Fit Happens: The Executive Search Podcast

Jason Baumgarten 00:00
I'm Jason Baumgarten and you're listening to Fit Happens, the podcast where top leaders, investors, and board directors share the stories, surprises, and hard earned lessons behind finding the right fit. Let's get into it.
Jason Baumgarten 00:18
Welcome to Fit Happens. I want to talk a little bit about a problem MIT had. There was a talented entrepreneur that they wanted to work with and there wasn't a job, not a professorship. He didn't have a PhD, not a research post. He'd been building companies, wrote some code, so they made something up.
Jason Baumgarten 00:36
They called him the hacker in residence, the first one. And years later, that same person runs a research institute ranking the S&P 500 on how deeply they've absorbed artificial intelligence. He teaches some of the most oversubscribed courses at MIT, and he's written a book with a title that should probably have you worry if you have a job. It's called No One Works Here. My guest this week is Paul Cheek, a person whose skills fit no existing slot, so MIT had to build one around him.
Jason Baumgarten 01:11
And the book he's written says the era of humans doing the work is ending, and the era of humans designing the intent is starting. Paul, welcome to Fit Happens.
Paul Cheek 01:11
Hey, thanks for having me, Jason. Super excited to be here with you. And look, I want to be very clear from the onset that when I say no one works here, that's the title of my book. Well, two things. First of all, I didn't choose the title. And second of all, that does not mean that nobody has a job. It simply means that work has changed. And as you said, it's all about designing intent and not just actually doing the execution.
Paul Cheek 01:44
So I love it. Thank you for having me. And yeah, I'm excited to dive into everything that relates to AI maturity, AI literacy, what it means at the highest levels of an organization and why that's so critical. But also what I don't know, the state of the art org chart looks like. Anything else that's in there that we can dive into, I'd love to talk about. So, yeah, thanks for having me.
Jason Baumgarten 02:01
Well, Paul, I appreciate it. As listeners know, every week I do a sort of one on one, more theoretical academics episode. And then the next week we have a practitioner, whether it's a leader, an author, an academic, come on, talk about kind of the real life.
Jason Baumgarten 02:16
But as a part of that, it's always important to talk about the real human. So I'd love to start with some of the early days of Paul. Where and how did you grow up? What are the early shaping moments of Paul Cheek?
Paul Cheek 02:33
Yeah, I grew up just outside of Boston and as a kid I had a love for telephones. I don't know what it was about telephones, but if you gave me a telephone, I would sit quietly and play with said telephone. If you gave me two telephones, even better. That was my thing. I loved them, I absolutely loved them.
Paul Cheek 02:46
That, building Legos, really building things from zero to one. And I feel like that's kind of carried through professionally. I love building things from scratch and getting them off the ground and running, which is probably, as you alluded to, some of the entrepreneurial nature of my career so far, which I think is so important. I think it's one of the most important characteristics folks can have professionally, is that entrepreneurial mindset and the drive to build things from zero to one. But I also learned there's a big difference between building a product and building a business.
Paul Cheek 03:11
And the latter is what has real impact. I don't know, what else do you want to know? I spent much of my late teens working in a grocery store. I loved it. I became produce Paul in the sense that I worked in the produce department, worked my tail off at the grocery store.
Paul Cheek 03:27
Loved that, had so much fun with that. And I think part of that was kind of my interest in business, in the way not only that you could stock the shelves with produce, but also the ways in which the overall business of the grocery store worked.
Jason Baumgarten 03:44
What was your first job? First paid job?
Paul Cheek 03:45
Painting houses.
Jason Baumgarten 03:45
Painting houses. I love it. You say first paid job, what was your first unpaid job?
Paul Cheek 03:55
Oh, yard work around the neighborhood, helping people out, raking leaves. Very New England thing. Go around the neighborhood with a rake and offer to help people, stuff like that. And you'd sort of hope somebody would drop you a five dollars every once in a while. And I probably did some babysitting very early on. And then we started painting because we realized we were pretty good at it and we could charge a fraction of what the big companies did.
Paul Cheek 04:12
And it was a great gig, but it was really physically taxing. We would eat not one but two foot-long subs every day for lunch. And I still have a fond memory of subs, because that was like the summer staple for two years. So that was the first job. And then I started my own businesses. I started DJing and other things.
Jason Baumgarten 04:29
Oh man, well, you got to drop a beat. Wait, so you said New England, where?
Paul Cheek 04:36
New Haven, Connecticut.
Jason Baumgarten 04:36
Oh, love it. Okay, cool, cool. I mean, New England for me is like shoveling the snow, right?
Paul Cheek 04:42
Totally.
Jason Baumgarten 04:42
More pizza, less donuts. So, Paul, we talked about the hacker in residence thing. All seriousness aside, what did they want you to do? What was the job?
Paul Cheek 05:07
I think it was one of those things, like on day one, people would look at me and be like, what's he doing here? And then I'd be like, wait, what am I doing here? There was this broad sense of what that role was. And it didn't have a title at the time. It didn't have comp tied to it at the time. Day one, I think it was kind of like, what are we doing? But the general gist was we wanted to build software that would help entrepreneurs at MIT get their start faster, more easily, basically set them up for success as they went to go turn an idea into a business, go commercialize their research, what have you. Because there's such a wide variety of entrepreneurial resources at MIT.
Paul Cheek 05:34
So let's build some software around it. That wound up turning into, over the course of seven or eight years, what is now the AI platform that helps entrepreneurs accelerate their overall process. But at the early days of it, day one, it was an idea that turned into a software platform.
Jason Baumgarten 05:56
That's amazing. I mean, the number of incredible companies that were born out of MIT that people don't realize. Bose Corporation, E Ink, the foundation of e-readers. The list goes on and on and on. It's really incredible.
Paul Cheek 06:22
And I think one of the things about that too, Jason, that's really special to me, having spent so many years working in entrepreneurship at MIT, is MIT's investment, not only financial investment, but investment in education. Not just to create those companies, but rather to create the humans that go on to build those companies.
Paul Cheek 06:23
And that's something where, like, I think it's very hard to put a number on success. But we felt we had to, because we didn't want to go off of just venture capital raised or things like that, which is also an important metric, of course. But let me ask you this. What percent of startups do you think are successful?
Jason Baumgarten 06:42
It depends what kind of startup. And going back to your book for a minute, startups in sort of the real world versus technology startups, but I'd say less than 1%. If you really look at all startups in totality, I'd put it at fractional small numbers.
Paul Cheek 07:04
Yeah, and there's a million and one surveys out there about that, and they have a wide range, but they're all generally low single digits. Right. It's a very small percentage of companies that get off the ground and are successful. So we ran a ten year longitudinal study of the companies that had come through our program at the Trust Center, which is MIT's entrepreneurship center. And what we found was that of the 181 companies that had come through that program, the accelerator program, 61% were successful. And what is success?
Paul Cheek 07:29
Right, the definition in that study was they were still active or had been acquired five years out from founding. And while they raised a billion dollars, we were super proud of all these founders for that, the thing that at its core that research study was designed to surface was not just how did the companies do. The question was, how did those individual entrepreneurs do, the humans?
Paul Cheek 07:52
Were they educated in a way that would make them better at the process every time they did it? And that's core to the MIT methodology and a lot of what my colleague Bill Aulet put together with the disciplined entrepreneurship framework. It's a question of what did those 692 individuals go on to do after they built those companies? They wound up starting 130 additional companies that raised not one billion, but two billion dollars. So when I think of that as the impact that those individual entrepreneurs can go off and take, that gets me excited. But I also, you think about, you've started some businesses, it's like, was the DJ business better than the painting business or the yard work business?
Paul Cheek 08:33
It's like every time we do it, we get better at it, right?
Jason Baumgarten 08:33
Totally. And sometimes, you said something in one of your own startups, you said the idea is worth almost nothing, it's the team.
Paul Cheek 08:43
It's one of those things you learn, is that so much of your own happiness, not just the financial success, but your own, would you do it again? Is like, did I have any fun doing it? Because people often ask me, is it worth it doing a startup? And I said, listen, you have to assume you will make no money. Because when you do a startup, if you go into it assuming you're going to be Mark Zuckerberg or Jeff Bezos or Bill Gates, you're kidding yourself. To your point, statistically, whether it's one percent or ten percent, if you look at the odds, you're going to be less well off than if you did a normal job.
Jason Baumgarten 09:11
But if the fun of doing it, if the learning of doing it, if the excitement of doing it is really high, going back to economics, your utils will be off the chart. And so you've got to account for that in your own personal economics. Because I do think some people, particularly when the economy's in a boom cycle, a lot of people got lulled into doing a startup because it was the way to get rich. And that isn't the way to get rich. The way to get rich is to go get a good corporate job or a good professional job and don't spend too much and sock it away. That's actually a much statistically better outcome. But if you really can look at it holistically and think about that learning as part of the fun, it's probably unmatched in many ways. And then that compounds, to your point, your next startup, you've learned so much, you're curious.
Jason Baumgarten 09:59
And I think with AI, what we're seeing interestingly is a lot of people who are, I'll call it not fully employed right now, I won't call it unemployed, because it's really just they're doing something other than a traditional full time job, are often much further ahead of the curve on AI knowledge and learning because they're spending almost all their time absorbing these new tools and technologies. And somebody who's sitting at even a very progressive technology company may be like, well, I'm using the two tools I'm approved to use, but I can't use it to do much, and I'm working 80 hours a week, so I can't spend that much time in, I'm trying to sleep and keep up with my kids in the side time. So I do think there's these interesting moments of learning as well that we all go through, almost lemmings on the curve.
Paul Cheek 10:43
You just touched on so many different things that are very central to my world right now, which I love. So a couple things I want to touch on there and follow up on. One is the question of, okay, personal finance and entrepreneurship.
Jason Baumgarten 10:58
Yeah, absolutely. Right. You can't go into it assuming you're going to make a ton of money.
Paul Cheek 10:58
But we also have to make sure that that business makes money, because if it doesn't, we have nothing more than a really expensive hobby, and that doesn't get us anywhere. And I always tell students on the first day of class, it's like, if you're here to profiteer, the door is just that way, because if we're thinking about it from a profiteering perspective, we give entrepreneurship a bad name. However, if we think about it through the lens of the company making money is its path to economic viability, amazing. But we also need to think of that as a reflection of the impact that the organization has had in creating value for others. So I like to think of it through that lens as it comes to the topic of entrepreneurship.
Paul Cheek 11:33
Yeah, tons of people are going out and pursuing entrepreneurship right now, which I love. And I think of entrepreneurial mindset, the ability to adapt quickly when faced with change, and systems architecture engineering mindset, as those things that best position the next generation of leaders, which I think we'll probably wind up talking a little bit more about. But when I think about it, one of the things that really strikes me is the fact that you're exactly right, you're on point. Anybody who works a full time job right now, they have projects, deadlines, deliverables, organizational annual goals, KPIs, metrics. That structure is a wonderful thing for all of us as humans to be productive in society.
Paul Cheek 12:16
But at the same time, that does limit our ability to justify taking time away from the day job to go and explore and learn this new emerging technology or the newest, latest and greatest emerging elements of AI. And I think what that means is that we are in many ways restricting the learning that is happening in these organizations amongst every individual employee. Which is why, when I go in and do an AI transformation program, my number one thing is time needs to be carved out by the CEO, because that says to everybody else in the organization, you can also carve time out of your day to go learn, which is something that you could do over a long period of time. But here, we cannot wait a long period of time, because that learning compounds, which I'm sure you're seeing too.
Jason Baumgarten 13:03
Yeah, absolutely. And the technology is changing quite rapidly. And so I'm sure you've had this experience where you tried something six months ago, a year ago, and suddenly you try it again and it either works or it works dramatically better, occasionally worse, and you go, oh, I can now use it for that thing I didn't think I could use it for, or I can try it in a way that wasn't successful before. So I think there is something to that. Before we get to the book, because I do want to spend some meaningful time there and some of the implications, you spend time at MIT, a place with a lot of PhDs. And I'd be curious, just in your own journey, when you're teaching, what do you think you're, and I'll use the word better, what do you think you're better at? And what do you think you're genuinely worse at? What do you bring to the classroom that somebody with a purely academic background doesn't? And what do they bring that you're like, that isn't my thing?
Paul Cheek 13:54
I think, historically, when I was teaching mostly entrepreneurship, I would be sitting there with a rocket scientist or a PhD with tons of years in machine learning experience. And I may not know everything about rocket science. I know quite a bit about machine learning. But regardless of what their domain is, brain, cognitive sciences, chemical engineering, one of the examples I love is a company creating what would go through a variety of blood vessels to get up into the brain, a thin fiber that can do non-invasive deep brain stimulation to help Parkinson's patients.
Paul Cheek 14:33
I know nothing about how that works, but I know quite a bit about how to think through commercialization of that, how to turn that into a business. And so I think of that process through the engineering mindset. My background, as somebody who understands software engineering, systematic approach, variables, like the scientific method and entrepreneurship, every element that gets a business off the ground running and then to scale, that's what I bring in. An ability to distill that in a way that people, generally speaking, actually enjoy. And I think a lot of that could be taken in a very, okay, now do this, but it's a much more flexible process than that.
Paul Cheek 15:13
So while, yes, I can get super curious about rocket science or deep brain stimulation through non-invasive fibers, excellent, that's not my specific domain. So what I think I do is I make entrepreneurship accessible to those who maybe wouldn't have thought of themselves as entrepreneurs necessarily coming into it. And today, I think the other thing, just for your context and for others listening, I spend a lot of my time working now with C-suite executive teams and boards, thinking about how do they run an AI transformation. And I do this in the classroom at MIT and outside. It's like, how do they do that when they're managing teams of tens of thousands, hundreds of thousands of human employees? Meanwhile, my entrepreneurship students are sitting on campus right now in Kendall Square and they're building companies from scratch that are stealing market share away from those S&P 500 executives, building companies that don't really have huge teams. They have huge teams, but they're huge teams of AI agents.
Paul Cheek 16:10
And so when I take those two perspectives and those two vantage points, what I think about is how does that relate back to what needs to happen in a large organization. And so that perspective for either one is something that I think I have this, a little bit different than a lot of other experiences out there, is what do those two different situations look like?
Jason Baumgarten 16:40
Yeah, I love that, and I think it's so important, and we'll touch on governance in a little bit. But one of the things that a lot of boards are stressing about right now is the risk of AI. And I think for so many, it's also the risk of not AI. Right. Like, what is, to your point, that startup that isn't worried about all those things, they're just going to focus on how to steal market share. They're not going to focus on, boy, I've got some IP risk, or I've got some data risk. Now, it doesn't mean throw the risk away, but you have to look at risk as a positive and a negative. Somebody asked me the other day, I had the privilege of working with Elon Musk for many months in the earlier days of Tesla, they were presuming what one of his great skills was. And I said, I think one of his great skills is risk taking. I think actually his most defining skill is risk taking.
Jason Baumgarten 17:16
He may or may not agree with me, but my learning of watching him operate was he was willing to take risk in ways other people were not. And that in and of itself is a really profound journey. Because I think people often assume it's like, well, I need the great idea or I need the great thing. It's like sometimes you just need to be willing to say, the whole way we do this is crazy, we should throw it out and start again. And that's something that most people aren't willing to do.
Jason Baumgarten 17:40
And you know, I know you touch on that in the book. But let me get to one of your central premises.
Paul Cheek 17:54
And I love it that you have that and you're able to refer back to what's in the book. You picked it up, you read it, that means everything to me. I want to say that up front, like, so important. And the Elon thing, his ability to take risk, I love, because I teach a case on the Cybertruck launch, I've taught it countless times, focused on the entrepreneurial experiment they ran. But one of the biggest things that always comes up is the risk that was taken to throw the ball at the window only to find out that it actually broke, even though it wasn't supposed to. And that's one of those things I always take a sidebar whenever I teach that case, whether it's to undergraduate students or to executives or PhDs or whatever. And I say, look, yes, we're going to talk about the entrepreneurial experiment here, but I also want to flag that he just sent a very strong signal in doing that, that risk taking is acceptable in this organization.
Paul Cheek 18:33
By taking that risk himself, he set the stage for everybody else that works with him. And I think of that as being so important, especially in cultures where innovation is less of a day to day activity, or risk taking is less of a day to day activity. To drive entrepreneurship within organizations, we need that spirit that risk taking is acceptable. So I love that you bring that up.
Jason Baumgarten 18:52
Yeah, and I think one of the things you touch on in the book, to link back to that, is I think this notion of risk taking and entrepreneurship is not always about a new business idea, but about a new way of doing something. And I think with AI, so much of the transformation isn't we're going to have a wholesale new business, but we're going to reinvent the way we do business and be willing to take risks in it, as opposed to, and dare I use the old re-engineering the corporation language, but we're not going to pave the cow paths. And I think that so many people are like, we're going to automate the cow paths. And I'm like, if the cow paths are in the wrong place, start there, right? Start with the fact that you've got them in the wrong place.
Jason Baumgarten 19:32
So in the book, which I just finished, and we'll get it in frame, remind everyone when it's coming out.
Paul Cheek 19:48
By the way, August 25th, coming in hot, and I can't wait, there is so much energy behind this book launch. I just can't wait. But the thing I'm most excited about, and the reason I wrote it, because initially I wasn't going to, was getting into the hands of people who can really change the way that organizations run. You refer to it as the re-engineering, or like not redoing that, but I think of it as organizational redesign, which is, in my view, a requirement for the AI era. And the big reason why is because now the actors in an organization are no longer just human. Now we have AI working in a probabilistic manner, taking action on behalf of the organization, representing the organization publicly.
Paul Cheek 20:22
That is a risk if not properly governed, managed, and that literacy doesn't exist. And that's really what I want to help make sure that every executive, every board member is thinking about as they move forward with their organization's AI journey.
Jason Baumgarten 20:46
It's such a profound thing. And you talk a lot about your employees being both human and agent. I also like to remind people that humans are not perfect. And I think this is a distinction we see in cybersecurity. I remember years ago when I was in my consulting career, we were running some projects for a very well known organization, and they were very concerned about cyber risk. And we ran a simulation to show them that actually just robbing the mail in a systematic way would be cheaper, easier and more foolproof than trying to hack their incredibly complex systems. And everyone ran out of the room kind of going, oh my god, oh my god. I was like, yeah, it turns out you pay a teenager twenty bucks to steal the mail.
Jason Baumgarten 21:21
It's pretty easy. And you can multiply that by hundreds of thousands of people a lot easier than you can find the one hacker who can get through your system. And so I think sometimes when we make a digital transition, we up the bar, and we're seeing this in self driving, we're seeing this in robotics. I mean, I can't throw a ball very well, so when I see a robot fail to throw, or I'm an okay dancer, so when I see a robot kind of okay dancing, I'm like, hey, it's like me, we're good. When I see somebody say, well, AI agents fail ten percent of the time, I'm like, how are you as an entry level employee? I suspect worse. So I think that one of the things we all have to do is, I'm not saying lower the bar, but appropriately adjust the bar to the training we've put into it. When you have an early employee or an intern and you give them moderate or no training and ask them to do something and they get it right half the time, you'd go, that's okay.
Jason Baumgarten 22:13
If you have an AI agent that you spent three seconds giving a prompt to, with no accoutrement to the tool, why do you expect it to succeed? And so I think that's one of the things we all have to get better at, which is, if you put hundreds of hours into a human employee, how many hours do you put into your AI employee? And if the answer is like three minutes of a bad prompting class, you failed it. You failed it, not the agent failed, you failed it.
Paul Cheek 22:47
I love that you say that, because I run this executive simulation that tries to get at exactly that, where there's two hundred twenty-five million dollars on the line, you have an angry customer whose name is Maximilian. Maximilian is upset because you sell odeur, the scent of gold, to him for his fragrance house. But you also have an industrial SKU that industrial SKU has another name in French that translates to new car smell. When it gets in his scent of gold, he's upset.
Paul Cheek 23:13
And we add additional context into the system from your organization's mock data, the internal team chat, the brand guides, the internal wiki with all the policies and procedures, and then ultimately your email. As we add more context, the responses get significantly better, much like you would expect of a well trained employee, until it doesn't. And coming back to your cyber point, you add in your own email inbox and a cyber threat that we all assumed was addressed at this point in time, phishing emails all of a sudden lead you to draft an email back to Maximilian that says, we would like to approve your two hundred twenty-five million dollar refund and offer you a two hundred fifty thousand dollar courtesy payment just for your troubles. And one of those things where we actually live, feel and see the way the AI systems are, with increasing context, improving the quality until they don't.
Paul Cheek 24:06
And feeling that with so much money on the line is something that I think a lot of individuals have yet to feel in the high stakes nature of many organizations, especially those in regulated industries. And that's where I think there's a lot of opportunity to create those real experiences that don't put everything on the line.
Jason Baumgarten 24:31
I think it's so important, and cyber is one of those things. The cyber risks of a human and the cyber risks of an agent, it's much easier to put policies and procedures in for agents, as long as they are done in ways that can see their way through, not just a probabilistic framework, but also the deterministic guardrails that actually do prevent agents from doing things that are rogue. I should tell you about my rogue agent simulation sometime, which, it's wild, it's amazing what happens when we have to react to an agent that's gone mad.
Paul Cheek 24:57
Yes, and we should do the rogue employee simulation at the same time and see whose blood pressure went up more.
Jason Baumgarten 25:04
So one of the core concepts of the book is humans going from doing the toil of work. We all had that early job where you cut and paste data from one place to another, or you're doing something that you really wish you weren't, to humans designing the intent of work. Walk me through, when a leader really believes that, when you work with somebody and they really go from a fuzzy concept of that to truly embracing it, what's a concrete thing that they do differently the next day?
Paul Cheek 25:58
Yeah, well, and see, this is the thing that's so hard is, like, changing the habits and culture within an organization to enable that. Because even if we totally believe it, actually putting that into action is extremely difficult. And it takes a very different set of transformation guidelines for an organization. I share that up front because what I see is very similar to what you would have seen from the first person who ever wrote an email.
Paul Cheek 25:59
They sent an email one day, but then the next day they go back and they start writing a handwritten letter that they put, and send off in the mail. Because changing the habits is the really hard part, to go down that path. So much so that I literally built a tool for employees that monitors what they do on their computer, and anytime they start doing the data entry, the copy and paste, it flags. It's like, hey, that's something that you could be doing with AI to save yourself time.
Paul Cheek 26:23
And it actually puts a dollar value on it so that they see that and they think, oh my gosh, I'm actually going against the habits I'm trying to build.
Jason Baumgarten 26:29
Right, kind of like a coach in the gym. It's like the old fashioned rubber band slap when you use a filler word. I don't know if you ever, that was like an old fashioned speaking trick, when you'd wear a rubber band and every time you used a filler word, you'd snap the band. In a few days you're like, no filler words.
Paul Cheek 26:46
I love that, I love that example. But no, a concrete example, I want to come back to your question, but I think it's important to flag that up front. The simplest example would be, whatever KPI they have in front of them. Now, I would also say many people jump to something that is not actually an organizational priority. And we could talk about the four characteristics that I look at for any opportunity to use AI.
Paul Cheek 27:05
But I see a lot of people who look at whatever that very clear KPI is in front of them, and so it might be, let's just say revenue, right? Company revenue overall, we need to increase revenue. It's setting the intent, designing that goal, because that's how an agent works, is we define a goal instead of just writing a prompt. And that goal is something that an agent will perceive, what's going on around them, the scenario, the environment within which they're operating. Think current revenue, think areas of the business that have increasing product revenue or decreasing product revenue, business units or business lines that perhaps have margin that is either increasing or decreasing over the course of time. It might take in all the sales data, it might pull all that information in, will perceive it, will then plan, think about, okay, what should we do if we wanted to increase revenue?
Paul Cheek 27:53
It might act by going out and actually saying, we need a sales enablement plan that embeds our sales workforce with AI capabilities and AI literacy that's going to enable them to build agents to be more effective. I'm making this all up, but you get the idea. And then it's going to come back to us with a real output of work, which is they designed a sales enablement training plan. That sales enablement AI training plan is now in the hands of the person who runs sales enablement. So as the CEO, we're now going to take that output of finished work.
Paul Cheek 28:22
We're going to be able to report on that to the head of sales or our CRO or whatever the case may be. And so they're going to go to their KPIs first. But I would also say that I see a lot of individuals go to things that are really impactful for them personally, but that may not actually move the needle for the organization at large. And I say that because when you're enabling folks throughout the organization, when the cost of creation has dropped so significantly, where anybody can create whatever they want, what you wind up with is the strategy of the organization and the tactics of each and every one of the individual employees diverging, when in reality, if we set each employee up for success with the proper AI governance and AI systems and AI data, they should be able to much more easily achieve the organization's strategic goals. So I look at what people jump to, and I often think, that seems great for you.
Paul Cheek 29:11
What does that mean for the organization at large? And I think that's such a common first step that I see as a mistake. And if I might add one more, and then I'm kind of curious what you see when you would look at that and work with companies. It's like, I see a lot of folks jumping into AI strategy, or I see them jumping into data cleanliness or compliance or governance, when in fact the first thing that everybody should be doing is investing in AI literacy, because we can build the fastest AI strategy, but if we don't have real AI literacy, we don't understand the ways in which AI actually works, then that strategy, that's kind of like garbage in, garbage out.
Paul Cheek 29:48
And so I think about it a lot in that context, which is why I like to not jump so quickly to what should we do differently and what intent should we set. It's like the literacy is the foundational part. You think about that, and even maybe it's in your own role, or in some of the clients you work with, or what have you, how do you see people setting intent, or do you, or maybe that's not something that you have exposure to quite as much, I don't know, I'd be curious.
Jason Baumgarten 30:07
Yeah, it's pretty all over the map. I mean, I think that many are doing a lot personally because it's where they can have the most freedom to create, because they don't have IT constraints or risk constraints. And I think this point of people are becoming more productive holistically. But your comment on, to get the organization more productive, you actually need much more orchestration work. And that's why we're seeing it in certain functions like engineering, customer service, where they're very paved systems, very locked down, the rules are very clear. Imagine if instead of having normalized computer language, every company had a different computer language, programming language. That's the world of business, right?
Jason Baumgarten 30:54
Because even something like recruiting, the way we do recruiting, or even the way I do recruiting, may be very different than the way a thousand of my clients do it. And what I have to do is say, okay, how do I, or how do my clients, when they're looking at AI problems, how do they get that multiplier effect, so that they're not just getting the supercharged human, which I think is a bit of an overused concept, and really getting supercharged teams. Because I think what we're seeing with AI is exactly what you said, getting these individual high node behaviors, but the divergence from strategy, I thought that was really eloquent. Strategy, the alignment to strategy is waning, because it's like, well, if I can now score the produce of every grocery store I'm in, I'm going to be like, great, I'm taking photos of all the produce and I'm working on this cool website and it's scoring the quality.
Jason Baumgarten 31:44
Who cares if that's not part of my corporate agenda? If the agenda is sell more bananas, because that's the highest margin thing, then what I really want to do is figure out how to sell more bananas. And I'm sitting there figuring out whether the mushrooms were high quality this week. So I think that requires more work, though. And I think that's where what we're seeing is that there are examples of real adoption.
Jason Baumgarten 32:08
And this is one area of the book I pause hard on, you have a great chapter, WTF do I do with AI, and it's where I see a lot of people getting stuck, because they're doing the automation task, that's easy, and it's driving some productivity gains.
Jason Baumgarten 32:26
It's real. And then they're stuck, because they don't know how to go from that automation of the thing to any kind of transformation of the way they're working more. An SDR in sales is a great example. So they take their outreach strategy for customers, they create a digital SDR, they empower the SDR, they tune it, they have this great, and then they're like, great, now what?
Jason Baumgarten 32:48
Your revenue might have actually not improved, because you just introduced a bunch of low quality leads to your system, but that requires data you don't have, or something else, or your marketing materials are terrible. And so I think this idea of how do you create that cross disciplinary team effectiveness is still hard for a lot of organizations.
Paul Cheek 33:13
Yeah, it is, it's so hard. And that's where I see the major unlock right now. There's a lot of companies that have invested in what they think is AI transformation, but really it's just AI tools, deploying AI licenses and AI training. But those two things together do not necessarily equal AI transformation. The transformation that's required for an organization to see, not just a ten percent productivity gain, rather what they really want is the ten x, not the ten percent, ten x efficiency boost that sets that organization up to be competitive in the forward looking marketplace.
Paul Cheek 33:41
Which I worry about a lot, because I think a lot of people are thinking, oh, people who use AI are going to replace people who don't in the workforce, which I generally agree with, that's a takeaway that a lot of folks have from executive sessions that I do. But I think the real risk, the real threat, is that an organization that deeply embeds AI in their operating model will replace organizations that do not, not in the workforce, but in the marketplace. That's the real fear I see.
Jason Baumgarten 34:08
And one of the things that I think is a real challenge is that many large established firms do not have, in many cases, leadership that are as well incentivized for long term strategic production compared to the short term, what's the stock price. And I think that's something that becomes really important as we look to the rate of change in the marketplace. And one of the questions I often ask people is, what percent of the S&P 500 or Fortune 500 that is on that list today, do you think will be there in twenty years?
Jason Baumgarten 34:37
And I get answers that are all over the board. I got a lot of optimists and a lot of pessimists. But regardless, the thing that I ultimately come back to is, at the end of the day, it requires both individuals to do things differently and the organization to do things differently. So when you say, what value gets unlocked when the whole team or the whole company or every employee in the organization is leveraging these AI agents, these data assets and AI generally, I think I see a real unlock when organizations go from, I call it the state of the art org chart, which embeds both human nodes and AI nodes in the org chart, because that's how we've organized historically with just human actors. But now that we have AI actors, we need to organize in a similar fashion.
Jason Baumgarten 35:17
They need to be embedded in the org chart. But the real unlock that I see, in terms of that, like ten percent productivity boost to the ten x efficiency gain for the organization, is when we go from what I refer to as the personal swarm. And Jason, that's like you using a bunch of AI tools or AI agents, or me using a bunch of AI tools or AI agents, to shared agency, where the team is collectively working with an operating model that is very different and looks nothing like the established firm and the playbooks that many leaders have run for years and years and years. It looks much more akin to that of an AI native startup, too, which I also think is one of those things that you kind of have to experience, which is why I love doing the thirty minute startup exercise, where I have people literally build an entirely new company using a variety of AI tools in thirty minutes. You see that, and you're like, oh my gosh, wow.
Paul Cheek 36:07
The way we operate is nowhere near the speed with which a new AI native startup is emerging, which is so exciting, it gets me going.
Jason Baumgarten 36:40
No, totally, and I think that, part of that's also the way in which the paradigm of a lot of the AI tools was really Google, in terms of, we have a search bar, we're prompting, we're asking questions. I think what you're going to see is much more collaborative tools, where it's, hey, we're all going to be interacting with these agents and using these agents the same way that if you and I are working on a project, we don't go into our silos and then come out and go, look what I finished.
Jason Baumgarten 36:41
We actually work on the project, we break it up into pieces, we pull it back together, and those tools are kind of emerging, they're not quite there yet. I will say that, to your S&P 500 question, if you look at the lifespan that we're at, at the moment, we're in the Bronze Age, we have a corporate life expectancy that would put us in the Paleolithic to Bronze Age of humans, which is not a great place to be, I'm pre-penicillin, so we gotta get a little improvement if we want these entities to last a little bit longer.
Jason Baumgarten 37:10
Paul, there's one collision in the book where I did have a moment of pause, which is, so much of the argument of this show has been around fit, right? And when you find the thing that you're great at, it is much more personally fulfilling, but you're also much better at it on a relative basis. It's the individual guns and butter exercise. Because, you know, I wasn't destined for the NBA, I suspect you weren't destined for the NHL.
Paul Cheek 37:37
And so we all have these, you never know, you never know, it could happen. I know, an agent might help you.
Jason Baumgarten 37:44
I'm going to posit a guess. But we all have these moments where you realize, hey, I'm really good at this, and I'm kind of not as good at that. And self awareness comes late to many of us. But this idea that fit is important in our own human lives, both personally and professionally. One of the things I want to get to is there's this risk that people take away from AI, that it's not about having skills, but about this generic operator talent.
Jason Baumgarten 38:09
You know, I'm just going to be the generic operator, Mission Impossible style, moving stuff around the board, and that, along with a lot of smart agents, will beat things. I'm not sure I'm all in on that. Talk about that need for still having real human fit.
Paul Cheek 38:25
Yeah, oh my gosh. Well, I think the human fit is ultimately going to be one of the most important things. And I think when I look at the role of an organization, and whether it's a one person company, a two person company with a bunch of AI agents, or a zero person company, even what I talk about in the epilogue, a company that is operating autonomously with no humans. And I think that's important, by the way, for established firms that have plenty of human employees. Because I am not of the mindset that we should be replacing human jobs with AI.
Paul Cheek 38:54
I am of the mindset that we should be enabling every individual in every organization to do the work that is most impactful and will bring them the most joy. What we talked about earlier, if we're not doing that, then we are doing the humans in the organization a disservice, which I am fundamentally against. When I think about that fit, I think about humans finding the things that will ultimately be impactful, that they enjoy doing. But how do we supercharge each and every human individual to do more than is reasonable with the resources they have control of?
Paul Cheek 39:22
Something that we always define in entrepreneurship, that is basically an entrepreneur, somebody who is doing more than is reasonable with the resources they have control of. So I think a lot about that, I think fit is really important. But I also think, if I think broad strokes in the economy at large, there's a lot of value in thinking about what are those companies or jobs that AI could do on our behalf, or even better than we can. And then, where does that free up human talent to do the things that they care a lot about, that are far more impactful, that AI is not going to solve for us.
Paul Cheek 39:58
AI might help us in that, but so I think a lot about, can we enable AI to do more of the work? If you've ever seen the HBO documentary on Theranos, you'll know that they describe Theranos as part of either the chat app companies, or the companies that are solving what I think of as MIT-size problems, think autonomous driving, healthcare, a lot of stuff in the life sciences, the development of new drugs and therapies. I think that AI could do a lot of the management, and this is meant to be a generalization, going back to the documentary, but AI could run companies that are the chat app companies of the world, in some regard.
Paul Cheek 40:37
There's still a need for human creativity and judgment and communication and critical thinking in all of those organizations. But I wonder, could AI help do a lot of that, which would free up human, and our overall collective workforce, to focus on the things that really matter, to focus on the things that ultimately AI is not going to solve for us. Those things such as life sciences, climate change, things like that, that really do require a lot of human attention, if we are going to solve and get right as a population. So I think a lot about that. And ultimately, what I want is I want every individual in the workforce to work on the things they care about.
Paul Cheek 41:13
That's what it all comes back to. And that's what, when we teach entrepreneurship, we always say, start with the thing that you are so obsessed about. Find the thing that is going to get you out of bed before the alarm clock goes off, every single morning, for the next seven to ten years of your life. Because if you do that, wow, you will wake up even when things are not going well, and you will be ready to go.
Jason Baumgarten 41:34
And Paul, that is the premise of the show, which is when you find that fit, everything's great. I just got back from a trip in Italy, and I had this moment that I've thought about a lot. If you've read about Da Vinci, one of the things I always reflect on is Da Vinci was such a broad thinker. But part of the reason we know about him, and we know about how broad his thinking is, is because he could draw. And I think how many people of that era who couldn't draw, but who could envision the greatest weaponry, or the greatest transportation machines, or the greatest whatever, were limited because they didn't have that technical skill of drawing.
Jason Baumgarten 42:25
And this idea of the human capacity to create, versus some of the tactical skills, whether that's, oh, I can't use CAD-CAM, or I can't do the advanced math required to design a rocket. But what if I had an idea about a rocket that nobody's ever had? If I can now use AI to augment the piece of that I don't have, can I actually do something that's truly groundbreaking?
Paul Cheek 42:59
So I do think there's something powerful in what you've just said.
Jason Baumgarten 42:59
All right, so we talked a little bit about what changes when a leader actually believes human intent is more important than toil and cutting and pasting cells. And one of the things you touch on in the book is the human manager, I won't just say CEO, but the human manager who wakes up every day, checks the numbers from the day before, delegates some work, sets up some meetings for a few weeks out, has a lot of friction in all of those things, right, and a lot of time.
Jason Baumgarten 42:59
And then the digital manager, the digital CEO, who has no friction, who is running that loop constantly, all night, on the most recent data, not waiting to coordinate a meeting with other agents. One of the things I was thinking about is what does the human do in that context that the non-human can't do? What is that boundary that we need to preserve, that we need to think about?
Paul Cheek 43:40
When I think a lot about that, whether you're talking about a team or a company or what have you, I think what ultimately becomes the most important element is something I talk about in the multi-layer latency stack, which is the governance layer. I think governance is the most important human role moving forward.
Paul Cheek 43:41
Because if we set the intent, and we allow AI to determine what needs to happen next, do everything that an AI agent from a technical perspective is set up to do, our job becomes to govern, to, amongst other things, manage risk, right, the governance layer. We talk a lot about ensuring safety, ensuring compliance, ensuring that we're aligned with regulatory or other ethical considerations. So I think human judgment becomes the number one consideration, the number one thing that is required. I am very much of the mindset we should offload work that can be done by AI to AI, where it makes sense, provided the context of an organization's core values, what different stakeholders would expect from us.
Paul Cheek 44:19
For example, what do our customers want, what do regulators want, if we're working in the financial services space, for example. But enabling us as humans to do the thing that only humans can do, and that humans should do, leaving everything else to AI. Humans most certainly need to govern AI. And I think a lot about this within the context of what the workforce could look like over the course of time. So if you think about it, right now, any one human can only reasonably manage so many human employees.
Paul Cheek 44:49
Each individual human could have a certain number of direct reports, but I think one of the misconceptions I see is a lot of people think, oh, well, I can only manage as many AI agents as I do humans. And I think of it as quite the contrary. Any one human, and this all depends on agent quality and architecture and governance and observability and logging and audit trails and all that, but any one human employee should be able to manage far more AI agents than they could human employees. And when that's the case, what we ultimately could wind up seeing is an environment in which the workforce is far more AI agents than it is humans.
Paul Cheek 45:23
And I think we will see that very soon. But that's not because we've taken humans' jobs away, it's because every individual human should be able to manage far more AI agents than they can humans. And so with that context, I think what becomes necessary is the development of those AI agent governance skills, those AI agent management skills, that enable humans, regardless of what role they're in, to manage a fleet of AI, or a swarm of AI agents, a network of AI agents, which is not the same as people leadership, it's a little bit different.
Paul Cheek 45:54
So right now, the best AI agent managers are what we have historically thought of, and currently think of, as individuals who have systems architecture and software engineering roles, or backgrounds, or skill sets. Well, I think those roles will change. The jobs will certainly still exist, which is why I put a lot of emphasis on the roles that many companies are kind of laying off, or not hiring as aggressively right now, which are software engineering roles. Because that's what AI has taken first, in the sense that it could generate code. But if the vast majority of the workforce in twenty years is AI agents, then all of a sudden we need people who have the skill sets to manage AI agents, not just manage people.
Paul Cheek 46:35
So I think of software engineering skill sets, as we think of them today, as being critical to any organization moving forward into the future. Governance ultimately becomes the number one thing for humans to take care of.
Jason Baumgarten 46:35
Paul, it's so interesting, I had this passing thought when I was reading this section of the book, where I thought, so much of the narrative right now is kill middle management because of AI. And I thought, actually, in some ways, it's the good middle management that has become so critical with AI, and it's not the skill most companies develop.
Paul Cheek 47:08
Well, right, because if you think about what leadership, a lot of people get promoted not because they are good at making decisions and creating clarity of work, but because they are good at getting people to feel good, right? And we've all had that manager where you're having a tough day, or they told you to do something that made absolutely no sense, and you go to talk to them, and at the end of the conversation you leave feeling good. You're like, god, I feel so, I love Paul, Paul's so great.
Paul Cheek 47:36
And the next day you wake up and you're like, but I still have no idea what I'm supposed to be doing, or I still think that was totally idiotic. But they lull you into complacence, that somehow it was a good idea. And I think one of the things we're realizing is that when you have the management of AI agents versus humans, that becomes a truly unnecessary skill for the agents, right, the agents don't care if they feel good.
Paul Cheek 48:00
In fact, there's funny memes online about telling your agent, thank you so much, and how many tokens it cost, which is bad for the environment.
Jason Baumgarten 48:09
Yeah, exactly, bad for the environment, yes.
Paul Cheek 48:10
But it's this interesting question of redefining leadership, which at the core, and I actually, I would disagree slightly that software engineers are the direction. I think it's people who can disassemble work. And you make this point of jobs versus nodes, disassemble work into the component pieces, based on what the right objective function is, and provide really clear instruction as to how to get that done, and what does success look like. And when you look at a lot of leaders, our human nature doesn't judge them on that ability, it judges them on whether they made us feel good after telling us something that made absolutely no sense.
Paul Cheek 48:47
And so I think that it's going to be really important for us to all think about, is a manager, and again, I'll use that term at the most basic level, because to your point, it could be a manager of a thousand agents and no humans, is a manager good at understanding the job to be done, like the what, what do we want to get done?
Paul Cheek 49:06
Can they break that down into the component tasks and work components required? Can they measure whether that was successful? And can they pivot when it wasn't, to provide either clear or different instruction? And it turns out that's what most managers are not trained to do.
Paul Cheek 49:24
They're trained to do a whole host of other things. And so I think that it will shift what we believe is inherently good management, and this sort of cult of personality of management will become deeply less compelling.
Jason Baumgarten 49:43
Plus one, retweet, I love it, I love it. And so when I think about that, what you describe as the ability to break things down into smaller pieces, to define the different functions or jobs to be done, things that need to happen, and then to orchestrate them so that they do occur, we measure the outputs, we refine, what you described there, what in some regards should make a good manager, or at least a good manager of AI agents.
Jason Baumgarten 50:24
A good manager is something that right now, and it's really interesting, just like to hear your perspective on this, is something that software engineers are doing on a daily basis, except they are not doing it for other humans to do, they're doing it for historically deterministic code to accomplish. And they are, in some regard, evaluated on whether or not that code does what it is supposed to.
Jason Baumgarten 50:24
And I don't want to say it's exclusively software engineers or whatever, but I do think that is really what they are evaluated on. While people managers are evaluated, as you call it, do they make people feel good, and do they motivate people and lead, which is also extremely valuable, of course, because I don't believe we will ever have an economy that does not have human work, that will still always be a really necessary skill set. But I think it will be not necessarily sufficient, in that they also will need to manage AI agents in that overall architecture of the firm. So I think of that as being really, really important.
Paul Cheek 50:56
Yeah, I don't know, I can't wait to step back from this conversation, reflect on what you just shared, because I think that is an important distinction. And one of the questions that I love asking people is, do you remember your favorite manager, and do you remember your least favorite manager, and what would make the best possible manager for you? And I think it is a combination of humans that understand, make us feel good, and that mentor us and coach us and give us that desire to do more and to learn and to grow and develop professionally. But there's no world in which any human manager could ever be looking over your shoulder twenty four seven.
Paul Cheek 51:31
They couldn't know everything that you've done, and coach you on every single thing with all that data. When I was asked this question recently, about somebody I was hiring who said, what I'm really looking for is really good mentorship from somebody who's really senior and this, and I said, that's interesting. One of the things that I would encourage you to be thinking about, in your job search or in your whatever comes next for you professionally, is not only who would that human mentor be, but what does the organization you're joining have, in terms of infrastructure, to set you up for success, in terms of the always on, always watching mentor AI, who may serve in different ways as a much better mentor than your human mentor, that can really be taking in all of the data about every decision you made, every action you took, every way you communicated something positive, light, and a negative, whatever the case may be, that can be there forever, and see everything.
Paul Cheek 52:22
Because if that exists, there is the potential that you will accelerate professionally much faster than maybe even you imagine. And so I think about that a lot too, in terms of the dual human and AI mentorship, coach, manager, leader roles. What happens when you combine those two?
Jason Baumgarten 52:44
That's really powerful. Well, going back to your rubber band tool, I mean, that is, I think we're going to see that, I don't think we're seeing a lot of it yet, but I think we're going to see a lot more of it. And also the boundary of human capacity for ingestion, because I only have so many hours a day for somebody to tell me I can improve, at some point I'm like, okay, I suck, let's get over that, let's move on. I don't want that all the time, but I think we should all have that ability.
Jason Baumgarten 53:03
You know, there's a dial on our computers, so to speak, that says give it all to me, or give me ten percent, I'm having a rough day, I didn't sleep well.
Paul Cheek 53:23
Well, and I think that's the difference between the day to day, and if your manager, human manager, is telling you, human employee, every single day, you're doing this wrong, you're doing that wrong, you're doing that wrong, like all of a sudden your motivation is going to diminish significantly. You're not going to want to do anything or say anything, because you just know where it's going to lead you. But that's where the collection of that performance data, what you've communicated, what you've done, what decisions you made over a time period, to help inform, let's say, a performance review, a performance evaluation, for example, can actually be really helpful, as long as it's not all delivered the second it happens, right.
Paul Cheek 53:54
They take the rubber band out of it, or maybe turn the rubber band into a catapult, or something that, at the end of the quarter, or the end of the year, it's like the dunk tank at the state fair, but you only do it once a month.
Jason Baumgarten 54:24
I do think that there's this issue around winning that we don't address a lot in the workplace. And with startups you have this vicious fruit fly cycle of, if you don't win, you go out of business so quickly. In a lot of organizations, when I ask people, going back to your question about who is your best boss, there's usually two reactions. There's somebody who helped you grow a lot, and then there's somebody where you won, where you were part of a winning something, and you felt like you didn't just win, but your role in winning was really clear, right?
Jason Baumgarten 54:24
You were a part of it. And I think going back to this question of what makes a great leader, so often it's not just they made you feel good, but they made you feel like you were part of the winning, and you won. Because if you didn't win, it turns out at the end of the day, I'd rather a rough ride to winning than a fun, wonderful ride.
Paul Cheek 54:48
Oh yeah, it is not good.
Jason Baumgarten 54:48
It's like blackjack, you get great drinks and you lose all your money.
Paul Cheek 54:54
I'd rather be taught, count the cards, suffer, do the math.
Jason Baumgarten 54:54
One of my favorite professors at Stanford wrote a book about friction. And you sort of allude to this in latency and friction, that AI takes friction out of organizations, right? You don't set up the meeting in three weeks, you just get it done. What is something that you've seen already where you're like, that, in most organizations, let's get rid of the friction right now, zero it out, use AI. And what's something we're like, where you've already seen, no, no, no, keep that friction in there, you need a lot of it?
Paul Cheek 55:33
I mean, the first one, meeting notes, meeting notes, action items, follow-ups. The process of somebody walking out of a meeting, having taken copious notes, a list of action items, and then individually transferring that to their personal to-do list, to then maybe get around to it at some point. Now, I can't imagine that, I truly cannot imagine that anymore. I have a system I've architected that takes, well, off the shelf tools, take the meeting notes.
Paul Cheek 55:56
That all gets ingested into one central knowledge graph, and walking out of it, any task goes directly into my to-do list. In my to-do list, every task AI is evaluating to figure out, can it do it by itself? Does it need my input, to set its intent, its direction? Or is it something that I need to do personally, because it requires human judgment, human execution, something in the real world or otherwise?
Paul Cheek 56:21
And that level of friction and latency, it's like I should not take a week, because there's other things going on, other projects, deadlines, deliverables, what have you, to follow up on an action item that I could do in an instant. So that's one that I see latency decreasing on significantly. The follow-up emails, done. The work plan or the document or the ad campaign is launched, done, don't even have to think about it.
Paul Cheek 56:47
The things that I really feel do still require that kind of human judgment, the number one thing, the one that's most obvious to me, is a lot of what you do, recruiting, hiring, bringing in the right people into the organization. And I say that because I think there's a lot that needs to be done to build that human connection, to build that human trust. By the way, I did not pay Paul to say that.
Jason Baumgarten 57:09
I did not pay Paul to say that, I'm just putting it out there.
Paul Cheek 57:09
Hey, I didn't even know that question was coming. I can see the argument on both sides, because I do see a lot of companies doing the dual agent interaction in recruiting and hiring. I think that's great to an extent. But ultimately what I've also seen and heard, plenty of stories of, oh, that's actually just producing wildly unqualified.
Paul Cheek 57:27
Not to say that I don't think that the AI agent recruiting model can and will work, but I think at the end of the day that is something that will always require a human in the loop, because that's something where you are really, what you're doing is building trust in both directions. Human trust gets built when humans interact, and that comes over the course of a period of time. A storyline gets built for either side of, here's what this can look like, here's why this is a fit, here's why this is not a fit. AI could maybe help accelerate some of that process, it can help in that process.
Paul Cheek 58:00
But ultimately a human should be joining an organization because they believe in the leadership and the mission of that human leadership. And also on the company that's recruiting side, they should be building a lot of conviction that that human is going to add so much value to the organization, not just financially adding value, but culturally, energy, all of that should be very clear to both sides. And I think that comes through the way in which trust is built over the course of time through those human interactions.
Jason Baumgarten 58:29
I agree, and I don't think AI will not have impact on recruiting, but I do think that the way in which we as humans build relationships is very unique, and it's not just in our work lives. I think that'll be true in any professional pursuit, actually. I'll ask you a question, if you think outside of your work life, to the two or three biggest accomplishments, I'll leave it very broad, in your life, one of the points that I always try to make to people is that you needed people to get that done.
Jason Baumgarten 58:59
And your book's a great example. I was really touched by how many people you acknowledge, the pre-readers, all the people that helped you bring it to fruition. And there's probably great recruiting stories in that, where you had to go find the right person for the right part of that journey, where I don't need somebody to read it for this, I need somebody to read it for that, or I need somebody to help me pick the title or whatever. So I'd love to, if we go a little off topic for a moment, given you open the aperture to this human nature of recruiting, is there a fun story from just an accomplishment you're really proud of, where you had to pull somebody along quickly touching on the book?
Paul Cheek 59:40
It's like everybody always asked me, oh, did you? And I was like, no, I think I will actually be the last person to ever write a book without ever having used any AI tool in the writing process. That was for my last book, for this one, of course I used it to help me.
Paul Cheek 59:50
But what I always tell people, because they're like, oh, an AI book, you must have written the whole thing with AI. It's like, okay, I took all my thoughts, put them in bullet form, super draft, super rough. Use AI to help me actually construct it into something that is really cohesive. But the real value came when I took the whole thing in a Google Doc and I gave it to all those pre-readers, roughly one hundred or so people, and they went at it in the Google Doc, and I'm so thankful to them, because the book would not be what it is today if it weren't for them.
Paul Cheek 60:21
I will say, once I started going through all their comments in the Google Doc, that's when I was like, oh gosh, I may have gone too far, in the sense that it took me so much time to go through all the comments. I still remember a flight back from Norway, where literally, I sat there for roughly eight hours just going through Google comment after Google comment, revising based on everybody's perspectives. It made it a much better book. But I think that human touch is really valuable, and it required that a lot of people come along and come on board to make that happen, which I think of as being extremely important.
Paul Cheek 60:49
You said outside of professional pursuits, I don't know, we were talking about kids earlier, it's like my mother was over here helping watch the kids tonight while I'm doing this with you. And I think our family, that I think requires a whole crew to really make it, they always say it takes a village, and I think that's one example. Raising a family, and all that, it most certainly takes a village in our household. So, like, those are some of the things that I think about. Could AI have made my golf game better?
Paul Cheek 61:15
Yeah, I wish, honestly, it probably could. And I still think back to one of the fun hobbies that I worked on for a while with a buddy, it's like, yeah, we said basically, we want to use AI to analyze golf swings to make them better. Was the tech there?
Paul Cheek 61:30
Yeah. Was my personal muscle memory, in terms of my golf swing? Yeah, not so much. And I've kind of accepted the fact that it's kind of like your reference to blackjack earlier, it's like, I'm not doing it because I'm going to be the best at golf, I'm doing it because it's an experience and I enjoy it, and that's what I'm going to do.
Paul Cheek 61:46
Right, and so while I would love to have gotten better at that, yeah, not so much, not quite yet.
Jason Baumgarten 61:59
Switching gears, one of the things I love that you're working on is rating companies outside in, looking at how they're doing on AI. And there's some companies at the top of the list that are probably not surprising to people, like Nvidia did okay. But there's also some companies that are a little more surprising. Maybe take us through one, that first of all, explain what you've done, and then what's an example that maybe surprised you, and what did you learn from it?
Paul Cheek 62:00
Well, so first of all, what I would say is when I look at a lot of the surveys that have gone out of where AI literacy is amongst board members and C-suite executives, I looked at the data and the results, and I was kind of like, huh, that doesn't totally add up. And so I started asking around, and asking questions of, hey, how often is AI coming up in this company's board meeting, or that company's board meeting?
Paul Cheek 62:37
And I was chatting with board members, and the answer, and this was well over a year ago, it's like, yeah, several times, every board meeting, right, it's on every board meeting agenda. But the fact of the matter was that most of those board members weren't actually AI literate. And part of a board's role, of course, is to manage the risk for the organization. And my thought process on this is, well, AI agents operate very differently than humans, both are imperfect, let's say.
Paul Cheek 63:06
But AI agents have a very different set of considerations, guardrails, privacy, risk, all sorts of different considerations, in terms of the management of them, that are more akin to that of software, rather than human leadership. And so, I was thinking about this, if a board's role is in part to manage the risk in an organization, but they're not necessarily AI literate, my belief is that they are actually increasing the risk to the organization with limited AI literacy, because those AI agents can go rogue, much like a human could, but in very different ways, with different constraints, different considerations. So, one of my co-founders, Felipe Csaszar, he'd been doing a lot of research on the relationship between innovation and AI adoption in organizations.
Paul Cheek 63:49
I said, hey, let's take the board angle on this, let's look at the S&P 500. We did just that, we started this company called the AI Driven Enterprise Institute. And we measured, using all available publicly sourced outside-in data, the AI maturity of every company in the S&P 500. We collected this data through structured systematic data pulls, but also through AI research agents that scour the vast internet for everything related to a company.
Paul Cheek 64:15
We profiled every board member and every C-suite executive, that's over ten thousand five hundred people. And we combed through well over half a million documents. We looked at hundreds of millions of data points, to really understand what is the say-do gap for each of those companies, and how do they compare to each other. We wanted to address the fact that right now, and before we released this overall data set, there was no company-level, publicly observable, objective, evidence-based rating or score for how a company's performing with AI.
Paul Cheek 64:49
We put that out, we put it out to the media, put it out with the team at Squawk Box, that's where we launched it. And that was amazing, we got a ton of outreach. It's been covered now widely.
Paul Cheek 64:58
CEOs talk about it, it's been used in board meetings for S&P 500 companies, it's been referenced widely in the media and investor days, that sort of thing. It now provides an objective, evidence-based evaluation of a company's AI maturity. Now that's kind of how it came to be.
Paul Cheek 65:16
Now the question is, what's the say-do gap between what an organization's leadership are saying about AI, and what the organization's actually doing with AI. So while, yes, there were some surprises for me, maybe, in terms of who hit the top of those rankings, by the way, the value is not really in the rankings, the value's in the data that sits underneath it, that we can use to inform what an organization should do next. But the surprises for me were more so, where were there significant differences between what that leadership, the board and the C-suite were saying and knew about AI, and what the organization was actually implementing? Because one of my biggest fears is that organizations go out and talk the talk, but aren't walking the walk, which means that they are being thought of as an AI leader, but the reality is they aren't actually doing much.
Paul Cheek 66:03
And then, on the flip side, organizations that are building a lot, but don't necessarily have that AI literacy at the board and leadership level, which means that they could be building things and deploying things to production that expose the organization to risk. And so those were the bigger surprises for me. There's a couple surprises that I might call out. I think a lot about companies in the tech sector, Airbnb, for example, we have four archetypes.
Paul Cheek 66:27
The early adopters, the AI trailblazers, the stealth AI adopters, and the AI visionaries. Airbnb fell into the AI visionaries, meaning they are talking the talk, but not building and implementing quite as much. Whereas on the flip side, one that surprised me for sure were home improvement stores. When you look at companies like Home Depot and Lowe's, within their sector, we see them showing up as doing more than I ever would have expected, compared to other retailers. I feel like I walk into a home improvement store, and like you would expect a home improvement store.
Paul Cheek 66:59
There's a lot going on, and there's kind of, in some regards, stuff everywhere, they are quite advanced. You may not see it when you walk in the store, but it's happening in those organizations. And the leadership literacy and advocacy is relatively high, or higher than I would have expected. So I was really excited to see that. But then, you're always going to see, because we believe we have to look at it sector specific.
Paul Cheek 67:19
For example, in energy, SLB was top of the list. In healthcare, Johnson & Johnson was top of the list. But the thing that I'm excited to look at, Jason, is what happens over the course of the next ten years. How does that shapeshift? How do we see some organizations leapfrog others, especially given that they may know where they want to wind up.
Paul Cheek 67:39
But the hard part is understanding where they are to start. Once they have that data, and once we kind of put that out there, they can then figure out where to invest, or where to deprioritize, to ultimately get where they're looking to go, in terms of their AI maturity. We've had so much fun with this, and I've been blown away, just because companies want to know where they stand. They want to know what they should do based on that data. The investment banks and the hedge funds, they want to know, oh my gosh, this data.
Paul Cheek 68:03
How do we go put this to good use. But the thing I want to do is, I want to make sure that every board, every C-suite is equipped with the reality that lies in that objective data, so that they can make the best decisions rooted in the most informed human judgment discussions, rooted in data, not in over-reported surveys. That's the thing for me. But it all lies in literacy, AI literacy.
Paul Cheek 68:26
What is AI literacy? I think about it a lot, and one of the things I like to hand people is, I have a, sounds silly, but I made a dictionary, a book that goes along with No One Works Here, and the dictionary, I basically framed it as every term that an executive needs to know about AI, so they don't get caught off guard in a board meeting or a C-suite offsite or strategic retreat. It's like literacy drives everything. We're going to put out some data shortly that shows just how influential board and executive literacy is for AI implementation.
Paul Cheek 68:59
I can't wait to get that out there, because I think that's one of those things that will really drive the agenda for most organizations on their AI upskilling and AI literacy levels for the board and C-suite. There's a lot more to come on that. Let me ask you this, this is a question I'm fascinated by right now. Our data shows that corporate boards have very limited AI literacy across the board, S&P 500. What do you think we should do about that, Jason?
Jason Baumgarten 69:23
What do you think the company should do about that? This is a great question, because I think we live through this with the internet, probably the most analogous period in history, where if you go back and you do the same exercise, to look at who was saying what, everyone was e-businessing everything. But there were some companies that really took it very seriously and invested thoughtfully, and some that didn't. There were some that said, oh, it's just a channel, some that said, it's more transformative than that.
Jason Baumgarten 69:52
I think that one of the things we also sometimes mistake in these transformational shifts is somebody who's got the buzzword bingo, or they're a good practitioner of the tools, versus they're a truly deep thinker about how this impacts a business. And I think one of the things that is missing sometimes is that you have to, at the end of the day, and you touched on some of this in the very beginning of your own story, you have to understand what makes a business good. Right, at the end of the day, it's selling something for more than the supply costs, where ideally you have some advantage in the pricing you can charge, and ideally some advantage in their cost of supply.
Jason Baumgarten 70:38
And at the end of the day, that's business, that is every business, no matter what it is, you're charging a little more relative to your competitors, and you're paying a little less for the cost of producing the widgets or the service or something else. And I think what people confuse is, they're like, well, this person knows a lot of bingo buzzwords, they must know about AI. And one of the questions I've started asking boards is not, what do you know about AI?
Jason Baumgarten 71:01
But I actually say, if you had a list of things, jobs, tasks, activities that your company does, do you think you could accurately place them on a spectrum of whether AI can do it, could do it soon, or could never do it, based on your understanding of the technology. When you say to me, are you AI literate, it's not, can somebody build a website with Cursor, can they use a Replit, can they do Claude Code, can they do a better prompt in Perplexity, I really don't care. What I really want, at the board and governance level, is can they understand the impact to their fundamental business of this technology?
Jason Baumgarten 71:46
Yeah, the things that move the needle, right? And what I find there is, sometimes people are not, they're not a fast, hyper, to use that example, but they understand the implication of not having to write by hand. And I think that we're in this moment where boards and management teams can get confused between the person who's the good practitioner of the tools, and the person who can actually say, based on your business, here's where there's real core risk, and where there's real core opportunity.
Jason Baumgarten 72:16
And at a board and executive level, that's how I would define AI literacy. So what do we do about it? I think one, you've got to train people. And I think that a lot of this is going beyond the parlor trick prompting and the like, oh, isn't it cool you can do this thing, but actually going back to, what is it that AI can and can't do?
Jason Baumgarten 72:39
What is it that your business actually does? Because, by the way, sometimes the gap isn't the AI gap, it's the fundamental gap of where the value is being created in the business. If the value is ninety-nine percent brand, I'm making that up, then the question is, how preservable and defensible is your brand in a world where AI can both dethrone it and create new brands faster? And is that real?
Jason Baumgarten 73:03
Is that a real risk to you? We've never been in a time where brand is both more valuable and more at risk. Right, because you look at brands that, MrBeast didn't exist. Amazing brand equity didn't exist twenty years ago, didn't exist ten years ago, really. So I think that this is one of those things where we have to go beyond the, how is AI as a personal productivity tool, and what does it unlock for your business? And how well do you understand the drivers of your business?
Jason Baumgarten 73:35
And for a lot of board members, I think the gap is that they are reactive to management's presentations, as opposed to starting with first principles, which is what makes this company valuable. I did a post earlier this year about a board that called us about a CEO succession. And their premise was that they needed a great CEO to move to this very, very difficult, out of the way city. And I asked them a very provocative question. I said, why should the company be based there?
Jason Baumgarten 74:07
And it was a really, really unpopular question. I mean, many of them had been recruited because they were the business leaders of that community, and they felt like the pride and joy of the company was being based in this town. But it turns out that town was not a compelling place for a CEO. It wasn't a compelling place for a management team. And in fact, most of the employees were no longer based in that place.
Jason Baumgarten 74:30
And I said, isn't this a moment for the board to take the unpopular decision of considering whether you should not be based here anymore? And I said, there's lots of companies, Chipotle, Boeing, and others, for good and bad reasons and for good and bad outcomes, have moved the company headquarters. And is that the real decision that precedes, can we recruit a great CEO? And they looked at me like I had three heads, and said, no, no, we have no interest in exploring that question. And then when I put that on LinkedIn, it was amazing how many board members and CEOs wrote to me and said, I wish we could have those kinds of conversations, right?
Jason Baumgarten 75:07
I wish we could talk about, are we even in the right business, or do we have competitive advantage, or the things that are really keeping them up at night, they're not able to talk about. So to digress, I think, what do we do about it? We start by saying, do you have a board that can attack the first principal questions, going back to, what is really the risk, the opportunity, the governance? And then the second is, can they move up the value stack of AI, from the tools, and really at the more conceptual level, understand what is the risk and opportunity to company.
Jason Baumgarten 75:38
I also don't think we're quite there yet, where most companies, the impact AI will be in five to ten years, not one to two years. I think it's going to take a little while for us to see tectonic economic shift. It's not that there won't be lots of new companies sprouting up, but if you're a fifty billion dollar revenue company, it's going to take a little while to see the cracks, right?
Paul Cheek 76:05
Well, and I think that's one of the fundamental problems, everybody's looking for the short term ROI. And now, whenever I engage with a company, I say, look, if you're looking for the short term ROI, I'm sorry, I'm not, generally speaking, I'm not your guy. And the reason why is because if we're looking for the short term ROI, we're expecting the ten percent productivity boost that comes when you educate people and empower them with some tools, and expect wild transformation, that doesn't come in the short term.
Paul Cheek 76:25
That comes when all of a sudden we're switching from reactive AI, as in AI reacting to what we prompt it, to proactive AI, which is more autonomous, where AI is operating with a heartbeat throughout the night, and all of a sudden picking up on what it needs to do next, reasoning, acting on that, while we sit here, you and I have this conversation, AI is working right now. I think the other thing on that is, I equate it basically to a high school student deciding to go off and get a four year undergraduate degree. They're thinking, okay, I'm going to make a significant investment of my time, the next four years, and money, tuition.
Paul Cheek 76:59
If they were looking for the short term ROI in that, they'd be looking for the salary multiplier on their hourly job, working on campus in the cafeteria, which is not going to provide payback for the investment they're making. They're looking for the long term salary multiplier. And any company that's looking for short term ROI is looking for that hourly wage increase in the on campus job. And not to say that we can't see some productivity gain in the short term. But if we're literally looking for what is going to fundamentally change the economics of the organization, what is going to appear in the balance sheet, the P&L, the income, those sorts of metrics, that will take, to your point, years and years and years to get to, when we are not starting from scratch.
Jason Baumgarten 77:47
I'm like so aligned with that, and I think of it as not the return on investment, but looking for the avoidance of loss in the long term.
Paul Cheek 77:47
That comes back to the obsolescence of companies.
Jason Baumgarten 77:57
Totally, totally. All right, Paul, I've gotten carried away, we're going way over. I mean, I think we both got carried away, and I love it. We're going to do the speed round. So five questions for you. You're twenty-five again, and you just finished your book. What industry and or job would it compel you to want to do?
Paul Cheek 78:12
Ooh. Based on what I've written in the book, I would want to be an entrepreneur that is going and building a hyper lean AI native startup that has the potential to have phenomenal impact in the world. But ultimately my role in that is as the architect and the designer of intent of what AI agents do to execute the work alongside us as human employees. But ultimately I am the architect building that AI native startup.
Jason Baumgarten 78:42
All right, the most overrated credential in business right now.
Paul Cheek 78:42
Oh, I think people who have managed thousands and thousands and thousands of human employees, not to discount that at all, I think that's extremely valuable experience. But I think that becomes overrated as we look at a workforce that is predominantly AI agents. And I think that requires a different skill set.
Jason Baumgarten 79:05
The most influential book you've read in the last year.
Paul Cheek 79:05
I don't read tons of books. Instead of one influential book, could I give you maybe two influential readings that I've done?
Jason Baumgarten 79:07
Sure, sure, fair game.
Paul Cheek 79:07
So one would be, Satya Nadella wrote a memo that was on X, that I thought was just phenomenal, and it talks a lot about learning loops and intelligence within an organization. One that I always recommend that folks read.
Paul Cheek 79:35
That one was called A Frontier Without an Ecosystem Is Not Stable, and it is excellent, everybody should go read it. The other one, while I generally am aligned with it, there are certain parts of it that I might look at slightly differently, was from Jack Dorsey of Block, and back in March he wrote kind of like a manifesto, if you will, from hierarchy to intelligence.
Paul Cheek 79:58
And that is, both of those are very aligned with my thinking, and what you find in the book. I think both of those are really important. And that one looks more at what happens to an organization when humans are not necessarily the mechanism for coordination, which I think is really important to consider for leaders today.
Jason Baumgarten 80:21
Certainly both very well circulated, and at least one of them very controversial. Business advice you hear often, and you think is truly terrible.
Paul Cheek 80:22
Ooh, and specifically because I'm hearing it so frequently right now, comes back to just what we were talking about. We need to find the ROI on AI this year before X happens. I think that is so shortsighted, and that is not the big picture.
Paul Cheek 80:40
The big picture is scalable intelligence.
Jason Baumgarten 80:40
And then finally, and I will not let you take an easy out on this one, a job you would be truly terrible at.
Paul Cheek 80:56
Ooh, okay, so I went to a business school, one that specializes in finance and accounting. I would be a terrible accountant, not because I don't understand it, but because I would just lose my mind. I would lose my mind and I would lose all motivation. A lot of respect from my friends who are accountants. I couldn't do it, is probably one of the things I'm very happy about with AI, is that accounting is one of the things agents can be trained very well to do, is to translate my brain to credits and debits, which no matter how many times I took accounting, I was like, credit, debit.
Paul Cheek 81:24
It really is, I work in dollars in, dollars out, I think, your brain, and that's one of those things I am so glad to have learned, and to learn the fundamentals of. Because I feel like without that, my ability to understand what drives a business would not be there. I just know I couldn't do it all day, every day.
Paul Cheek 81:41
That's where I'd lose my mind, and that's why I'd be terrible at it.
Jason Baumgarten 81:41
I couldn't agree more. Again, for anyone listening, No One Works Here, it's coming out August 25th.
Paul Cheek 81:51
The twenty-fifth. Go to Amazon, grab a copy. It's a fun read, and there's a lot of extras, as somebody who just finished, from discussion guides to cheat sheets to takeaways that you can try.
Jason Baumgarten 82:06
And it was a great read. I appreciate you spending the time with us. As always, this show's all about fit. And Paul, I can't think of somebody who's doing something that fits you more than what you found, not only in the authorship, but all of the education and the research and the sessions that you do with people. So I'm just delighted to spend time. And I think we're going to look back in ten years and go, boy, we got a bunch of this right, and a bunch of this wrong. But it will definitely be an exciting journey, and I can't wait to see where we go.
Paul Cheek 82:33
No question, thanks for having me, more to come, take care, bye.