Prefer audio? Listen anywhere
Three years ago, boards did something unusual. They didn't just push management to act. They pointed at a specific technology and said: go.
That almost never happens. Boards govern, challenge, approve. They rarely lead the charge on a directional bet. But every board member had tried ChatGPT on their phone and the reaction was visceral. Pär Edin was sitting on the KPMG US board while leading their AI go-to-market. Both sides of the table. And what he saw was a pattern now playing out across more than a thousand board members: full throttle and full brakes, at the same time. The question most boards still haven't resolved is whether they're governing a company through a technological shift, or still governing the company that existed before it.
🎙️ Pär Edin
Pär Edin is a Principal at KPMG who spent five years on the firm's US and Americas board, chairing the operations and technology committee when generative AI hit. Not advisory. Votes. He's since worked with over a thousand board members on AI. What makes his lens unusual is the dual view: boardroom and management floor, simultaneously. He knows exactly where the two misunderstand each other.
🔥 Key Insights
✅ Use cases are a sign of immaturity
The more a board asks for specific AI use cases, the earlier they are on the curve. Advanced boards focus on the haystack, not the needles. What happens when every player in our ecosystem adopts AI at the same pace or faster? That outward-looking question separates governance from curiosity.
✅ The power play you're not running
Think of AI like an extra player on the ice. Would you rather play short-handed or in constant power play? That extra player doubles in performance every three to six months. A competitor with a year's head start sounds small. It compounds.
✅ No company transforms faster than the pace of trust
Employees afraid to use AI because an error might cost them their job. Shadow usage exploding because official tools are two generations old. Top talent leaving because the policy signals a company standing still. Not three problems. One trust failure.
✅ Demanding ROI right now is counterproductive
The circular argument plays out everywhere: management says the board needs a number, the board says we asked for a conversation about value creation. Forcing precise ROI assumptions that don't exist yet wastes energy and produces fiction. Most AI investments so far were approved on strategic intent. That's not reckless. It's rational.
✅ More work, lower cost. The productivity flip.
Traditional productivity means fewer hours per task. AI inverts this. When agent hours approach zero marginal cost, you can triple the volume of work while cutting what it costs. A European telco went from targeting 30% improvement in offshore finance to automating the function entirely and bringing it home. But only because someone redesigned the process from scratch.
✅ The board that beat every human
An INSEAD experiment in HBR: a fully agentic AI board versus human boards, blind-tested by human jurors. The AI board scored higher on every dimension. Pär isn't saying boards get replaced. But if you're asking your teams to rethink which work is uniquely human, at some point that question comes home.
Read the full transcript
Last time we met, we had a really good discussion on how AI is, um, kind of bleeding into to every day. Uh, that was a recording that we did in Swedish, and this time I thought I'd leverage your extensive knowledge when it comes to AI and the boardroom. I thought it might be a good starting point for you to kind of like, how come you can speak to this topic specifically.
What is the work that you do together with boards?
Yeah, absolutely. Yeah. So I'm principal with KPMD, and as, uh, part of my role, I was elected to the board of KPMD in our US firm and our regional Americas firm, um, for, for five years as a standard term, if you'll, and I rolled that off a little bit more than a year ago. So, uh, I happened to be on that board at the time when the generative ai, sort of revolution really happened.
And I was just looking back to it. It's almost exactly three years to the day, um, when most of our clients and corporations sort of woke up and started. Asking questions about this. And, and so March 23, when, when Chat GPT moved to the, uh, GPT four model, I think was, was really when, when, when our clients woke up, if you'll, and we had a period there of, I mentioned it on the previous podcast where, you know, in, in eight or nine months, we had, you know, many, many hundreds of, uh, of, uh, of clients, uh, reach out and actually a large share, I would say majority of them said something to the effect, uh, that, you know, our board is on our back.
Our board is asking questions around what are we doing with this technology? Every board member and every management had tried it on their iPhone or had, you know, somebody teach them how to to use it on their iPhone, and they were blown away. That was really one of those rare moments where boards were really pushing, um, management to explore, uh, this, I thought that was interesting to be on the board at that time.
And you've been on doing transformation work for, for 30 years, how can you remember a time where the board has been as pushy to get a transformation started as this time?
No, I can't. And it was quite unexpected. I think most professional services providers found that, that, you know, that there was a quite different, uh, demand, if you will, and not all of it, but a lot of it came from the initial reaction from, from, from board members. Um, and, you know, uh, a lot, a lot of that reaction had to do with are we piloting?
What are we doing?
We, we fear we're missing out. I think was, was, was a big fiber, but it didn't go very long until the next question came. So my, my, my, my board is now asking me to down,
Yeah. Compliance issues. Yeah. Yeah,
So I, I think there was more than a year of sort of like full, full throttle and full breaks every back and forth.
Right. And, and I, I think that both of those aspects are great. So from a being on the board, and I was at the time also leading our go to market efforts for, for AI in the US and how we are helping our clients, I could see it both from the management perspective and from the board perspective. Uh, so it's really an unprecedented time in that sense.
And we also made, uh, some very big decisions back then that, that, that, um. We're a little bit, you know, early. So we, we changed our corporate structure. We, we created a new role called, uh, essentially the chief AI officer and took one of our most senior, uh, leaders outta a very big revenue role and to, to drive this transformation.
Um, luckily we had already a committee, the operations and technology committee that I happened to chair at the time. So we had a little bit, uh, more of a structural advantage to be able to handle some of the, time required both in the full board and the committee, et cetera. So there were a lot of things going on there that really, and most importantly, we felt that we needed training as a full board. Um, and therefore we developed our internal training. We have the six hour, three hour, one hour sessions that we did for ourselves that, that we are now, uh, you know, taking to to our clients.
And how many boards have you trained, uh, in general when it comes to board work, because I know that you do a lot of board work and specifically now on ai.
I mean, training is a big word, but I've interacted with more than a thousand, uh, board members on the topic of ai. Uh, in, in, yeah, in the last. Year, at least two years. And that goes from one-on-one coaching sessions, uh, with clients or, or other stakeholders. It goes to one of many, uh, board sessions where we get called in to talk about the trends and how to think about that.
And even all the way up to, uh, you know, large webinars, et cetera. I did with, with Eric Brynjolfsson that you also had on your podcast earlier, and we talked specifically for an hour with, with a thousand board members, you know, uh, dialing or, or at least, uh, on, on the virtual line about the topics of AI and, uh, how you can think about that.
So, yeah, it's been a number of discussions. I, I, I caution anyone to say that they're an expert on anything to do with ai, given how quickly it evolves. And I certainly don't, don't sort of tend to be that, but I, I have learned a lot and, and, uh, I have a lot of thoughts around, you know, what, what seems to be working and not working from a board perspective.
Fantastic. And I, I completely agree with the, with the fallacy of trying to be an AI expert, it doesn't seem to matter how much you, you try to keep up, you're, you're always gonna be be behind with the, with the speed of change. Right. Um, I'm wondering, could you give us some type of, you know, state of the union for how mature is the average board when it comes to understanding the fundamental changes that AI introduces into our business?
I mean, first of all, there, there, there's obviously wide spectrum, um, wide spectrum, uh, that depends on a lot of things like the, uh, you know, the, uh, the original composition of the board. So remember, I mean, t is only, or generative ai and now Agentic AI et cetera, has been around for less than three years, right?
In, in sort of a corporate sense. So most board members were already elected to the board before. there's an increasing number of board members that will, were elected to, uh, increase the level of competence. And, and, and, you know, AI seven as of the board. But most boards are, you know, governing this with their board that was elected before.
It must be a very difficult like recruiting task for boards as well to even find relevant new team members to hire because it's so new to all of us, right?
I see two types of approaches there. The, the first one, the more sort of reactive one is that get somebody on the board that has technical competence, that has built something with AI that understands what AI can do when, and the limitations around, et cetera.
And I, I think that certainly has value. I, I, I sort of lean more towards the second category that says, you know, let's have people on the board that understand how, um, significant disruptive technologies can impact companies in the world around them, and how the work of the board needs to evolve and focus in that world.
So people that have been part of major, uh, digital transformations have seen technology, innovation driven growth or, or disruption. Uh, I, I, I tend to look a lot towards, uh, valuing that type of competence we're looking for, uh, for new board.
Yeah. Higher than specific AI technology skills. Yeah. when you meet a new board, what are the way that you'd introduce how their function as a board changes needs to adapt as a consequence of ai?
I think there, there are five topics that, that every board to some con, you know, to, with, with some, uh, changes are, are dealing with, um, uh, at any given time.
The first one is just understanding these technology trends. 'cause they're moving so quickly, not so much exactly how the AI is built, but what AI can do and what implications that can have for, for the company. And in these three years we've had several iterations of that. So it's not, you know, it started with generative ai and then we had the copilots and we had embedded software, then we had agents agent system.
Now it's swarms and physical AI and, and you know, a GI and um, SGI and all these acronyms, right? So no matter how quickly you move every three to six months there, there is something else. And on top of that, the technology itself is doubling in price performance every three to six months or so. So whatever you thought you were dealing with a year ago is now, uh, significantly more, uh, powerful and potentially helpful and also potentially, potentially more risky.
So that's sort of first one around understanding the trends. I think every board, uh, kind of feels they have some degree of, but is very varied in the board. And I, I recommend having a dialogue around to level set so that all of the board members are feeling at, at that they're on the right place there.
I would imagine that most boards have or board members have heard about all of these terminologies. Agentic, I think is a brilliant question. Uh, or example, but there's this sense of, yes, I've heard the hype, but I haven't really seen it yet in, in its full implementation or in my industry, and, and how to kind of separate hype from reality and what we should push for or move towards.
How do you hold that discussion in a good sense? In a board?
yeah, the first indication of maturity is whether the boards ask for use cases. There's a little bit of a personal, but the more they ask for use cases to make it real, to make it credible, the the less mature they are. Um.
the more mature bores focus more on the haystack, how big it is, how quick, quickly it's moving, uh, than the individual needles. If you'll,
That's a little bit counterintuitive, I would imagine, like it's a signal of interest asking for use cases.
It it is, but early interest. Right? So, 'cause the next level question becomes, well if this is a practical thing that some person can save some time with, is that relevant for the board or not? Does it, does it really move that needle, uh, on revenue or cost or risk or ability to maneuver in the market or the market around us?
If no, it's not a board topic, right? If it could disrupt our entire business model, if, if what others can do and what we can do, then it certainly is an urgent, uh, sort of board topic if you'll, so that is one level of, of maturity. If you'll, the more. Implication questions. The more, bigger questions around, um, uh, not only productivity, but also growth, innovation, um, quality and, you know, changes in business models and market around.
I think the most advanced boards are also asking, we are only one of many players in our ecosystem. What is going to happen if every one of our customers, competitors, uh, suppliers regulators, et cetera, if they are adopting AI at the same or even faster pace, right? This outward looking, I think, is an indicator of the more mature boards right now that are thinking ahead.
Uh, if you'll, there's also this balance between risk and opportunity and, and the nature of the board is, is typically risk minimizing in many cases. I always encourage and look for boards that what means at, at least spend equal time thinking about how could we use this in our offense? How could we disrupt somebody else, right?
How can we grow, um, in addition to how can we drive productivity and how can we defend against both, uh, lawsuits and, uh, you know, brand infringements and all that stuff, right? So this balanced view, if you will, I think is a ma indication of the more mature boards.
It interesting because, I would imagine it's quite a difficult conversation to have in a board, so one part of it is.
Do we dare to fundamentally challenge the, the underlying assumption of our business model? So I think there are to, to your point of what happens if all of our customers, all of our regulators, like our whole ecosystem adopts these tools. So if I were to be in, in recruitment for example, I would look into the recruiting process and really look for what fundamentally is the value that only a human can bring?
And, and how much would probably all of our cus customers be able to do themselves? That used to be the value drivers that we could deliver and, and charge a premium for. And like, do we dare to have a truthful discussion around the, the kind of next level consequences of, of AI bleeding in, right? I don't think that many companies would need an HR specialist externally to write their, uh, you know, job descriptions.
Something that they might pay for previously, but now it's just available in a chat model. So, so one thing is like the courage of, of really looking into what fundamentally changes and, and dare to speak those things. And the second part is, how quickly do we feel that we should push for this change?
Because you also have, like we, we do have our current market that is still willing to pay for our old services. And this is interesting for all transformation. How quickly should we force a pivot into where we think the market will be in in three years
Right. And, and this, you know, puts the, the spotlight on one of the biggest challenges of being on the board versus being in the management. The, the board is not the management, the board's role is not to take the decisions reprimand, you know, uh, direct in that sense. The board provides governance and oversight. it oversees the management's decisions and at the end of the day, has the ultimate responsibility of hiring and firing the chief executive officer, et cetera, in, um, it also, you know, approves major decision and you, you literally had to vote on some of the bigger decisions. But a lot of the discussion is board on the board is about asking the right question.
Making sure in our oversight that management has considered, uh, many of those aspects. And if management has considered those aspects and came to something that you might personally disagree with, that's not your role as a board member to opinionate on. It's the process that leads you to their, you know, in to, to reflect the shareholders, you know, longer term value, if you'll, so particularly in those topics, it's if you have old, uh, consultants or, or, or managers, uh, like me on the board, you have to unlearn, uh, you know, the recommendation and direction skill.
Relearn the listening and asking the right questions, skill. Um, and, and particularly around those business model questions and the bigger questions that, that, that comes into light a lot. There's, there's nothing more sensitive to a strong CEO than the board leaning in too much hands-on, right. At the same time, it's nothing more risky for the shareholders and the board that doesn't lean in at all and, and stays too far, you know, away from, from, uh, uh, challenging or, or, or, or probing and asking the right questions.
How, how do you find the right balance in, in practice on those?
With difficulty, I would to say, and you know, it, it is both, uh, in your board training and in your experience as a board member, you, you get pretty clear signals when others' opinion is that you are wearing too far away from, from being hands-on or being too hands on. If you'll, um, it's also the role of the chairman or the lead director to direct the dialogue and manage the dialogue in such a way that we stay within those lanes.
But it is by definition somewhat of a gray zone.
And that gray zone is where a lot of the risk and a lot of the value, uh, is. So I think, uh, I, I think the mature boards will both have experience in dealing with it and also have some, um. Generosity in somewhat, somewhat passionate board members or or management members going a little bit astray and then bringing them back in because those outlier voices can be really, really critical.
Particularly I, I mean the importance of, of governance, uh, and particularly strategic governance. Long-term governance increases the more, the more uncertainty there is, uh, the more rapid development there is, and the more upside or downside there is of getting it rights in the world of ai.
Yeah, absolutely. It feels like such a, a important moment. I, I reflect already 'cause I, I'm too young to really feel that I was part of the digital revolution, even though of course I've, I've been on, on, in, in the kind of digital space in, in my whole career. But this is the first time where I feel that I'm in a management position through a big technological leap.
And it, it really feels like one of those pivotal moments where my kids will ask, uh, you know, in, in 10, 15 years, what did you do? Yeah, exactly.
understanding those trends and what it means from a board is, um, is really critical. I think the second topic that both most boards deal with is how big is the opportunity? And I always like to talk about that. Second, because most boards naturally go to how big is the risk.
the way the brain works also we talked about in the last episode is maybe that you are, you are more sensitive to risk than opportunity. You're more sensitive to loss than gain. And therefore, uh, humans and all board members are human. Need to, in my view, push ourselves a little bit towards looking for the opportunity as much as the risk that that is within our mandate.
And that could be debated and I think a lot of my board colleagues might, might favor more the risk side. But that's also the composition of the board should favor, uh, may blend of more opportunity focused people. And on the opportunity side with ai, we've always struggled with this inability to quantify the opportunity,
Yeah. 'cause it's, uh, hypothetically endless, but it's also very hypothetically endless.
Yes. And there, there, there is certainly enough, um, confusion and difference of voice in the market and in different reports around how, you know, everything from how world changing it is to how it's not happened and how it's overblown, et cetera. And any board member with, you know, takes input from the market would rightly be confused.
What, what is it, what is it for our company today? What could it be? Withholding it back. So this ability to quantify the opportunity that's real today, uh, let alone could be real tomorrow, is, is this important?
Do you have some frameworks to introduce? Uh, like how, how would you go about to, to the average board to start quantifying this?
so there are at least four or five different components to the value equation.
Is it the, the most. Tangible and easy to measure, at least estimate is the labor productivity. So
basic time freed up, um, reinvested in the business. Uh, and how, how much is that? So it's a time perspective. I also talk with board members. I think think about generative AI in its early instances as a time machine, not bringing you backwards or forwards, but creating more time.
Just as you create more investment budget, you create time that can be reinvested. It's not really a good tool to, um, do headcount reduction. And this is the first fallacy. I think if you, if you're, if you're deadbent on using that productivity for headcount reduction, you, you will only be able to, in our experience, to get a very small share of the full potential.
The big potential comes from reinvesting, uh, time into the business to do other things, drive growth, uh, innovation, time to market, quality of the processes, even employee satisfaction and so forth.
Hmm.
And for that there are tools. We have some others, have others, you know, to quantify that if you'll, and, and and that's just a portion of the opportunity.
But that opportunity in itself, um, at least theoretically is very significant. We estimate the white papers, we put out between four and 18% of EPTA potential, uh, for the time freed up, reinvested in the business, considering, you know, the salary cost of an hour depending on where you sit in the world, which function as a measure of economic, uh, uh, opportunity.
There are a lot of ifs around the way, and most companies there are still in the making the tools available with some training and some type of incentives. Uh, many of them in my view too negative and, and sort of forceful, but ways to stimulate that. And we have, um, we are seeing in pockets of the world, some companies, some parts of companies really getting that.
And in many right now, they're struggling with sort of realizing that benefit. But to give you some examples, um, uh, we've measured disability for, for 30 companies that employ a hundred thousand people in, in a hundred companies. And at least in portions of those companies, we're seeing replicable 22% productivity, 54% quality improvement in the work that they're doing.
And 44% employee satisfaction uplift. So, you
know,
it's early days. But it gives comfort that when done well, you can, you can rely on those, and that's from basic tool usage, you know, some kind of, uh, uh, GPT model, copilot model or, and simpler task, like not the fancy future, uh, um, AI, et cetera.
you need to have a conscious thought around the reinvesting of the time into something, even, even higher value. I think that's a piece, at least to me, more looking at it from the, the management team perspective that we, many management teams doesn't really have an idea of what do we reinvest towards?
It's more like we're looking to save up time.
Yeah. And this, this goes into the sort of like, uh, well, how you reinvest that time depends on your corporate strategy. What are you trying to do? Where are you trying to grow? And how could that evolve? Right? So if you don't have a corporate strategy, you don't know where to reinvest it, and if you don't have a deliberate transformation effort that encourages people to invest in, in the direction of the company, they will invest that in things that they think are the most helpful to them and or the company not always the same.
So,
No. Exactly.
um, both the approach to which companies unlock this usage, the, the adoption, uh, which is the biggest barrier in my view right now, uh, and using proven behavioral science approaches, not top-down force and shame and blame, uh, which doesn't seem to be working, um, is a prerequisite. But then the, so from a board perspective, you ask simple questions like, okay, where are we?
What share of our employees have access to a decent model? Right? How much are they using them? And it's not enough to have a, have you ever used or are using monthly? I think right now you should be measuring how much, what is your daily or frequent use, um, to, to get any impact, right? But then, okay. How are we guiding, uh, that organization to reinvest?
What do we want to achieve? Do we wanna grow more? Do we wanna increase our market? What, what are we doing with it? And, uh, there you quickly get into management questions, right? So you need to be careful as a board member, not to get to need too nitty gritty and into it. But that high level question on where are we on the adoption journey to get labor productivity?
So, so for me, again, looking at it more from, from the operator's lens, when I meet a company that says, yeah, we have copilot license. I, my, my like, gut reaction is great. You've gotten absolutely almost nowhere, uh, because you're at model levels that, that aren't really the, the, the, the latest versions and you probably haven't done that many advanced new cool things with it.
I get that question a lot and I, I think, you know, the every three months, the answer may change what your preferred, you know, highest performing, uh, model is. Yeah. And that will continue to change. So from a board level perspective, I, I, I think it's, it's, it's less that than sort of what is the, what, what is the age of the model, right.
Um, because there's also risk here. If you give employees something that's older than one or two generations back, they will use their private.
Yeah, exactly.
And that creates not only that creates risk and lack of transparency. So, you know, the maturity, the age of the models, you know, how many have access to a later model, et cetera, I think, I think is a, is a great indicator of providing the right tools.
But, but even more, you know, the models already three years ago were good enough for, for board members to wake up, right? And think about it. So any model that's out on the market today is capable of, of getting you those numbers, right? 20 ish percent time, uh, even almost without any data or invest because of the basic grunt work of knowledge workers that are reading, writing, synthesizing, right?
So. Um, so I think that, that, that potential is, uh, is, is clearly there and the boards need to, um, uh, at least ensure that management is going after that with sufficient, uh, uh, you know, vengeance. But beyond that, it's what you do with that. You know, we're in, we're in the period of Stanley cup here and I used to play ice hockey.
I tried to explain to a Canadian, uh, CFO and board member how, you know, how to think about this. And it's more like, you know, would you rather try to score the same number of goals with four players against a team of five? Or would you constantly want to be in power play and have one more core
player on the ice without
take,
yeah.
gets it, right, yes, I'd like to be, and that's what AI can do.
If you are the first to get extra player on the ice, then, then, uh, you know, you, you, you have an advantage.
I love that analogy because that's exactly it. I've been, I've been struggling for a good analogy to, to kind of exemplify the compounding, uh, like advantage that the AI gets you if you get into it early enough. I see on, on clients that I advised that we started the work with a, a year ago, like what they do versus the, the clients that I begin with now and the, and the amount of, of operators within these companies that, that truly make completely new things because it takes the time in the company culture to, to really understand the, the potential of where our market is going, how this fits into it, how we practically build things.
Uh, and like over a year, you can make massive difference within a company, but if you have a company, so even though it's like. Fast, which is interesting. You can say, yeah, they're only a year ahead, but like that year, it's hard to catch back up to a competitor who's who has a year's Head start on you.
'cause they will continually have that year's Head start.
Exactly. And then if you play the, the, you know, the, the, the likeness forward, right? That player, the extra player on the ice, uh, doubles in performance every three to six months.
Yeah, exactly.
You still have to pass the puck, right? Between the current team and them. Uh, but you want a few of them and you want more them on your side than on their side, right?
So it, it's a lot of that going into it. So you gotta find, uh, you know, ways to think about that. But again, that is the, you know, the lower bound of, of, of the opportunity. And that is big enough for most boards to approve the level of investments that we see. So we're measuring intent to invest every quarter, right?
And now it's somewhere between a hundred and 200 million per year for large Fortune 1000 or 500 companies.
Uh, I put that into a percentage number? Because I do think that would be valuable for the, the average board member. Like what, what does a significant or, or, or like a realistically good AI investment, where does the money go? Is it only to tech vendors or how much is training and other stuff? And, and how much of share of revenue would, would the investment be, would you say?
this is usually intent, if you will, and, and in the big scheme of things, that's sizable to matter, but not earth shaking. Most companies, most companies didn't plan a budget for ai, you know, one or two years ago it took it mostly from the IT budget, and now they have, and they, they redistribute that the, you know, investments, the way they see it fit.
So it's not yet sort of, you know, bet the firm type of things for most user companies. I'm not talking about the
builder course the like. model builders know.
my personal guidance is if you look at the quantifiable, lower end of the opportunity, take that labor productivity opportunity, right?
And, and let's say that that equivalent to four to 18% of EBDA half of that probably requires more advanced agents than we have today. So you're down to two to nine, and about a third of that is the more low hanging fruit that you can certainly go after with existing technology. So, you know, you're looking at a 1, 2, 3, 4%.
Annual EBTA potential for most companies varies by, you know, function, et cetera, but, uh, whatever that number is for you, take 10 or 20% of that and, and make that your budget. So if you are investing less than 10% of your annual low end productivity, you're probably under investing and you're gonna get that extra player on the ice slower than your competitors.
If you are investing more than 20, 30% of that, that could be make total sense for the, for, for the period you're in. Uh, but you also might be, uh, you know, uh, ahead of os and there are no really hard mention, but that's my personal, when I look at a company's transformation and or when, when, when, you know, talk to board members, I ask them to kind of think that way and it's very rude and rudimentary.
But, but I, I believe it's accurate.
How diligent should you be with, um, you know, business cases, return on investment, uh, scenarios because it's also a transformation that you can have to invest into a little bit speculatively, because a lot of the innovations that you'll end up finding won't be the ones that you're expected to find, I would assume.
I mean the, the, the ROI is really, really difficult right now. I mean, let alone the r the, the, the urg, right, that we just talked about. And, and the, i the investment depends a lot on how you do it. And as you said, what is your starting position? Do you need to completely revamp your IT infrastructure, get access to models, data, et cetera, or have you done a lot of that?
Do you have all your data in your, your data cloud? Have you invested in, in getting, you know, some of the latest models, uh, et cetera. And then, uh, the biggest part is change match.
Yeah.
The effort it will take to drive that adoption is usually underestimated because board members and management and everyone immediately adopted Chat GPT on their phone and we expected that the worker would do the same.
The process of work, and that has not happened or isn't happening at the same pace for many, uh, in hindsight, explainable reasons that have to do with behavioral change and so forth that, that we talked about on the previous episode. Um, um, so change management and, you know, you and I both talked to, uh, professor Brynjolfsson of, at Stanford and you know, he, he has proven in previous transmissions that they're somewhere between, I dunno, one and $9 of services and change that needs to go to every dollar of, of, of, of technology and product, if you will, to get the benefit if you'll could be different for ai, but, you know, in that order.
So I think those investments are, are, are, uh, are very important, uh, to include in it. But when it comes to ROI, the honest answer is it's premature to expect management to build a very solid defendable ROI case for the full transmission.
And I'm not saying that to get a lot of the, uh, management out out of the pressure of the board.
Uh, but it's also very counterproductive for any board to ask, uh, for unrealistic accuracy and detail. That takes a lot of effort from management and essentially forces them to make a lot of assumptions that may or may not be, be right. Um, so far most boards have, um, and most management teams have largely felt comfortable approving the type of investments that have been done so far.
Largely out of strategic intent. Some say fear of missing out, um, obvious, uh, point use cases that indicate the larger potential and the sense that there might be an opportunity or a risk to, um, that that needs to be captured
when we move into the next three years of investments. That investment, uh, can go up and it makes it more important.
But I would ask for sort of a balance scorecard of what do we expect the outcome to be, uh, and, and the magnitude of investments. Um, what do we expect in productivity? Um, what could be some reasonable expectations on, uh, time to market, process improvement, uh, quality, et cetera, that are part of our strategy?
And maybe the goals already exist there, but this will make it easier to, to, to get to there. And then who are we trying to convince? Are we convincing the shareholders? Are we convincing the board? Are we convincing ourselves? Are we convincing the employees? A lot of di dialogue, uh, I is around assuming that somebody else wants this and that they need, uh, to do their work.
And, and it's largely true, but it's also becomes a little bit of a circular argument. Um, you know, I, I can't count the number of times where, where, um, you know, where you're on both sides. You know, management says the board needs this number, right? And then you sit on the board and it's like, I, that's not what we asked for.
We asked for, you know, uh, a di dialogue around, you know, how are we thinking about value creation? Where are we, what are the uncertainties? What's the magnitude of upside, downside? That type of thing. Right?
Yeah. I think one thing that I'm missing a lot is like, it's so easy to default down to the kind of productivity number or, or the increase in EBITDA number, or reduction in headcount number. Whereas I find in the lot that, uh, a lot of the work that I do at Grail is finding new opportunities because the way that I look at it is like the, the productivity thing.
You need to do that, but everyone's gonna do that. So it's just gonna be a new price floor in the market, and it's gonna be priced in between now and three years from now, depending on your market. What's, and, and that's not really a, a, a new and improved competitive positioning, uh, 'cause it's just gonna be equal for everyone.
But what's
well, well, just to reflect on that one, you know, you justifying the most common way to differentiate and create value in the history of Uh, so I, I wouldn't undermine that, the value of that, that that could be enough, but obviously the potential could be even bigger where you're going.
so I have this triangle that I work with, uh, with my clients on, which is, let's look at any business process from, from speed, quality, and cost. And everybody seems to look at speed and quality, which is how can we do our existing jobs faster? Um, and for me it's all about the, the, the quality.
Like what is fundamentally new, what is a, a And we both have a relationship with the lovable guys, right? And I, I think they were the ones that made this click for me that previously in the market of, of software engineering, what used to be the squeeze point was the access and the cost of engineers.
What happens to this market if that's not a limiting factor anymore? like what does that unlock? And I, and I think for any business process innovation, that is fundamentally the question that I try to get to, is there something that used to be a squeeze point or a, a high cost center that all of a sudden isn't? And what becomes possible as a consequence of that, not existing,
Right.
anymore.
Yeah. That's super interesting. So I would say, uh, you know, the most mature boards and management teams, they spend at least a significant, some degree, significant point of discussion on that topic, right? And, uh, we are seeing examples where when you re-engineer the process almost from scratch, but with the, the, the people that do the process today, part of the journey, not as a, a victim of, or, or a defender of if you'll, um, that you can get some pretty dramatic results.
In one concrete example, we were working with, um, you know, with a telco in Europe and they had an offshore, um, finance ops center, uh, in another country if you'll, and the task was to improve productivity by 30%. But by redesigning the process, um, that work could be automated to a hundred percent and moved back to the home country.
Hmm.
Uh, but that required a fundamental rethink of the work. And, and there's one sort of intellectually very stimulating point that if you think about the, the notion of productivity and the history of, of humankind, uh, other than CapEx, it has to do with reducing the number of hours of labor that goes into a process.
Since labor is still a significant portion of most company, not the only one, you can cut out the costs, et cetera, but literally that's been at the core. How do we do this with less people, less hours. If the marginal cost of a AI agent hour equivalent token, um, uh, goes to marginal cost of zero or the cost of compute energy. You will find that there are types of work where you can double or triple the amount of work, but dramatically reduce the cost of that work. So by increasing the amount of work, you get more productivity. And that is so counterintuitive for most companies, boards and, and human beings, if you'll,
Hmm. Yeah,
requires that process redesign.
So there, there're essentially, I, I think two schools of, of, of thought, of companies and boards like you have, I would call it sort, sort the, the here and now people that are, you know, skeptical naturally to all that future potential. And they just want to see enough credible. Productivity to justify the investment.
And then we'll see. Right. Lot of, lot, lot, lot of value in that opportunity. And then there are those that say, well, that will not get us anywhere. Right? We need to completely blow up and re-engineer some parts of our organization the way we do things, right? And we need a board on board to do that. And then, and not only then will we see the full benefit, so, uh, we're not interested in that daily productivity grind.
We're, we're, we're, we're just looking at this disruption of, and they also have a marriot, you know, less proven yet. But, uh, in my view, the, the, the, um, the right answer is do both, right? Do you have portfolio of initiatives and questions and strategies that help you tap into, uh, you know, those different types of creation?
Its very risky.
Yeah,
Uh, and, and to some degree you can do it with the same resources and develop investments and tools, and in some cases you need to dedicate the different ways of doing it. But yeah, no, the, the topic of value where RI for board is one of the most difficult ones right now. And, uh, having said that, um, when you do see all these surveys, right, everyone says that, you know, um, concern about ROI or, or ability to, uh, measure that is a top concern for every company and every board.
Uh, but you also seeing that, at least in our surveys and the, the recent one is just coming out now, this month, um, the intent to invest still goes up. So there's this, which means increasing frustration and, and, and, you know, not being at ease in board management, but yet seeing the need or, or, or the desire to continue that go up.
So it's a very interesting period of time right now.
you do a lot of service at K towards the boards, what are more of the kind of top concerns of the board?
Yeah. So as, as the list we started on, right, the tech trends, the value, uh, second one, the third one are the risks. Um, uh, the, the fourth one is how do you translate opportunities risk into strategy and transformation?
And then finally, how does it affect the work of the board? Those five sort of trends, opportunities, risk, strategy, transformation, and, and work of the board are the top, uh, five topics that for, for every board I've interacted with to some degree, when we get a question from, from a client or somebody on the market, it, it is usually either a.
Risk question, come and talk to us about this, but when, when you, when you ask those five questions, are you interested in this? Almost all of them say yes to all five. Um, uh, second most comment is to have a discussion about just, you know, tech trend, et
cetera. But then you go, the feedback afterwards, the, the, the, the part of that that's most highly valued is the last one, which is the work of the.
Yeah, we're, we're for sure coming there. I'd like to double click on, on the risk section before we do, because risk, uh, to me has been, um. I understand why it's there and it, and it is good that it's there and, and risk when it comes to ai, at least to my experience, is very misunderstood and, and built on, on kind of legacy ideas of, of what is, how does these models work.
And, and I feel it's, it is almost a little bit cartoony. Can my competitor now Google whatever I, I chat with in my, uh, in my chat model? And, and the risk is, is both way bigger and, but also elsewhere, like how would you structure a risk discussion with a
Yeah, no, you have, you have to tailor careful to the industry and the nature, if you will. But right now I said the three hot topics in that risk bucket, or trusted AI bucket as we call it, um, are, um, sort of governance and compliance, cybersecurity and shadow ai. So on governance, compliance, most companies by now have some degree of framework, uh, procedures.
They've gone through, they've gotten advice on, or they're following some of the recommendations around, you know, how, how to set yourself up to do that. Um, uh, and that's important. Uh, compliance is almost the same. Uh, my colleagues would, would not agree, but related at least, but you, you know, you, you need to speculate what the risk is, um, uh, for you to not be in compliance with regulatory requirements that are, in some cases clear, in some cases, not clear or intellectual property, the evolution of, uh, lawsuits on intellectual property, et cetera.
So it's, it's, you know, but most companies have, departments have skills, have people that are good at that risk management, not perhaps the AI generated risks yet. But there's a place to land it. And, and it's an ongoing discussion. I think in that topic. One of the reflections I've seen come up more than once is, um, at some point many companies separated the overall enterprise risk that you always publish in your annual report that you have to do with adding a new AI risk.
And it was very convenient way to handle the AI risk. But you also realized very quickly that all of your traditional risks are impacted by ai and you kind of need to rewrite, uh, you know, all of them in light of ai, right? Which takes more work time if you'll, so that's on the, on the governance compliance, risk on cybersecurity.
Um, you know, it, it, it was very important before. Gen AI and acceleration on agent ai, and it, it continues to be, um, uh, exponentially more important for the very reason that even if you don't do anything with AI yourself, the bad guys out there, uh, are very savvy AI users, and they will get that productivity and quality.
Uh, so you will get more attacks and more effective attacks. Uh, the more potent the models are. And again, they're doubling in price performance every three to six months. So the threat to some extent is very rapidly increasing. So far, and this is not my opinion, but, uh, uh, opinion of somebody, I won't name that, that used to be in, um, uh, the, the FBI, when you look at, at least on the, on the US side, the nature of those risks, it's more about, uh, you know, a bad hacker becoming average or good. Then that the best hacker, you know, becoming, uh, uh, you know, dramatically more effective
So that would translate into more attacks rather than the sophistication of, of the attacks.
for, for now. Uh, at least that is a perception of one, one expert that I've found, you know, credible. I don't have data on that, but, uh, it may not true be true tomorrow, right? So you need to kind of think about both if you'll, and that's if you do nothing, right? If you do do something, which you'll forced to do, if you wanna be on this journey, you'll also, uh, open up more attack vectors.
So it's almost impossible to kind of like stay on that. You need to constantly, uh, as a board member, um, try to ask and ascertain not only where are we today, but what is our plan to increase. Our level of defense or handle the risks that are there because it's also, you know, uh, risk is part of the business model.
What is the level of risk that you are willing and able to accept, uh, for certain outcomes, right? And there are whole industries like that depend on the ability to price that risk, right? That most other companies, um, are not always thinking about that. So that cyber is a huge and growing, uh, part of the agenda.
We can get a lot of help from AI though, uh, like, uh, anthropic for example, has a brilliant, uh, suite of, of uh, kind of security skills and so forth, and we saw yeah.
If you if you deploy them..
right.
them. Yeah, sure.
So the opportunity for the attackers goes up, your opportunity to defend goes up.
So you have this
arms raised to some
extent, right? Which makes it an evergreen topic. Uh, further accentuated by, by ai. Uh, most companies have focus on cybersecurity. They may not have enough or, or update it, uh, frequently.
I, I would, you know, I, I would reflect on whether once per year is enough for a board these days to have this dialogue. Right. I, I would, I would opt towards more of a, uh, frequently review. And then thirdly, um, shadow ai. So essentially what, what your employees do with their private AI tools at work, whereas part of their work.
Yeah. It is unfortunately a very common phenomenon risk, um, both because companies have not yet provided them with, uh, you know, a, a strong enough model, if you will.
Uh, often under the kind of shadow argument of risk, which is a fantastic, uh, funny coincidence. Yeah,
irrational. Yeah. So I would say one of the biggest risks right now is not having your employees on your control platforms. Uh, if you'll, um, and, and there's, there's a lot of mostly survey based, uh, uh, input on that. We have asked 40,000 people around what they actually do behind closed, uh, doors, uh, when, uh, when, and, and it's quite scary, right?
Or it could be quite risky, but it also indicates the productivity they feel from using. They wouldn't be doing that if they didn't feel that it was relevant for their work. So the people that question the ROI also have to explain that in light of how, how could, how could both be true?
Yeah. There's something broken in that equation that, that we seem to struggle with adoption at the same time that we have a lot of shadow usage.
Exactly.
So it's not always the same person. It's not always for the same purpose. Uh, and there are a lot of drivers that, that you can affect there. But looking at it as part of your adoption strategy, you get the double benefit of of, of, you know, reducing or managing your risk.
Yeah, are many o I would bring in also the, uh, kind of talent risk into this equation as well. I have a lot of conversations with people my age, high achievers that are working at companies that are AI laggards that are saying, I take too big of a career risk staying here, uh, because this is obviously gonna shape the, the next coming X years.
And if I'm not hitting the ground for running, uh, I need to get out of, of wherever I'm working currently. And that's a huge risk if you start to having a, a brain drain from your company as a consequence of a a a too scared AI policy.
Absolutely. And on the other end of that same spectrum, you have people saying, I'm, I'm actually nervous to use these tools because if I rely on them, I may be held accountable for some error that my boss will fire me for. So I'd better off being less productive, but knowing that I'm equally, even if I were more productive and, and better with the model at least.
So then trust is a key component across all of this. And, and you know what, what binds it together is, um, I think that the, the best boards and companies they see trust not as a, um, as a break or, or, or, or something that slows down transformation. Uh, but we like to say that then no company can transform faster than the pace of trust.
Hmm.
The pace of trust that their employees have in the company, their job and the tools, uh, customers have in their use of their data regulators have, and, you know, following regulation investors in deploying this, you know, in a prudent way. So, um, the, the, the best way I feel is to look at trust as an accelerator, not a decel, in that, uh, it takes a little bit of minding.
Can, can we discuss, uh, the, the kind of investors to the board, um, touchpoint as well? Because from, from the investor perspective, obviously, and a a company valuation perspective, we see a lot of companies kind of scrambling for really the answer to the simple questions. Are we winners or losers as a consequence of ai?
And I've had in, in kind of behind closed doors, a lot of discussions with CEOs saying more or less that I've kind of bullshitted my way through this answer. But now I, I really need an answer to the question. So how much pressure are boards facing from their investors? I know they're composed from the investors a lot of the time, but, but can my, my experience at the board level, how this actually works is very limited.
Can you talk me through how, how this would
Yeah, I mean that, that varies enormously depending on the maturity in the company industry. So there's almost not, not, not, not a single answer. That's, that's right. But I talk about some of the extremes. I, I think at the most basic early stage company in the industry, that's not going through the most dramatic, right.
Um, there's more of a. You know, let's, let's not get ahead of our skis and overinvest type of sentiment from investors, right? Show me the ROI, uh, you know, yes, you know, this may happen over time, et cetera, but, but let's make sure we're not last, but we may not need to be first, right? Um, and that could be a very prudent strategy for some, it could be a very dangerous strategy for, for others, right?
Um, if you go to the other extreme, you are basically, uh, your company's valuation becomes, uh, exclusively tied. To AI story.
Right. Um, and we've seen several companies in the tech industry being both able to redefine their narrative to get an AI multiple and not being successful in that.
Right. And that is, uh, you know, and that's a constant, uh, and I'm not an investor and if I was a good one, I might not be on this call here, but, you know, it's a constant struggle for any investor to keep up with the value flows and the perception of the winners and losers in, in, in, in that case. Right. But
yeah, and as you said, we've seen this in the tech industry very clearly with, with like source multiples. So, during February, specifically in March, do you think it's feasible to think that this will hit most knowledge worker industries similarly? Just that the tech industry is a little bit further ahead?
Or was it like, oh, it was the big, but also the first scares, so it, it
Yeah. I, I, I think the prevailing sort of mood of thought is that exactly that, right? Tech industry is leading, it's both building and using and enabling all other companies to use and therefore what happens in tech will happen to other companies, right? It's tightly coupled with where we see the most productivity.
And it's a number one use case for AI is coders, right? And it's beyond doubt, you know, that, that they're being transformed and, and so forth. And now the prevailing thought is that what you see to coder knowledge work will gradually, uh, happen to any knowledge work and, and potentially even physical work with physical
and then on their own. everything else. Yeah. Sure.
So I, I think there's a lot of merit to that way of thinking. Um, the reality will be very mixed depending on where you're and who you're, et cetera, right. Uh, to, to do that. But I, I think that's clearly the way to, that the most are thinking. The irony is that a lot of the biggest tech companies that are deploying and building these are so focused on building the technology that they're not themselves fully using their own technology to drive the more basic productivity, which is so secondary to their valuation, right?
It's growth quality of their products, and whether they have a 10% more effective back office doesn't really their their needle, right? So they're focusing on the right. Things from their perspective. Reality is different in many other matures, not moving, you know, that that 10% back office efficiency could be, do or die in, in, in that. So, so that's different as well. Um, so I do think we're gonna see a lot of, lot of that. Then when you hear, um, a lot of the, the speculation in, in media and investor communities, right? You, you have to separate two things. One is the underlying, um, impact on competitiveness and productivity that AI unleashes.
And the other is the perception of investors or where the train is going and which is somewhat decoupled from the here and now. Short-term value creation, right? Both are influenced. And on top of that, you now have geopolitical macro factors that, that can override even the biggest, um, short-term productivity gains or losses. it becomes very, very difficult. And, and boards food. Do well to ask themselves those questions, right? Is this a, is it, is it an, is there an underlying productivity? Um, difference here that would motivate some of that. Is there a investor sentiment around winners and losers, uh, bubbles or not?
And is there a macro political driver overall? Which of those are the ones we need to care a lot about? And here's where AI and geopolitics and, you know, uh, external market uncertainty comes into the boardroom, usually at two different types of the agenda. But I think increasingly, uh, you know, being, being one right to help guide that, that through it.
So the latest discussion of course around, you know, all the. The chip manufacturers, the data center investments, and the SaaS companies, you know, valuations that, that have been, you know, looked at, uh, and reevaluated in many ways in the eyes of the investor. Um, some would say that there is a lot of overreaction and, you know, uh, and irrational behaviors.
I would say whatever it is, it's, it's where the train is going and
the truth is somewhere in the middle.
Yeah, exactly. And that is probably true both for, for the investments racing into anything that smells AI and, and the investors fleeing out of anything that smells punished by ai. Uh, that's interesting. I'm gonna open up the, the kinda last chapter, but perhaps the most practical chapter as well, uh, which is the inner working of, of the board.
Like how is that practically changed as a consequence of ai?
that's really a good question and sort of maybe even before we go there. How, you know, the, the implication on that is how does the work on strategy and transformation, um, uh, change because that impacts how the boards work with change. Uh, and this has been a hot topic for boards for, you know, the board's role in strategy, governance and oversight is, um, longstanding discussion, right?
There are many opinions around how much ownership the board has of shaping and, and sort of developing the strategy versus overseeing and, and, and sort of, uh, you know, asking for updates on the strategy. And I, I, I think, you know, um, learned scholars will differ a little bit on what the optimal is there.
Um, but it's one of the areas that put most sort of gray differences between boards. And this comes to head with AI because AI not only requires in strategy. How to deploy ai. It impacts the corporate strategy that existed before ai
Yep.
and in many boardrooms, and we were one of them, we, we had a lot of debate initially around, do we need an AI strategy or do we just need a corporate strategy influenced by many things, including ai.
And if you randomly pick any board member, you'll find they fall clearly in one of the two camps.
Mm.
Uh, I'm personally in the camp of, you need both. And I, I I, I, I joke a little like, does your company have an HR strategy or a vacation strategy or location strategy? If you have that, then you need an AI strategy, which answers the question, how will you leverage AI to improve your company?
Um, your corporate strategy is than that, and it includes many drivers, not only ai, but. Almost any company that within reasonable time has a perfect corporate strategy that goes unchanged in any way with such a massive, that would cause pause as me as a board member if that happens. Now, strategy is all about the how.
It's not, you know, it's not the why or it's, you know, the what happens when you're there. And the, you know, the how gets translated into most companies into some kind of transformation, uh, initiative.
Either a new one, your AI transformation or a better old one, your ongoing digital or company transformation, uh, or even just a set of portfolio of initiatives where you change stuff in that, whatever it is, all boards would do well to reflect on what is it right, what do we expect from it?
How does it interact with all other initiatives that we had doing and money, money companies had a separate effort looking at productivity and cost reduction. Separate effort at AI and et cetera, and they didn't have not yet merged the two. Uh, which I think is a, is a good board question, right? So you, you get into that and, uh, there are many ways to articulate that strategy.
Um, but if you look at, um, some of the clear ones, it, it, it contains some components, right? It contains something around what you will do in the market differently. You know, where you expect to grow, what you, what you expect to sell, how you know you'll price that, et cetera. It has some components on how you will deliver your products and services
differently. And it has some components around how you, how you drive your company differently. Those three things. If they're not present in your corporate slash ai strategy, then boards would do well to
would you add the fourth around changing customer expectations and behaviors?
Yeah, it, it's, you could, to me that those are outcomes are embedded in the three,
um, on one hand you can try to set customer expectations by the way you price communicates,
features, et cetera. You don't control it. And the, the perception of customers and competitors are influenced by how others, you know, try to do that as well.
So it's hard to do it in isolation if you'll, but it's certainly part of the, the, your, your strategy should include. What, what do we think would be beneficial for our company's ability to compete now and, and thrive in the future? Uh, you know, a very common one is, is pricing. Right? Um, if, if you, we are, uh, one of the world's biggest professional services companies.
We're certainly in the middle of this epicenter of change. A lot of our revenue comes from, um, you know, uh, the same companies year after year. Some are new, uh, and uh, usually at sort of hourly based work, input based, uh, agreements that have been formed, uh, over many, many years. And often at the request of our clients and customers,
Yes.
particularly the purchasing departments of those kind, you know, are are very good at enforcing a certain way. Now there's, there, there is a clear direction to move away from that not only professional services, but in more other aspects towards more of an outcome-based fixed price, you know, ways to, to pay for the value, not the input that gets you the value. And that's a pretty clear trend that leads in perception and in communication in all aspects of that.
It's, it's not yet as clear in reality and, and partly because of how companies like to buy or have been trained to buy. And that takes longer to change, but it's certainly capability you need to have. Um, and you as a board member, you need to make sure that the company, uh, kind of reflects on. I also remind people that, you know, there's a little bit of a fallacy, like if, if I just jump to a fixed price or an outcome-based price, I will be able to retain some of the margin for myself.
I won't be squeezed out by my customer, et cetera. And that is usually. Not true, or at least not directly true because it's, it's your ability to deliver more value and your ability to do that, deliver that value at a faster pace or at a lower cost that enables you to set a price that maintains or increases your margin.
Yeah, absolutely. It's not a cosmetic itself. the top. Uh,
Exactly. So I often ask, you know, okay, we're moving to fixed pricing. Okay, okay. Uh, at what level? Yeah, we'll figure it out. But how would you figure it out? Right. Well, you know, we wanna win in the market. Okay, but will, will that enable you to sustain margin or will those margins die if you win in the market? Uh, it, so I think there is a little bit of a mental crutch to jump to to that if you'll Right on the
I think, uh, because I, those are some a lot with, uh, with kinda service firms and, and most of them have, you know, 60, 70% of their revenue comes from some type of bread and butter type of deals where we could invest quite heavily in the efficiency of the delivery vehicle to get to that outcome. And then proactively price a fixed price that is lower than the market and then get into a really sweet position where we offer something that's faster, better, and cheaper, uh, versus market.
But for this very kind of defined subset of our services, that is the, the, the high volume, high repeating business
exactly.
and then keeping the
so you know.
more hourly or, or whatever.
Yeah, exactly. So, so as a board member, you know, you wanna understand, are we doing everything we can to create the flexibility for us to have different pricing strategies and tactics that would win in the market? And, uh, and are we using those into the way that we front customer One of the two is not enough, uh, because it actually hurt you in that sense.
Uh, but if I end on a final, sort of like we, we, obviously, we, we, we talk about our client zero story. We, we happily share in, in various waveforms how, how, how we have thought about it, good and bad. But,
And for the listener, the the the, the, client. Zero Is, is KPMG implementing the
what do we do to ourselves?
Yeah.
So we try, we like to try things on ourselves before we, we tell clients what to do. Uh, and we've learned a lot. We've done a lot of cool things, but also things that, you know, that didn't work out. But I, I think the most exciting things right now that are happening is, is thinking about the professional services, uh, portfolio of offerings and the, um, pro productization or potential productization.
Although there's a lot of discussions around services as software, not software as service, but services as software where some portion of the professional hours could be delivered through more of a product. Uh, you know, a build a tool, a technology, uh, while still having, uh, you know, all the services required to make that work in the tech industry that's existed for. decades Right. Products plus service in the services industry has been service Now it becomes service and product to some extent. So, uh, um, one of the coolest things I, I think w we've done is, uh, first of all, of course, we have tried to reimagine wherever in the world we can do our work more effectively with, uh, AI and now agentic
ai.
So we built a global agent delivery platform called Workbench, where all our. d igital task task workers and coworkers will live and breathe and,
you know, get their energy, et cetera. And so in theory, if we develop, train an agent in Australia, we can deploy it in Stockholm or New York the next day while being compliant with all the regulatory constraints that we're heavily under, of course, as a, as an audit firm.
second, we are building, um, uh, in numerous, I would say at least somewhere in the order of nine agent factories, right? Uh, where we will take work and try to automate portions of it as quickly as possible in sort of a mass production, you know, facility. If you'll,
And what, what, get a little bit more practical. What, what is an agent factory?
An agent factor is a, is a repeatable, um, structured, um, process coupled with the latest technology that, that enables you to say, here's how we delivered a.
Finance transformation project in the past. Here are the work steps, here are the activities by the humans, here's the data, here are the frameworks and the knowledge and extra property we used. And then translating that into, uh, tomorrow with current technology and using all the latest, we could, um, you know, deliver 30, 40, 50, 60, 70% of that human work with technology.
In this case, agent AI and AI technology, if you'll combined with our human services.
E every, every service then has a different current point of that. Some cases it's zero. In some cases it's a lot, and it will change over time as the models get better. But the more we can do that, the more we can focus the human labor, almost uniquely human value added and interesting to do.
So, uh, the, the work gets better, right? So that's kind of like our bulk production facility for knowledge work, if you'll,
And it's interesting because you're, you're, you're what? You're 150,000 something employees. You're a huge, huge company. Uh, there you go. I I, is this a kind of moving circus of, of like transformation people that, that are trained on, on like this process and the tools and, and how, how to make this a repeatable thing that moves around from, from Stockholm to New York to Sydney?
Or is it more like we find one process in Sydney just because of serendipity of, of something happening in Sydney and then we try to mass export it to all of the 250,000 or the, the 80,000 that would be affected by this, uh,
Yeah, it, it, it's somewhere in the middle. It's more, more of the second. It's a federated approach where we concentrate infrastructure investments in a, a, a finite number of places. Assume you know, US, Germany, uk, Australia, a handful of companies that are central in the region and that have the scale and, and, and capacity, um, uh, to run these factories if you'll, and then what comes out of the factory can be used by everyone.
Um, so, so pretty much like you would, you would think of a manufacturing plant type of approach. It's just that it's a knowledge worker factory, not a widget production factory, if you will.
That begs another question, uh, that I've found a lot, um, from CEOs and especially, uh, companies where you have like regional CEOs or country managers, which is relating to how do we properly structure AI investments? Uh, because as we mentioned, uh, some time ago, there isn't typically a dedicated AI budget.
There might be now for 26, but most of the time any AI investments comes out of your own profit and loss, which then incentivizes not really the kind of build here and export to everywhere. It's a little bit, I know that we're all in the same boat, but if, if, uh, if you want to make a, a investment, you, you'd prefer it to be some type of umbrella investment, but there, there isn't really.
Can, can you speak to, to like how you structure investments correctly?
first of all, we, we, early on adopted the principle of innovate once, you know, use everywhere. Um, which, sounds really good and it is good, but it, it is hard to do. Um, and we also have the added complexity that, you know, we are not one global firm. We are a network of global, independent legal entities, uh, that, that operate under, you know, as independent companies under, you know, global set of collaborative rules and, and investments.
So, so our, we do have, of course, a vehicle to, to, to share investments globally, but it hence the federated approach works really well for us. If you're a truly globally into integrated company, you, you may need to be more centralized than that, but it also increases risk if you'll so it, the, the investments come from, um, you know, all, all the member firms.
Uh, uh, into what they individual would do. And then there is a shared investment, uh, that goes across the global network, uh, to support those things that are truly global in nature and central, et cetera. Right? But it's also very important and tricky to handle the intellectual property, and they handle on this, that we're independent entities in, in many countries.
So that's kind of like the bulk of the approach, but this is moving so quickly. So two other strategies that I'm very excited about is that, um, we have also taken. Clients into the account. So, um, in several cases we've gone to clients to say that, you know, we agreed to do this work manually and, and, you know, we are really happy about the, the outcome, et cetera, and you can trust on that.
But if you're willing to co-innovate with us, we'll invest in trying to do this even better with the latest technology and if there is benefit with the
difference, right? Um, so we've done that, uh, on a number of, of engagements and found really, really attractive, um, returns on that, which has sort of spurred us on to do not only centrally planned.
Production distribution, but already changing at the edge while the factories are being built out. Right. But the even more interesting thing now is, is our, what we called our, our sort of asset based portfolio. And our red teaming approach is not called that anymore. But essentially we wanted to make sure that we are not disrupted by any one of our competitors and that we leverage the full innovation power of entrepreneurs, particularly in Silicon Valley, where, you know, the center of the world is right now and to be right now.
Um, so we employed an approach where we took. Existing really high performing partners out of their line delivery roles, which was a huge risk of course, because they generate a lot of our revenue. And a handful of them were put into startups in Silicon Valley, bringing their knowledge, data frameworks, et cetera, but then, uh, became founders for three months to build the new company or a product that could disrupt themselves at, with a fixed stack of the startup and the speed, uh, of the startup, et cetera.
And the first three of those products that then succeeded, uh, uh, are now coming onto the market and we're deploying them to paying customers this month
actually. And one of the coolest one is called, uh, Maestro, if anyone was a Google at Nyst RO. And essentially it's the first five digital KPT employees, uh, that we will take to market.
We have trained. Uh, AI agents that can do sort of the, the a hundred percent of a role that it can either be part of our change team or part of, in this case an hr, uh, team to do opportunity assessment for AI to do all those things that they need to do for ai. So we'll see how that works, et cetera. But, you know, it's the first sort of, uh, fully digital co co-members that we put into the market and, uh, we, we we're trying those right now.
So those are ways that we are trying to transform ourselves, if you'll obviously using our trusted AI framework for everything we do, which we're very proud of, and the intellectual property that we have to try to codify that as much as possible. But this is a race. All the, this is the so-called consulting firms, professional services firms of any kind are in this, uh, you know, head-to-head race.
And, and, and really it's about how fast you run versus, so that's a little bit of, of sharing, uh. Public information more or less around the, uh, how we thinking about it.
Okay, so, so perhaps a, a a little bit of a, a tangent here, but again, speaking about the, the board and what should you and shouldn't you have opinions on from the board perspective? The last piece, the maestro piece would for me, beg the question of organization design for the future. Uh, how do you have that conversation or should you have that conversation at the board level?
yeah. That, that's one of the, you know, yes. But not in too much detail. I think, um, the experienced board members would find ways to ask a question around, uh, how our planned organizational changes align with our AI strategy and the changes to our corporate strategy to be able to enable that.
Yeah.
What changes have been made and are compli complement sort of contemplated to be made around talent profiles, structure incentives, training, all those aspects.
Right? Getting into the nitty gritty of what that is, is, is very much management and making the decisions. All of that is management's role, right? But having that interaction, both in terms of sort of getting the, the benefits, but also protecting, uh, the employees and, uh, you know, from, uh, burnout from too much, uh, change fatigue and, and being balanced in, in how you get the people along. 'cause at the end of the day, my humble opinion, people are now the biggest, both barrier and lever for, for, for growth and opportunity capture if you'll, um, so it's very hard to do this to the people without having. You know them with you. And that's of course, a board concern when it comes to the, the talent sentiment, um, you know, uh, people leaving or wanting to join or not, right?
And all that stuff. So it's at the highest level. I, I think, yes. Uh, but it's also very uncertain, right? Uh, it is any ways that you can change your, your, uh, organization and your talent profile. There's been a number of, uh, you know, famous professional services companies that have publicly stated that they have, you know, gone through all their recruiting data and completely reversed
their hiring profiles, right?
I, we, we, you know, we, we haven't quite seen that in reality yet. And I think, you know, somewhere in between is, is is prudent, but it's worthwhile thinking through. Um, you know, one, one of my pet things right now is that we are seeing a lot of companies pause around the early, uh, hiring, uh,
principles.
absolutely.
Eric showed that, you know, canaries in the coal mine paper, right?
That there are, there are some measurable effects, but yet only for a very few number of, of, uh, roles. Uh, but the sentiment is very strong that this is happening. And, you know, having kids in college, I sense that as well, and to some extent is very rational, right? You, on one hand, you're struggling with the change management of people that have been in the company for, for much longer.
And then you have this group of people that are AI-native.
Yeah.
Don't need any. help Right. They're cheap and they, with the help of AI, will perform as an average member very, very, very quickly. Right. Or more quickly, at least. A So rational organization would be to have more of them eventually than less. And, and there's this big, big discussion around, you know, a lot of job loss announcements that we are cutting our staff because of ai, et cetera.
And yeah, and if you haven't been in the boardroom, that may or may not be the primary driver, but I suspect for these many companies I see is that there are, uh, that, that are additional and even stronger drivers of that. It's not that AI has replaced those, it's it's more that the investments that are required for AI or, or, or anticipated to be required for ai, um, people are hedging and trying to cut costs wherever they can to prepare for that.
Uh, right now, I would say that is more true in general than, than job replacement, but it's certainly possible in the near term and something that Bos need to be very, very, um, you know, attuned to because of the pros and cons.
And talking about risk on, on the same topic, one thing that I, I do. In a year's time, two years time, talk more about this, the consequence on, on kind of succession planning. If, if you have nothing coming in from the bottom, you won't have your, your future middle managers that to promote. Right. Um, and that is a classic tension.
Right? Well, it's not a problem today, but it'll be perhaps a problem to, uh, for tomorrow. And then Okay. It's, it's future one's yeah.
Exactly and it's anticipation. And I mean, just this week that, I mean there were two senior CEO transitions announced
where the ability to govern a company through AI was among the reasons why that change happened. Probably could happen anyway. I dunno. But it's, it's starting to be mentioned at least, which gonna gonna drive that if you'll
Yeah. And I, I, I led a couple of teams where we were very aggressive in, in terms of pivoting towards AI nativity. And what you end up finding is that you, you run a much leaner team. Yes. But you're also way more dependent on those specific people staying, like each pay, uh, player becomes a key player in a sense.
Um, and so that's, uh, another risk to, to kind of consider. Yeah.
If you bring all that back to kind of what, how, how does that affect the work of the board, right? The, the, the, you know, the, the weekly, quarterly, et cetera. And a again, of course all, all boards are different in that sense. But first of all, if, if you are on a board and you want to, if you want to kind of make sure that, that you're talking about the top five topics that other boards are, are talking about, I offer those five, right?
The tech trends, the opportunities, the risk, the spread and transformation and the workload board itself. There are many others that you can talk about that at least you would not be an outlier if you covered those five. Uh, and, and at the high level, um, now that's a lot of work and a lot of time, right?
And three years ago that was, that was not the case. Uh, you know, you had. Important things to talk
about then. Uh, I actually looked up the first, or at least according to Perplex Perplex, the major paper that came around board governance, uh, for AI came in November, uh, 23 from Harvard Business.
Mm mm.
So there's been less than three years of any type of guidance to board members on how to be a good board and board member with this, right?
And most board tenures are longer than that. So essentially the vast majority of the boards are now dealing with this. Were not, um, put together or recruited at the time. You know, they, they were recruited before Geneve AI really hit, hit the, the sea. So we're all learning together again, and particularly as, as board management.
Um. So a couple of things that, that, uh, in that topic that a lot of boards are thinking about. So first of all, it's just the mundane task of structuring agendas and allocating time, uh, which is, you know, uh, very integrated, but very, very important because that is, you know, that structure by which the boards operate, uh, in the form of board setting, right?
And if this thing with AI is now going to take more time,
um,
you have very few, right? Either you reduce the time spent on other things, which most board members and companies do not want to do. You increase or you increase the number of hours put in by the board, which most board members are reluctant to do, or, and, or you increase the effectiveness of the hours spent so that you can, uh, handle this more and.
Uh, you know, one thought is then of course using AI to be more effective as, and, and, and, and that has been, uh, a slow evolution, if you will. It is happening. I'll come back to some of the coolest things there, but, but, you know, so one or two of those three need to give,
Yeah.
and none, nobody's gonna be happy with the answer to that one.
So important to have it upfront as a discussion. So when it comes to the time structure, et cetera, what I, what I at least see or recommend boards to consider is that even with the same time, think about the frequency and context of that time. So, take strategy or AI strategy or broader corporate strategy.
Many companies used to have an annual, uh, offsite and they spent a day or two talking about the strategy and that was kind of it,
right? At least have consider having that discussion at a quarterly basis or even more frequently, given that the pace of changes are, uh, when we went through COVID, right?
We, we went from, you know, quarterly in-person big meetings to sometimes weekly virtual meetings when, you know, the change was biggest,
maybe
some kind of mentality in the boardroom, uh, could, could come back without overreacting and overblowing it, right? Um, so that, but also how you group topics on the agenda. Should the corporate strategy and the AI transformation be discussed on different days or different hours or next to each other or in the same discussion? Right? You can group topics, which with pros and cons in that sense. And then there are, um, topics that you should increase, um, and before Gen AI that were benchmarks from many different places around what seemingly more effective boards or successful companies spend more time on.
And sometimes that's self-serving or not, but usually it goes to the bigger topics, the more strategic themes, um, uh, that are slightly more forward looking without reduced. So those are some of the directions of looking at that in many cases. Um, I see boards at least temporarily increase the time spent until they can figure it out with, with nobody being happier for it,
I think one of the evolutions that I've found a lot in, in, again, then in, in the context of management teams is how effectively can we prepare clear decisions as a consequence of, of using AI better in creating much better, uh, preparation for decisions, uh, in, in how we handle pre-reads and transcripts in, in, uh, the board meetings perhaps.
And, and, uh, running those transcripts through a set of, of post-processing agents that are, are clarifying things that we've left out that we should have discussed where we have confirmation biases. So just effectiveness of the idea of a governing body can be increased a lot with the technology, and I would imagine the same is, is true for, for a board, right?
it, it could certainly be. And, and, you know, that's, that's the third area. Um, so there, there's also a lot of, uh, risk and a lot of, um, resistance to bringing AI too quickly into the boardroom for very natural reasons of, um, uh, being compliant and being, uh, minimizing risk of confidentiality leakages and so forth, right?
So, um, that, that is very important to consider. And anything you, you do in the boardroom, uh, needs to be thoroughly quality, vetted and assured by from many those angles. Right now, most boards use a software tool to help the work of the board. There are a number of them out there. Uh, most big boards use one or two, uh, uh, if you will, and they naturally already have. Modules and additional functionality that uses AI to make the work of the board, the pre-reads, the questions, et cetera, easier. Um, in many cases so far, they're not turned off
Hmm.
and there's a debate between the board and management and the, you know, around which of these features that are there could be used.
Hmm.
Uh, I'm not saying that I know the answer to that, but at least it's not the lack of access to cloud. In some cases, it's, it's the willingness to use it and that should be prudently handled, but there is not productivity there against yet to be, yet to be had on, on the very futuristic and, um, you know, topics that most board members don't like to talk about.
Uh, you know, there are examples of very few companies that have even appointed, uh, an AI as a full board member to have, you know. Uh, more, more even do note takers and sort of like sentiment tracking. Are we talking about the right things? Are we leaving somebody out? But even, you know, full, full board member, voting board members have only a very few,
mom.
small companies, not, not in the biggest countries.
So
at this point, but it's an interesting
it's gimmicky et cetera. Right? But, um, the most interesting one I have now, which was an experiment, right? And it was, it was, uh, I believe it was insead, uh, and it was published in the Harvard Business Review. They, they, they ran a, a, a simulated board meeting, um, and they created a full agentic board with multiple board members, each represented by an agent.
Yeah.
And then they had, uh, human boards, um, I think there were two human boards, and they were, they were, they were prepped to do the same board meeting with
the same, you know, discussion, et cetera. And then you had blind, um, blind sort of blindfolded the blind testing, uh, a jury of humans
that would judge the performance of these different boards on a number of, I think it was seven or nine criteria if you'll,
and not knowing which board was which.
The humans scored the fully agentic automated board higher on every dimension on average ,
than the human board now I'm not saying that this means that board members will be replaced by ai, but as board members think about their companies and, and, and their employees and the leaders being augmented and in some cases the roles being fully, that will at some point be a discussion for the board itself.
Not today, maybe even not tomorrow, but, but technically that, that be, and that's the far extreme of where I will go on, kind of like the use of AI in the
no, what I, I, I
I think it's the fundamental kind decomposition that really any employee in any company is facing right now. Thinking about what value that I bring is fundamentally something that's uniquely me and my human kind of contribution that I should perhaps double down onto, like my human judgment, my ability to build relationship, build trust versus what value that I bring do I just bring out of old habit, but should in reality be, be quite significantly augmented by ai.
And I think it's very relevant for the board members to reflect on their own contribution in a similar fashion. Right.
No fully agree. And I mean if you, if you bring it all back to kind of where, where, where we, we started the journey. When you look at the roles the board have played so far in the three years and the bo the role that the boards will need to play in the next three years, which is within the board mandate of most board members today.
Uh, it's gonna be incredibly important, exciting, and difficult for all boards to go through this. Right? And it's important not only for the company to serve, but, but you know, for the whole. Economy, economy, the world that the boards, um, play this role, uh, with so much change, so much opportunity, so much risk.
Um, the interaction between the board and management on these topics are going to be absolutely crucial.
And having that balance in the boardroom every day and every board member feeling that they can be at their best, uh, because nobody's an expert on this one. That, that we're all learning together, taking the time, uh, to, um, to think through these aspects.
Uh, without this steady set, uh, opinion one way or another. It has never been more important and will continue to be more important. And that's why I think it's very exciting topic and one that I'm very grateful that, um, you allowed me to speak a little bit about.
Hey, this has been absolutely fascinating, pat. Hopefully not the last time that we'll meet each other over the podcast format. Uh, thank you so much for, for coming on today and speaking about the, the board's evolution in the age of ai.
There we're at the end of the episode. I think it's so amazing that almost 70% of you actually listen to the end of the episodes, especially for long format. I'm, I'm so thankful for that. Like the engagement numbers is fantastic. Before we leave each other, two things, first off. This episode was sponsored by Grail and I'll make no secret about Grail is the company that I'm starting.
So it's the future of AI and how can we unlock value that's not just automation, but actually building towards a future that we want to live in, like a future where I want to work augmentation. So head over to Grail Works to know more. Secondly, I'd appreciate so much that you listened to this podcast.
Even though 70% listen to the end, almost 70% are not subscribed either. So if you can give, a like button a subscribe to all my channels. We're on LinkedIn, we're on Instagram, we're on TikTok. Apparently, uh, we have an awesome webpage There's so much more content, like yes, the reels and so forth. But what, what I really would like to push for is that every week I publish a long format article, which is the pinnacle of my thinking.
So it's the most interesting thing that I come across that week, either in podcasts or in other endeavors. I try to write about it towards you, a senior leader, like what is the crucial takeaway from the most important part of my week, let's say. I publish that on LinkedIn and on my webpage podcast. Thank you so much and see you next time.
That was Per Edin on ThinkRoom — where exceptional minds think out loud.