as you've, you've, you kind of referenced there-
Speaker 2: Yeah
Speaker 1: ... um, because they've not really thought about where else the information may reside.
Speaker 2: Mm-hmm. Yeah.
Speaker 1: Um, I'm thinking, you know, for instance ... I'm thinking from the standpoint of marketing, but-
Speaker 2: Mm
Speaker 1: ... um, you might have a call center. Pe- customers are calling into it. Uh, you probably got people on Zoom calls speaking to your clients.
Speaker 2: Mm-hmm.
Speaker 1: Um, speaking to prospective buyers, handling, um, perhaps internal conversations-
Speaker 2: Mm-hmm
Speaker 1: ... as well. And, and there's knowledge and information that's, that's stored there, and there's another silo.
Speaker 2: Right.
Speaker 1: Uh, they might have a customer database or a CRM-type system.
Speaker 2: Mm.
Speaker 1: They might have, uh, numerous other systems and technologies within the business.
Speaker 2: Mm-hmm.
Speaker 1: For sure they've obviously got a website. There's something that's powering the website, and there's a hell of a lot of data there.
Speaker 2: Yeah.
Speaker 1: Um, they might have other campaign tools and, and so on. Um, but it would be my assumption that m- many businesses would be perhaps trying to ask questions of the data and get an answer, but they're just amplifying the bad because-
Speaker 2: Mm-hmm
Speaker 1: ... they're only as- asking question in one particular silo here rather than something that's joined up across that-
Speaker 2: Yes
Speaker 1: ... entire customer experience and, and that sort of customer journey kind of ecosystem is probably a nice way of describing it.
Speaker 2: Yeah. No, ex- that's exactly right. You know, and a lot of times even if you are interlinking data in these different silos, if it's not curated for a common purpose, um, you'll, you'll bump into the same problem you would have, you know, when you were trying to figure out a answer to a custom question. You know, like h- attribution. Attribution is notoriously difficult. You know, everybody thinks-
Speaker 1: Of course
Speaker 2: ... if we just connect data across all these silos, attribution will solve itself. Well, you know, if you've got, you know, a cadence over here of data collection that doesn't match the cadence over here or-
Speaker 1: Mm
Speaker 2: ... you know, you don't have the, the, the, the sort of metadata to say what was the origination of this, you know, are you really able to make the attributions that you wanna make? You may not be able to or you may be making false attributions. Um, and that's, that, to your point, is perhaps worse.
Speaker 1: Yeah. What are the typical use cases? What are ... What does good look like?
Speaker 2: Mm-hmm.
Speaker 1: Is probably a, a great question.
Speaker 2: Um, I mean, good from a client perspective is someone who's, you know, committed to the journey, right? They're not just trying to check a box. They're not just trying to ease some, you know, abstract fear that they're gonna get, you know, made obsolete by some competitor. So you're talking about cl- clients who recognize the power of this technology. They recognize its pervasiveness, and they're committed to that journey across, again, not just, you know, the technologies but across the opportunity afforded for reinvention.
Speaker 1: Mm.
Speaker 2: So they're really willing to challenge a lot of those legacy assumptions that they've had in the past and see what, you know, w- what they look like when they come out on the other side.
Speaker 1: Yeah. Um, and the ... And I guess the flip side of that, what does catastrophic look like?
Speaker 2: Well, catastrophic in, in our universe typically looks like, you know, a couple of different scenarios. One is, you know, we give everybody, you know, unfettered access to foundation models and tell them to, you know, go forth and, and multiply, and you hope that value falls from the sky. Um, you know, it, it rarely does. When it does, it's in pockets. You can't attribute the spend that's happening to the actual value that's being created.
Speaker 1: Yeah.
Speaker 2: So you basically created a Wild West. Um, but more problematic than that is that, you know, you haven't, in a lot of cases, set up the appropriate governance, uh, processes. And, and by governance I don't mean just controls. I mean, like, training people on what good looks like.
Speaker 1: Mm.
Speaker 2: You know, what are the things you need to take into account when you stand up agents, when you give it access to data? Um, and you know, without that governance in place, we see companies that are making catastrophic mistakes.
Speaker 1: Mm.
Speaker 2: You know, people putting th- you know, giving agents access to data that they shouldn't have, right? So you might think, "Okay. Well, you know, it'll be great if we could mine our, uh, internal data for, uh, expertise or efficiencies or things like that." But if it inadvertently has access to salaries and it starts releasing those salaries or using them as an input to its analysis, you get, you know, a massive violation of, uh, employee trust. Um, if you're talking about, um, you know, healthcare systems for instance, you know, you've gotta be very careful about, you know, all that personally identifiable information that's associated-
Speaker 1: Mm
Speaker 2: ... with records and with, you know, reimbursement codes and all that stuff. And it's not that anybody would set out and say, "I'm trying to release this data." It's that you would, you know, in trying to do something valuable and innovative-
Speaker 1: Yeah
Speaker 2: ... you have pointed at this data inadvertently and, you know, the AI doesn't necessarily know it's not supposed to take advantage of it.
Speaker 1: Mm. And I guess another issue would be around, in the past we would have called it employee tech.
Speaker 2: Mm-hmm.
Speaker 1: Yeah, but in essence, uh, employees using dark tech, i.e., their own AI.
Speaker 2: Yeah. Yep.
Speaker 1: Yeah. Uh, that-
Speaker 2: Shadow adoption is a term
Speaker 1: ... yeah, that must be a, a huge problem as well.
Speaker 2: Well, and that's a, you know, especially a thing if you, um, put too much governance in place or governance that is only focused on preventing people from making mistakes, um, because people will say, "Well, you know, but I have an opportunity to here create value or do something-
Speaker 1: Of course
Speaker 2: ... impactful," and so they'll go work around to get access to those tools or they'll bring their own license. Um, and yeah, that's massively dangerous.
Speaker 1: Mm.
Speaker 2: So, you know, if you think about, you know, prior epochs of software, you probably didn't have much access to, you know, software that the company hadn't procured and put on your laptop.
Speaker 1: Yeah.
Speaker 2: These days you can go get a Claude license for, you know, a few dollars a month-
Speaker 1: Yeah
Speaker 2: ... and, and access it anytime you want to. And if you don't, you know... And we ha- we've had senior C-suite leaders who were putting proprietary company data on a public Claude license just because they didn't have the most basic understanding of those privacy requirements.
Speaker 1: Yeah.
Speaker 2: Um, and so if you have people across the company doing that with absolutely no oversight or understanding, um, yeah, you expose yourself to a, a massive potential, uh-
Speaker 1: Yeah. Are companies lacking policies? So, you know, should they have proper policies in place in terms of governance, um, and, and be better at communicating this to, you know, the, the employees within the business?
Speaker 2: Absolutely. Um, I mean, and it's, you know, a lot of companies that we've worked with, you know, they'll say, "Well, we just need the basic policies," and that might be them trying to check a box, let's say.
Speaker 1: Mm.
Speaker 2: Um, but to your point, without the processes to socialize those policies, make sure they're understood, but also to enable people, right? It's one... I mean, if you only focus on the control side, you know, then you'll have people saying, "Oh, well, all the company is telling me to do is what not to do." But if you also give them opportunities to better learn how to leverage the technology, help them understand pathways where they see value that might scale beyond their own usage, then you start to change the governance converna- conversation from a control mechanism into a, a, a value creation mechanism.
Speaker 1: Mm. Yeah, I'm just thinking about that for a moment. Um, and the companies that get it right, I mean, do they have ... Have they put responsible individuals in place? It's probably a good way of describing it. Or is there some sort of hierarchy, so would they have a, you know, a chief AI officer-
Speaker 2: Mm-hmm
Speaker 1: ... as it were? Uh, or, you know, is the chief information officer taking on that responsibility?
Speaker 2: Mm-hmm.
Speaker 1: How does that work in terms of the, creating the, the correct sort of corporate hierarchy-
Speaker 2: Mm-hmm
Speaker 1: ... to be able to put effective policies in place?
Speaker 2: Yeah, I mean, that's a tough one because most, you know, compliance offices within companies, again, their job is to, you know, be more restrictive than enabling, right? So their job is to say, "Let's make sure that everything that we're doing compli- you know, is consistent with this policy or that policy," and anybody who's associated with it sees it as a necessary evil. Um, you know, the challenge is AI is not only, you know, not just, you know, something you have to do, have to do with certain guardrails, it's something that's enabling and touching and transforming every part of the organization. So, you know, it requires more than just, you know, restrictive covenants. The other problem is that, you know, AI doesn't fit neatly into a lot of the legacy functional structures in organizations. So we've seen a lot of consternation when, you know, a CTO, for instance, is given the full mandate of AI because it's a fundamentally different scope. You know, they... It's a fundamentally different budgeting mechanism, right? How do you, how do you assign, you know, investment dollars to that when the, you know, historically the CTO's office was a cost center?
Speaker 1: Yeah.
Speaker 2: Um, you know, are you en- enable- are you em- uh, empowering them to be the governance mechanism, right? Because that's something they have done, again, from a security perspective, but not from a human behavioral perspective necessarily.
Speaker 1: Mm.
Speaker 2: So, you know, and then, you know, we see other companies that put it in, like, a chief product office, uh, where, you know, those are obviously different types of companies, but then they focus primarily on, you know, how we can, you know, go to market as fast as we can with the latest, greatest-
Speaker 1: Yeah
Speaker 2: ... you know, features, and sometimes that creates other, uh, you know, violations like downstream problems with users that then result in unwanted complaints and tickets-
Speaker 1: Yeah
Speaker 2: ... associated with people just fundamentally not understanding how the technology works.
Speaker 1: Yeah. I, I don't think we c- you know, we, we can't disagree that AI is obviously very powerful in terms of transforming pretty much every function within a, an enterprise.
Speaker 2: Mm-hmm.
Speaker 1: Um, and I would imagine, you know, it could be HR, it could be marketing, you know, et cetera, are probably reaching out to their tech colleagues at this point in time. They might have the correct level of understanding, they might not.
Speaker 2: Yeah.
Speaker 1: Um, they might have that hierarchy in place. Again, they might not. But I would imagine even if they do have that hierarchy and they do have highly skilled individuals within the tech team-
Speaker 2: Mm-hmm
Speaker 1: ... where are they gonna focus first and why?
Speaker 2: Mm-hmm.
Speaker 1: So, uh-
Speaker 2: Yeah. That's a great question
Speaker 1: ... how would you go about that? What, what's, you know, what's the approach? How do you start making those decisions?
Speaker 2: Yeah, I think it's, uh, you know, it's really about, you know, right-sizing the approach to the organization and where it is in its journey, right? So, you know, companies that are earlier in that journey, they probably don't need, you know, massive, you know, restrictive governance policies. They probably don't need procedures or processes that require everything to go through approvals. Um, you know, but they probably do need to make sure they have, you know, people who are laser-focused on financial accountability-
Speaker 1: Yeah
Speaker 2: ... because they can't afford, you know, to lose money on a, on a bad investment in an AI initiative. You know, at the other end of the spectrum are the large enterprises where they've got, you know, thousands of projects in flight, and they've got, you know, people all over the place innovating and experimenting. Um, those kinds of organizations, you know, will have to, you know, really run the gamut from a governance and oversight perspective, again, as well as enabling. 'Cause their future is dependent upon the ability of their people to adopt and innovate with these technologies.
Speaker 1: Yeah. Yeah. I mean, I, you know, I look at some of my clients, they're marketers-
Speaker 2: Yeah
Speaker 1: ... and they can't even get a, a bloody page on their, on their website.
Speaker 2: Right.
Speaker 1: Because it's not, you know, not top priority and it never has been of-
Speaker 2: Yeah
Speaker 1: ... the, the H or sorry, the, the tech team who own the site. Perhaps it sat in the wrong place.
Speaker 2: Right.
Speaker 1: Perhaps it should've always sat with the marketers.
Speaker 2: Yeah, exactly.
Speaker 1: Yeah. 'Cause they deal with the customer in the first place.
Speaker 2: Mm-hmm. 100%.
Speaker 1: Which, which is, it's nuts. And I look at many businesses, and ownership often seems to sit in the wrong place-
Speaker 2: Yes
Speaker 1: ... for a, a whole myriad of things.
Speaker 2: Yeah.
Speaker 1: Um, do you experience that when you go in, and-
Speaker 2: Oh, yeah
Speaker 1: ... does that hinder the transformation?
Speaker 2: Yeah, yeah. Because, you know, like for instance, you know, who owns software in, for a company that, you know, sells software?
Speaker 1: Mm.
Speaker 2: The whole company owns it, right?
Speaker 1: Yeah.
Speaker 2: I mean, they all have a role in, in, you know, ideating it, developing it, testing it, bringing it to market, deploying it.
Speaker 1: Yeah.
Speaker 2: And, you know, to say that, you know, okay, it's this part of the organization that owns the AI without looking at all the cross-functional dependencies that are created when you innovate with AI, and you have to... You know, for instance, again, if you, and if you put a new AI feature in software and you send it to market, but you haven't, um, prepared clients for what it's gonna look like or how it's gonna change, you know, you go through all of that when you do your initial implementation, but you tend to under-think, you know, what's required when you make updates to it. But, you know, you can easily push an AI feature out in a software product, and if people don't understand it, your NPS scores go through the, the floor.
Speaker 1: Sure.
Speaker 2: Um, same thing with marketing. Like we've worked with clients where, you know, we've massively sped up the software development life cycle process. Um, but if you don't also equip, you know, uh, you know, marketing and go to market with the implications of, of faster release cadence-
Speaker 1: Mm
Speaker 2: ... um, or an understanding of, you know, what, how it might change the way that your products look and feel, um, you're gonna create other frictions within the organization that, you know, you're gonna get in your own way.
Speaker 1: Sure. Mm-hmm. Yeah.
Speaker 2: Um.
Speaker 1: Yeah. You will. Yes. So we're... And what sort of advice... Yeah, uh, yeah, what advice would you provide to, um, a business that's early on in terms of a transformation?
Speaker 2: Mm-hmm.
Speaker 1: Could be perhaps a mid-sized business or an enterprise company.
Speaker 2: Mm-hmm. Yeah, I mean, I would, you know... I, and I do say this all the time. I say, you know, "This is not an easy thing that you're setting out to do." Um, you know, it's very much centered in changing the, the, you know, the basis of your organizational structure, the way that people think about how the organization creates value. Um, and that's a very difficult thing to do. It's not about the technology, it's about the people.
Speaker 1: Yeah.
Speaker 2: And to execute change among a large base of people with a company, particularly when there's a long tenure and a strong culture, uh, it's challenging.
Speaker 1: Yeah.
Speaker 2: And, you know, the other thing that I see that happens a lot is you get a lot of early momentum because, you know, the CEO's on board with it, and, you know, they... Then people are, you know, excited about it or they're just, you know, playing along 'cause they don't wanna get in trouble. But, you know, if something changes, the CEO maybe loses focus on it or decides, "Hey, you know, we've, you know, we achieved this milestone, so let's declare victory," um, things will regress to the way that they were-
Speaker 1: Yeah
Speaker 2: ... before.
Speaker 1: Yeah.
Speaker 2: All of those gains will be lost.
Speaker 1: Yeah. I'm also thinking, say you've got the board, you have CFO, CEO, you know, COO, et cetera.
Speaker 2: Mm-hmm.
Speaker 1: Um, and they m- might all be... They're obviously going on a journey themselves, but they might be looking on ChatGPTs for some of things.
Speaker 2: Mm-hmm.
Speaker 1: Trying to get advice. How should we go about our transformation? How should we do this? And so on.
Speaker 1: Uh, and then perhaps they come with conflicting and also misinformed points of view-
Speaker 2: Mm-hmm
Speaker 1: ... that they feel are well-informed.
Speaker 2: Yeah.
Speaker 1: Um, and that causes hostility and, and conflict as well-
Speaker 2: Mm
Speaker 1: ... of course. Um, do you mediate in those situations? Do you come across situations like that at all?
Speaker 2: Um, less so. I mean, I think you do see a lot more, like, self-driven research where, you know, they're like, "Well, I asked Claude this or ChatGPT this," and put-
Speaker 1: Yeah
Speaker 2: ... put these challenges to it. Very rarely would I say that it oversimplifies it for the, the leader.
Speaker 1: Yeah.
Speaker 2: More often than not, I think it's, it just kind of opens the leader's eyes a bit to, well, this is harder than I thought it was gonna be.
Speaker 1: Yeah. Okay.
Speaker 2: And that's, that's a great entrance for us, right? To come in with that conversation so that, you know, the leader's already now d- you know, has dispelled the notion that this should be relatively easy, cheap, or short.
Speaker 1: Yeah.
Speaker 2: Um, and, you know, for us then to come in and say, "Okay, now that, you know, we're past that, let's talk about what's really involved here."
Speaker 1: Mm. Um- Just wrapping up, I normally-
Speaker 2: Mm-hmm
Speaker 1: ... ask a couple of questions. But do you have any go-to resources that you recommend that help keep you sharp, or perhaps just places that you might go to gain a level of inspiration or, um, to help you free up your thinking?
Speaker 2: Mm-hmm. Yeah, I mean, honestly, that's one of the great assets that our company provides, which is, you know, access to other people who are, you know, again, not just in my space, which is, you know, strategy and organizational change. Um, but we have an incredible engineering staff. And so, you know, through them I'm able to keep up with and understand in a quick and accessible way what's changing on the technological side of things. And through the interaction, you know, through us having regular conversations around, okay, well, what does this mean for... what does this technological change mean for organizations, you know, we're helping each other kind of unpack, uh, those implications. Um, on my own, I just read a lot. I mean, everybody-
Speaker 1: Sure
Speaker 2: ... should be.
Speaker 1: Yeah.
Speaker 2: Um, and you have to be increasingly careful these days, you know, of what you read and the sources you go to, uh, because AI has, you know, prol- proliferated a lot of, you know, slop.
Speaker 1: Sure.
Speaker 2: Um, so you know, I to- tend to follow, you know, a few people who I think are, you know, thinking the right way about, you know, this and the implications. But, um, yeah, I don't really limit it to any one-
Speaker 1: Yeah
Speaker 2: ... channel or source.
Speaker 1: And then do you have any advice for a younger individual perhaps they're at the start of their career?
Speaker 2: Mm-hmm.
Speaker 1: Um, or, you know, perhaps they've, they've graduated more recently-
Speaker 2: Mm-hmm
Speaker 1: ... and looking to get into the, the workplace, as it were.
Speaker 2: Mm-hmm.
Speaker 1: Um, I know that opportunities have changed-
Speaker 2: Mm-hmm
Speaker 1: ... um, for younger people. Uh, and it's harder for them to get into the job market.
Speaker 2: Yeah.
Speaker 1: Um, certainly in the entry-level jobs, because, hey, a lot of them have been automated-
Speaker 2: Yeah
Speaker 1: ... in some businesses-
Speaker 2: No, it's typical
Speaker 1: ... or types of businesses. So is there any advice in terms of, yeah, um, to a younger individual, how they should be looking to enter the, you know, the world of work-
Speaker 2: Yeah
Speaker 1: ... as it were?
Speaker 2: No, that's a great question. And honestly, it's not different from, you know, my vision for, you know, how most companies need to start thinking about, uh, how their, um, human machine integrations happen. Uh, and I would say focus on the things that the human does well, and if... as a student or as a young person entering the workplace, what are the things that you bring to the table that are difficult, if not impossible, to automate or replicate with an AI? And the second part of that is understand how you can work with the AI to make the, the aggregate output of your collaboration better than just the sum of the parts.
Speaker 1: Mm-hmm.
Speaker 2: Right? So the one plus one equals three.
Speaker 1: Yeah. No, great advice. Thank you so much.
Speaker 2: Yes.
Speaker 1: Uh, you've been a great host, oh, sorry, a great guest-
Speaker 1: ... on the show today.
Speaker 2: You've been a great host. Thanks for having me.
Speaker 1: Thank you so much. Cheers.