Technology Business Research, Inc.The Playbook for AI Reinvention
Episode highlights:
- The skills required for successful AI adoption
- The importance of CEO-led AI programs
- Potential points of failure
“AI isn’t really comparable to, you know, cloud or mobile. Those were infrastructure shifts. So, you could delegate them because they didn’t fundamentally change who did the work or how decisions get made. AI does. It changes the shape of jobs. It changes which decisions get automated and which ones don’t, you know, it changes headcount assumptions, and those are CEO questions by definition. You can’t delegate them without the whole thing turning into a science project,” said Cooper.
Listen and learn with TBR Talks!
Submit your Key Intelligence Questions for Patrick and his guests: https://bit.ly/3T9VZek
Learn more about TBR at https://tbri.com/.
TBR Talks is produced by Technology Business Research, Inc.
Edited by Haley Demers
Music by Burty Sounds via Pixabay
Art by Amanda Hamilton Sy
The Playbook for AI Reinvention
TBR Talks Host Patrick Heffernan: Welcome to TBR Talks: Decoding Strategies and Ecosystems of the Globe’s Top Tech Firms. Where we talk business model disruption in the broad technology ecosystem from management consultancies to systems integrators, hyperscalers to independent software vendors, telecom operators to network and infrastructure vendors, and chip manufacturers to value-add added resellers. We’ll be answering some of the key intelligence questions we’ve heard from executives and business unit leaders among the leading professional IT services and telecom vendors.
I’m Patrick Heffernan, Principal Analyst, and today we’ll be talking about business process reinvention and implementation of AI-enabled solutions with Milan Cooper, Head of Product at TWG AI.
Meet Milan
All right, Milan, thank you so much for coming on TBR Talks. Really appreciate it. Maybe you could give us a little bit of your background, then we’re going to dive into some of the topics that we’ve already been talking about with respect to AI, but maybe you could tell us a little bit about where you’re at right now and some of the experiences you’ve had professionally over the last decade or so.
Milan Cooper, Head of Product at TWG AI: Yeah, absolutely. Thanks, Patrick. It’s great to be here. So, I’m Milan Cooper. I head up the product and go-to-market team at TWG AI. We’re an AI products and services company, and we work mostly in regulated industries. So, think financial services, insurance, you know, like places where you can’t just ship something and see what breaks. So, before TWG, I spent about four or five years at J.P. Morgan, running AI for software development. And then for about a decade before that, I was at Accenture in technology strategy. I actually started my life as a network engineer, but kind of got sick of being in data centers at 2 A.M., fixing issues on the weekend. So migrated very quickly into strategy and have progressed a career since then. So pretty much my whole career has basically been about getting big, complicated organizations to actually adopt new technology across a number of different waves, which as it turns out is actually the hard part.
What skills are required for successful AI adoption
Patrick: And so, getting big corporations to adopt technology. So, when we think about AI, one thing I’ve heard, and I’ll just, I’ll float this by you and see if it makes any sense, is when it comes to adoption, particularly for a large organization, you need three things. You need the leadership to be bought in, you need the masses, meaning you need adoption or at least experimentation or at least familiarity, willingness across the masses. And then you need, for lack of a better word, a lab. You need people that are actually dedicated to taking that technology, whether it’s, let’s just say it’s AI, taking AI and actually making it useful within the enterprise. Is that, would you agree that those are sort of the three critical components? Is there something else? Is there something that we’re missing?
Milan: No, I think absolutely the top-down leadership is absolutely critical to drive change. The lab component, I would probably describe it as kind of an AI SWAT team. So, the models that exist today are phenomenal, and they’re only getting better. The trick is to have a team that knows how to manipulate those models with context, with data from your business and actually embed them into the workflows of the end users in whatever business line you’re trying to impact. And so, you kind of have to be a little bit of a Swiss army knife. You have to have systems engineering. You have to know how to integrate into systems of record, pull data in. You have to be an expert in the models. You have to know how to manipulate these things and how to build multi-agent systems that work together effectively. And then you have to know enough to be dangerous in the business, and have empathy with users that have existing workflows that kind of run operations today and understand what are the unique things about how they work and what their business is. And then be able to, kind of, embed and integrate that into new reimagined workflows. And so that kind of lab concept of just having research scientists kind of build these things in a dark room is kind of gone. And it’s now this kind of Swiss army knife of skill sets that is required to take something from, you know, that comes out of the, you know, the foundation model providers and actually embed it into an actual business workflow. But, you know, the CEO-led AI transformation is absolutely critical as well.
Patrick: I’m curious, when you talk about all those different components of what you need to be able to bring to the table as an, really as an individual, like how, was it something at Accenture or something at J.P. Morgan that sort of, that helped you develop those skills? And as part of what you’re doing now with TWG AI, is that, is part of that, like you’ve got the skills and the capability, the actual people, the humans, that can do it, and that’s what you’re bringing to your clients?
Milan: Yeah, I think the- I think pattern recognition across working with, you know, tens of companies that are all facing similar challenges kind of pointed us to a clear need to build a team that has these skill sets. So, the big question is, is how do you build an organization that can effectively do this work and what skills do you need? And kind of the experience that we had and myself with consulting and working with lots of different companies and then embedded in financial services industry for a number of years, kind of gave me and us that pattern recognition of, you know, you need all of those things to be effective. It’s not just- you can’t just have a research lab anymore. So yeah, that was kind of a key principle when we were kind of founding this company.
The underlying playbook for AI reinvention
Patrick: So how long have you been with TWG AI then?
Milan: Yeah, so TWG started early last year and we’re part of the overall TWG Global Holdings family of companies. And so, we own asset managers, we own investment banks, insurance companies. We own a huge sports franchise. So, we own the Dodgers, the Lakers, motorsports. We just launched Cadillac Formula One recently. Rodeo is part of our business as well. So, we have a huge, diverse set of of problems and challenges that we have to try and solve with AI. And we’re taking those learnings to market as well. And so, we’re working with other sports franchises, colleges, and then also heavily in financial services and trying to reimagine those businesses with AI.
Patrick: So, you mentioned earlier, you need to have enough knowledge about a business to be dangerous. And yet, I can’t imagine, or maybe you can tell me, where are some of the overlaps that you see between, rodeos and banks or, you know, Cadillac and the Dodgers, like I definitely see where those overlaps exist. But, the regulated industries and, well I guess Formula One would be pretty heavily regulated too. But I’m just curious, how do you have enough knowledge to be dangerous about a business when you’re involved with so many wildly different kinds of companies?
Milan: I think that’s the magic. And so, you know, we ask a lot of questions to our clients. And our clients ask a lot of questions of us. And I think the magic is really when you have a dense technical team with the skill sets that I mentioned earlier around context engineering, manipulation of LLM, and also the experience and the empathy to embed these things within a business. And then you work with SMEs. So, we ask a lot of questions to end users, and we understand what are the nuances of how they work and how they operate. And we learn from them. And those two things coming together is kind of where the magic is. I mean, the interesting thing is we’ve now done this in motorsports and insurance and financial services. And the surface details look completely different, but the underlying playbook is the same every time. A clean data layer that represents the dynamics of your business, clear decision points, human in the loop where it matters, measurable outcomes that you’ve agreed with the CFO before you even start the project. And the fact that the same approach works in the pit lane in a Formula One race and on underwriting desks in an insurance company kind of tells you something important, that this is a methodology problem, not just a technology problem.
The importance of CEO-led AI programs
Patrick: I might agree with that. I’m- actually, I’ll go ahead and say I definitely 100% agree with that, but I don’t know- how much pushback do you get from the companies you’re working with, at least initially, around that assertion that what- they’re not buying technology necessarily, they’re buying a change to their structure and how they work and their processes and their culture.
Milan: I mean, we, it’s an ongoing conversation, but I think one of the things that we evangelize from day one on an engagement is that an AI program needs to be CEO-led. It needs to be top-down. That’s how you drive real change, and that’s how you kind of get out of, you know, pilot purgatory where you’re doing many experiments, but you’re not seeing much impact in the P&L.
Patrick: Right.
Milan: And I think what’s actually changed in the last 12 months is that, kind of, AI has moved from being an opportunity on the CEO’s desk to actually being a liability on the board’s risk register. So, if something goes wrong with an AI deployment in a regulated business, you know, if a model makes a decision it shouldn’t have or customer data ends up somewhere it shouldn’t, an employment tribunal gets interesting. And that lands on the CEO’s desk personally and is something that they have to deal with. So, it’s not a CIO or a technology problem only anymore. And once you understand that, the question of whether the CEO should be involved kind of answers itself. They already are. They just don’t really know it yet. And the thing is, being involved isn’t the same as being in charge. And the test I use to figure out whether a company’s AI program is actually going to work is really simple. Who controls the budget? And if the answer is the CIO or technology or somewhere else, it’s probably going to fail. Not immediately, but eventually. And the answer, you know, if the answer is the CEO or the COO or one of the business unit leaders and P&L owners, it has a chance. The incentives are aligned. And the reason is that AI budgets sitting inside IT get spent on infrastructure, platforms, licenses, sandboxes, and all the stuff that’s easy to procure and impossible to measure. An AI budget that gets to sit with a business owner gets spent on, changing how work gets done, because that’s really what the business owner is judged on. Same money, completely different outcome, productivity, revenue generation, risk reduction.
Potential points of failure to look out for
Patrick: So, one thing I would have to push back on a little bit is you mentioned you were at Accenture for a while doing technology strategy. And during your time, no doubt you advised the CIOs and the CTOs you were talking to that they needed to think of themselves as and they needed to operate as part of the business. They couldn’t just be a cost center. They couldn’t just be somewhere where, you know, like you said, there’s money that just gets spent on procuring infrastructure and things that are easy to measure. So that advice at some point sunk in because the CIOs and the CTOs we talk to today tell us all the time that they’re part of the business, and they see themselves as driving revenue growth, not just cutting costs. But it sounds like maybe we’re hearing one story, but you’re hearing a very different one if what you’re experiencing with respect to an AI-enabled solution implementation isn’t really going to be successful if it’s just the CTO or the CIO in charge, yeah?
Milan: I think so. I mean, there’s so many different failure modes for these projects. This isn’t- this isn’t like software where you plan a big software implementation, you have a budget, you have a timeline, you allocate a team, you have outcomes, and then you just execute. And it’s almost binary. You know, you executed on time, on budget, or you didn’t. Data science is, you know, by definition, an experimental discipline. And so, it requires testing and learning. It takes you to places that you didn’t know you were going to go. You have to solve problems that you didn’t know existed. For example, a lot of companies have AI policies or control procedures that were written 10/15 years ago by kind of people that are no longer with the company. And so those are some of the things that can kill you or can really harm projects. You know, most big companies are kind of silently run by control procedures that were written, 10 to 15 years ago, by people who have long since left. And those controls have kind of permeated a culture to the point where nobody questions them. They’re just how things work. And then you try and do something with AI and you hit a wall and, you know, the wall is actually a policy from like 2014 that no one can remember the reason for. And so, when you kind of start these things in technology, you kind of enter into different, you enter into legal, you enter into the cyber world, you enter into the business world, you enter into audit and reg. You kind of step into all of these different departments and there’s always walls and challenges that you have to solve to actually get these things into production. And that’s really difficult to do from tech, frankly. You kind of need somebody who has influence over all of these departments and they can break down silos and break down walls in all of these different, kind of, diverse functions to achieve the simple outcome of pushing this AI that we built into production. That’s why it’s so difficult.
Telling the success stories
Patrick: But by now, you’ve had some really good successes where you’ve gone from pilot to production a lot, maybe faster than you expected. Is there a good example of where, you know, where a company sort of adopted an AI-enabled solution quicker than you thought and sort of immediately start seeing results?
Milan: Yeah. So, one of the most recent examples is with one of our asset management clients and they have completely reimagined their investment process with AI. And so that’s from, you know, deal origination and screening to analysis to even in the investment committee and how investment decisions are made. And then also into, you know, once you make the investment, the management of it. And so what was exciting and impressive, and it takes courage to do this, is a complete reimagination of the entire value stream of an investment, not just a point solution at one particular, you know, not just the deal screen and not just the management of the portfolio, but the whole value stream. And you start to see these things kind of comparing because they build momentum. Like when you’re doing point solutions that are disconnected, they all seem like completely uncorrelated problems, and each one has challenges, and they’re all different teams, and nothing’s connected. But when you work around a value stream, you get the same people working on similar problems. It’s all the same context, a lot of the same data. You’re seeing real value compound over time as you move, kind of, downstream within a process. And that builds momentum. And that allows you to actually move quicker. So even though you’re increasing the scope, and it’s kind of scary at first because you’re completely reimagining kind of how you do business in a particular area, you can actually gain huge momentum and huge value by doing that because of the reasons I mentioned. So that’s probably one of the best examples that we’ve had recently of large-scale reimagination and, you know, but also moving really quickly.
Patrick: Right. And when you start working with a client in, like, let’s jump to Formula One. So, Cadillac, new this year in the Formula One space. When you go to them, is it the asset management story that you tell? Or how do you relate what you’re doing with your other clients to clients who are in a completely different industry with a completely different set of industry challenges?
Milan: Yeah, I mean, many of the underlying components of how to do this are the same. So where is the data? What are the decision points that you’re trying to influence? How do you operate today? And we work backwards from the business into the technology. And so even if it’s a net new industry, if it’s a net new business problem, a lot of the components of how you execute this are very similar. You’ve got, you know, large data sets that are either unstructured or structured. We’re trying to get signal from that data and we’re trying to influence a business decision to, you know, do something more effective, improve productivity, better decisions, lower risk. And that’s the same in insurance when you’re underwriting a risk or in a Formula One race where you’re trying to, you know, build a strategy of, you know, for a particular track or based on certain conditions. And so, from a first principles perspective, a lot of what we’re, a lot of what we see across these different industries are very similar. And, we have a proven methodology that we kind of apply to each one and it seems to work.
A new data paradigm
Patrick: And we hear repeatedly that the biggest stumbling block, at least initially, is the data. Is, as you said, it can be structured or unstructured. More importantly, is it, I know Accenture for a while was stressing sort of clean core, this idea of, you know, you get the most out of an AI-enabled solution when you’re starting with clean data. Are you still seeing, well, I’m sure by now most companies understand they got to get their data in shape before they can really leverage AI, but are you seeing that happening on the ground or is it still sort of fits and starts or still sort of depends on where within the business you’re talking about?
Milan: Yeah, I mean, data is obviously the lifeblood of AI. We see a lot of different patterns in customers, but I mean, it’s kind of the boring answer that nobody really wants to hear, but it’s where most of the real work happens and it’s kind of a big failure mode. So, I don’t just mean like cleaning up the data or moving it between systems, although there’s plenty of that. It’s also the governance layer underneath. And so, one of the key failure modes is, actually third-party data contracts, for example. So really unsexy, but, these were signed, years ago and often make up the large majority of the overall data that you consume as a company. So, some of the data is generated inside, a lot of it is procured and sourced from outside. And a lot of these contracts have clauses that say, hey, you can’t use this for AI, you can’t use this for analytics. And so in six months time where you eventually get a phone call from legal that says, hey, you’re not allowed to do this, that’s the kind of you know, data problem that bites you kind of, you know, down the line in these projects and is a big failure mode.
The other thing is kind of the dependency on a large scale data transformation. And so, I think that was the pattern a few years ago where, you know, data is the first thing you need to solve. We have data in siloed places. We have a lot of different source systems. Let’s bring it all together. AI is built in the cloud, so let’s do that. And so, you know, companies would set KPIs and say, hey, like how many terabytes of data do you have in the cloud this month? And it would just be numerator, denominator, and just every month, how much data have you pumped in the cloud? And what a lot of companies created was kind of a swamp, just in a different place. And it didn’t help anything.
Actually, our approach, and I think it’s gaining prevalence at the moment, is don’t, you know, don’t rely on a multi-year data transformation before starting AI. There is still value in cleaning up your data, and that is a separate decision. But from an AI perspective, there’s a need for a platform that can pull in the data from all the different source systems, all the different technologies, and all the different locations into an intelligence layer that can be leveraged from AI. So, you’re almost creating a new layer of data in the organization that specifically supports AI use cases. I think that’s kind of the new paradigm that we’re seeing these days.
Starting with the business process, not the technology
Patrick: Yeah, and then so I’m curious because now you’ve mentioned governance and talking about the risk part of things and a lot of the companies that I spent a lot of time with, these are the consultancies that basically say they do very similar things. They’re focused on risk. They have the governance, risk and compliance layer. They have the consulting and the business industry knowledge and all that. What’s sort of the difference between what you’re doing now and what, you know, what an Accenture or a McKinsey brings to the table?
Milan: Yeah, so our approach starts from quite a different place to most other companies. We don’t start with the technology. We start with a business process. And we ask three questions. Number one, where does that decision get made today? Number two, what data feeds it? And number three, how would we change that decision loop if we could rebuild it from scratch? And only then do we work backwards into what the architecture needs look like and what AI can support that. And we have a joint venture with Palantir. And so that partnership matters because the hardest part of doing this in a real company isn’t the model itself. It’s actually connecting the model to actual entities and workflows and permissions of the business, the context. Who can see what, you know, what counts as a customer, how an underwriting decision flows through five different teams. And kind of Palantir’s Ontology layer, which is what I referenced earlier, is genuinely the best tool I’ve seen for that. And it’s the difference between something that works in a demo and something that works, you know, on a Tuesday afternoon in live operation.
Patrick: And I could only imagine that, I mean, Palantir has had a huge effect on the rest of the consulting and the AI market in the sense that everyone is trying to imitate what Palantir does well with forward deployed engineers and all that, to some degree of success by some of the companies that we cover, but certainly not all. How long is the- is it a joint venture or just a go to market or what’s it- what is it that you’re doing with Palantir and how long has that been going on?
Milan: So, we have a, you know, our leadership has a long history with Palantir. We have a joint venture. We go to market together. One of the best things about Palantir is the Ontology and the data platform. And we selected them as a partner because it allows us to move quickly. So, our data scientists, our AI engineers can drop Palantir into a business, integrate into their data, wherever it lives, whatever it looks like, and actually start building within 30 days. And so, there isn’t this long, drawn-out process of kicking the tires on compliance, of building custom connectors into different technology source systems, you know, connecting to different models. Everything kind of comes out-of-the-box. And so, we can really focus on what matters, which is what is the value stream, what is the business process, asking questions, sitting with SMEs and users, and actually building products that help them with their day-to-day. And so, it’s just an accelerant for us. I think that’s the key differentiator.
Patrick: Yeah. And the accelerant is critical. I mean, I know everyone’s sort of maybe a little bit fatigued with all the hype around AI, but going- that’s getting stalled in the pilot stage and seeing lots of money spent on things that just don’t ever scale, super frustrating.
I want to ask you about, I want to sort of take a step back and look at a long view of technology, because you’ve been around a little while, to say the least. And again, I don’t mean that in a bad way, of course. But before that, I am curious. So, is there a problem set or a sort of a business challenge that you consistently run into with the companies you’re working with that you sort of haven’t seen? Let me come at it a slightly different way. We hear so much about the ability of the companies that we talk to bring AI-enabled solutions to bear on pretty much every problem that exists. And to be a little skeptical, there’s got to be some stuff that’s still too tricky, too knotty, too confusing. And I’m wondering if you’ve run into that like consistently. Is there a certain problem or kind of problem that you see repeatedly at companies that you’re like, you know, AI is not going to solve this anytime soon?
Milan: That’s a really good question. I mean, I think it’s a well-known transformation principle to- like AI is the last thing you want to do as a general rule in transformative process. The first thing you want to do is figure out like why are we doing this process? Like is it a control procedure that was written 15 years ago that no one’s ever really looked at that says that this team needs to write this report? Where does that report go? Who looks at it? How does it improve our business? So, you get rid of everything that’s not required. The second thing is you try and make it more efficient. So, the steps within the process, like are we wasting time? Are those steps required? Do we need to do three reviews? Does it need to have an external team, you know, complete that step? Can we do it, you know, within the team? And so, you improve the process just by, you know, doing classic process re-engineering.
And then you try and automate. So like rules-based stuff. So, if A, then B. And in many cases, those simple kind of automation steps, RPA is all that’s required. And then once you’ve got a clean kind of almost like pre-AI optimized process, you can then look at how to leverage trained systems and large language models to get to the next level. I think if you don’t go through that process, it’s going to be really difficult to get to an end state that is truly optimal. And you’re probably just throwing AI at a problem that doesn’t really necessitate it. That’s, you know, that’s key. And that’s hard work. Like, I’m doing process re-engineering, doing value stream mapping, you know, everyone wants to do go straight to GenAI, wants to go straight to multi-agent systems. But if you take the time to go through those steps, I think the value compounds.
Building safe flexibility into processes
Patrick: So then, so let’s get into the long view of technology by actually looking forwards, not looking backwards. Because a couple of times you’ve mentioned this idea that there are procedures or there’s a process that was put in place 15 years ago and nobody’s, you know, policies nobody’s gone back to look at again. How are you thinking about not laying down those same 15-year mines for the companies you’re working with? So that 15 years from now, somebody’s not coming in and saying, who is this Milan guy and why did he set up this policy here? Like, why are we cleaning this mess up? Like, how are you thinking about making sure that you’re not doing that yourself?
Milan: That’s a really good question. I think it’s tricky, right? You can only hit what you can see. And so, as we rewrite policies for the AI world, you know, we’re making sure that the kind of legacy software development lifecycle type paradigms are completely broken down and they’re fit for a world that requires, you know, fast iteration, experimental approaches. I think that’s the main focus for us at the moment. Like, I think that’s going to be, you know, the future. I think that’s going to be how businesses operate for a long time. And I think probably building in safe flexibility into these control procedures. And specifically what I mean by that is kind of the ability to experiment on model training and data that is real. Like historically, that’s been a big no-no. Like you cannot use real data to, you know, in a system until you’ve got through kind of 10 steps and you’ve done all this stuff and it, you know, the system is proven and hardened and all of that kind of stuff. But that’s just completely backwards in the world of AI. Like the first thing you need to do is look at data to see if there’s signal. So, if you can’t access the data on day one, then you’re kind of in a catch-22 situation. And so, I think built-in flexibility into these control procedures built around an AI world is probably going to, you know, probably going to be valid for quite a while.
Patrick: For a while, right. Yeah, that makes sense. And I think back to the number of times that I’ve sat with, consultants who have said, they’ll spend two or three weeks trying to just get access to their client’s data because within the client, there’s all different kinds of people that own the data and have to sign off on the data. And so, they’ll start an engagement and three weeks later, they’re still waiting for access to the data, which is insane. And that just doesn’t, that can’t fly. That can’t be true anymore in an AI world. It just can’t.
Milan: That’s exactly right. Yeah. Squeezing that timeline between, you know, how quickly can you get real data in the hands of our data scientists is a huge accelerant. That is probably the number one failure mode of, you know, especially in the early stages of these projects, is just how long it takes to get access to data. And again, going back to CEO led, sometimes data lives across departments. And it’s not just one tech person or one business person that needs to kind of approve access to this stuff, it’s many. And then who do you get to kind of, you know, knock heads together to make sure that happens?
Patrick: Yeah, it all comes back to, like you said at the beginning, that question of like, who actually makes decisions and, what data are they using to make those decisions and can they make the- are they empowered to make the decisions?
What makes AI evolution different from previous technology evolutions
I do want to look backwards a little bit and sort of put the longitudinal view of technology in the context of where you sit today, could you, 10 years ago, when you were with Accenture doing advising, doing tech strategy and advising clients, did you imagine sort of where we are today? I mean, how and how much have things, I guess I’m answering the question, no, you didn’t, maybe you’ll say yes, you did, but how much of what you’re seeing today is sort of, is more than just an evolution of RPA, more than just an evolution of, you know, even blockchain or some of the other technologies that were so hot and emerging not that long ago, and how much of it is just truly revolutionary, truly different from what you expected 10 years ago?
Milan: Yeah, it’s a good question. I mean, some of the evolution and some of the technology waves that I’ve been a part of over the last kind of, you know, 15, 16, 17 years have been almost by accident. So, I started off right at the bottom of the stack as a network engineer. And so, I was in hardware. Then all of a sudden, kind of the early parts of last decade, all the infrastructure people at Accenture just became cloud people. They literally just changed our job title. So, on Friday, I was an infrastructure guy, and on Monday, I was a cloud engineer.
Patrick: Right.
Milan: So, some of these things just happen to you, some of them you get to make conscious decisions on. But the thing is, AI isn’t really comparable to cloud or mobile. Those are infrastructure shifts. So, you could delegate them because they didn’t fundamentally change who did the work or how decisions get made. AI does. It changes the shape of jobs. It changes which decisions get automated and which ones don’t. It changes headcount assumptions. And those are CEO questions by definition. You can’t delegate them without the whole thing turning into a science project. So, I think that’s really what’s different with this evolution is this isn’t just an infrastructure shift, which, you know, it isn’t just a kind of like cloud, for example, was a paradigm shift in how to optimize infrastructure and workloads. This is really introducing an intelligence layer into your organization and how you kind of leverage that is a key strategic question for all these companies.
Patrick: So, if you had to sort of balance like where we are now, where we’re going next on exactly that point of that sort of it’s- that intelligence layer is a key strategic part of any business. Are we at the point now where if companies have not already defined their AI strategy, already begun making investments in how they’re changing their business with that intelligence layer, that we’re going to see those companies left behind? I mean, is there a- do you see in looking at the companies you look at and then thinking about them in the context of their competitive landscape, are you seeing where that intelligence layer and that investment in the strategy now is going to make all the difference over the next, you know, five, six years to separate the companies that actually will succeed and those that won’t?
Milan: Yeah, I think so. I mean, it’s almost like, I can’t remember what the saying is on bankruptcy, where it happens slowly, then all at once.
Patrick: Right, yeah, a run on the bank, yeah.
Milan: Yeah. That’s it. So, you know, it’s, I think that’s- that day is coming soon, right? I think this isn’t, you know, typical, you know, productivity enhancements where you kind of take, you know, low single digit percent off your OpEx or whatever. This could be an enormous transformation shift in how companies operate. And so, you’re starting to see early signs of that where some companies are kind of completely re-imagining how they do business. There’s a lot of froth in the market at the moment. There’s a lot of dollars and capital flowing into point solutions that are promising big things. And I think at this point in the game, I think you need to have, you need to be way beyond just having or thinking about your AI strategy. I think there needs to be, you need to be pretty far down the path in fundamental business transformation using AI, whether it’s other traditional competitors that are just moving faster than you, or whether it’s net new companies that haven’t even been built yet that are going to kind of bite you at the heels and then ultimately take over large parts of your business. I don’t think it’s, overdone to kind of pose those questions to businesses and understand, how, like what’s the best way to defend against that?
Technology evolution vs business model evolution
Patrick: Yeah, and I think back to like how we, for years, we talked to companies about digital transformation. That was the buzzword before AI. Everything was about digital transformation. And we would hear these use cases where companies and their consultancies would tell us all about the digital transformation. And at the end of the day, it was, they adopted S4HANA, or they adopted Salesforce or something. And the business hadn’t changed. It’s just that the underlying technology had changed. Like their technology environment was different and improved and better, whatever, sure. But there wasn’t a business model change. And I think what we’re seeing now, and I think of what I heard you say is that AI is different in that way and that it’s allowing for a transformation that’s business model that’s fundamental rather than just technological.
Milan: Yeah, I think that’s right. You know, I think the saying back in the day was no one gets fired for buying IBM, I think. You know, you can kind of see a similar thing happening now with the foundation, with the large labs. Everyone’s, you know, the obvious thing to do is just go buy a chatbot.
Patrick: Right.
Milan: And introduce that into your business and, kind of hope for the best. And I think that is, that strategy is- and there’s been some recent research on kind of how, what the impact of that is on organizations. So, for example, I think there was a recent article in HBR that talked about how chatbots don’t improve productivity for work. They just increase intensity. And so, everyone’s kind of experiencing task expansion, blurred boundaries between work and non-work and a lot more multitasking. And the intensity of work is just increasing. So, the underlying process hasn’t been transformed. We just dropped the chatbot on top of it and hope for the best. And so, I think, there’s emerging research that’s kind of calling that out as a finding. I think we’re starting to see more and more of that. And I think that’s going to continue to point towards the need for underlying process reimagination.
Patrick: Yeah, it’s funny you use the word intensity. We’re about to roll out some new research that we’ve come up with that looks at what we’re calling human intensity reduction index. So, the human intensity reduction index looks at how well a company can continue to make profitable dollars with fewer people. How do you reduce the intensity of the humans that are going into, you know, and we’re looking primarily right now at services firms and services- consulting and IT services is a people business, but the sort of hype and promise around AI is that you’d be able to do more with fewer people. And so, the question is, can you actually reduce the human intensity within your organization and continue to make as much or more money as profitably as before.
Milan: Yeah, I think that’s right. And I think that’s, you know, this is all a really important signal for executives because it tells you that just deploying AI tools and waiting for productivity to show up in the P&L is naive. And if you don’t redesign the work itself, AI just becomes a tax on the people that you already have. And so that’s actually one of the reasons we’re so focused at TWG on embedding AI into core processes, rather than handing work as a chat bot and just hoping for the best.
Patrick: Yeah, that’s a really, that’s just a great way to think about it.
Following interesting problems and interesting mentors
All right, I have Milan, I’ve got one more question, and it’s a question I’ve been asking everybody this season on TBR Talks. And it’s about, to give you the context for the question, our youngest child, our daughter, is going to graduate from college in a month. And she’s going to be so happy when I no longer talk about her on this podcast. But anyway, she’s graduating, she’s 22, and she’s at that point in life, you know, 22 years old, you’re finishing up university and you look around, you imagine what you could do with your whole life, what you want to do with your life, where you want to be in the world. And talking a bit, you know, I know what I was doing when I was 22, but I’m really curious. So here you are now working for TWG AI, you’re doing this incredible work with these companies like Cadillac and Formula One and the Dodgers and all that. And when you were 22 about to leave university, did you think to yourself, I want to be an AI- I want to help companies use AI to change their businesses? Is that what you thought you were going to be doing at this point in your career? Or what was 22-year-old Milan thinking he was going to do?
Milan: That’s a really good question. I think the story I told earlier around how things have happened to me versus being conscious decisions is probably a theme throughout my career. I’ve been led by one, really interesting problems and two, awesome people. And so those are my two kind of like decision points and in whatever I do is do they solve one or both of those things? And so, there’s no planning involved here. It’s really reactive. But I think 22-year-old me is just asking open questions and being curious about the world and being curious about people and how companies work and kind of asking the questions that maybe some people are kind of afraid or embarrassed to ask because they might make you look stupid or uninformed or whatever. But I think that was a principle that I held. That’s how I do business. And I think you can kind of operate without a rigid plan if you just, kind of, ask questions, follow amazing leaders around the world and look for interesting things to work on.
Patrick: Yeah, that’s, I love the follow interesting leaders part of it too, because I mean, being open-minded and asking questions others don’t want to ask and looking for difficult problems is great, but man, you got to surround yourself with good people because that makes all the difference, doesn’t it?
Milan: Oh yeah. I mean, look, I was at J.P. Morgan and Jamie Dimon evangelized the apprenticeship model. It’s why he pulled everyone back into the office probably earlier than a lot of companies did, because being around colleagues, being around leaders, learning behaviors, good and bad, is really what shapes you professionally and personally. And finding those people that have followership and are inspirational and teach you continuously is really hard and it happens by accident. There’s no, you can’t just go out on a Monday and say, I’m going to go like build a really deep relationship with a phenomenal leader. It just happens over time. It happens by accident, happens through fortuitous interactions. And you got to, kind of got to put yourself in the game to take advantage of that.
Patrick: Yeah, 100%. And that the apprenticeship model has been so important to consultancies, to so many different kinds of companies that yeah, I’m glad. I’m glad we’re back in the office as much as we are now.
Final thoughts
All right, I lied, I’ve got one last question for you. I have to ask, having started to watch and follow Formula One a lot more this year, how many years until Cadillac is going to be the, you know, the manufacturer’s champion in Formula One?
Milan: *laughs* Yeah, so Dan Tarras would probably have a very good answer for this, the head of motorsports. As soon as possible, Patrick, is my answer.
Patrick: As soon- so, you’re not going to give me, it’s not going to be this season, it’s not going to be next season. Well, it can’t be this season. It’s too late already.
Milan: I think what the team are doing, what the team are doing back there from a standing start, you know, building a car and getting it on the track with all the regulations that exist within Formula One is a monumental achievement. I think, you know, step one, step two is winning races.
Patrick: Yeah, and it’s incredible. And this year they’ve got, they got the two races got cancelled, so now they have more time to look at all the data and make all the tweaks and stuff. So, it’s going to be a really, it’ll be, it’s going to be a good season. We’ll see how Cadillac ends up at the end of it.
Milan, thank you so much for coming on. This has been really- I’ve enjoyed this conversation immensely. And you’ve said some things that I think I’m going to- when I share this with my colleagues, they’re going to, there’s going to be a lot of, well, we may have to follow up with another conversation once my colleagues hear this. So, this has been really, really great.
Milan: Thank you, Patrick. Great to be here.
Patrick: Excellent. Thanks.
Tune in next week for another episode of TBR Talks.
Don’t forget to send us your key intelligence questions on business strategy, ecosystems, and management consulting through the form in the show notes below. Visit tbri.com to learn how we help tech companies, large and small, answer these questions with the research, data, and analysis that my guests bring to this conversation every week. Once again, I’m your host, Patrick Heffernan, Principal Analyst at TBR. Thanks for joining us and see you next week.
TBR Talks: Decoding Strategies and Ecosystems of the Globe’s Top Tech Firms
Join TBR Principal Analyst Patrick Heffernan weekly for conversations on disruptions in the broader technology ecosystem and answers to key intelligence questions TBR analysts hear from executives and business unit leaders among top IT professional services firms, IT vendors, and telecom vendors and operators.
“TBR Talks” is available on all major podcast platforms. Subscribe today!
Technology Business Research, Inc.
Technology Business Research, Inc.