Selling to Machines

Well maybe maybe to kick us off, Don, when when you were on the show back in twenty twenty three, you mentioned that, you know, we'd hit the hit the trough eventually with generative AI. we're we're sort of in it now, right? There's a lot of sort of false promises and confusion in the marketplace. what what feels different to you on the ground today versus what maybe you were imagining might might happen with generative AI back in twenty twenty three.

Yeah, thanks, Josh. And it's great to see you both. And thanks for having me on again. I I think a lot's changed in three years. certainly people know what generative AI is. they've used it. I mean, the numbers from OpenAI and Anthropic are pretty stunning in terms of use, not just at personal use, but also business use of it. So I think I think we're in that that stage where people are just trying to figure this stuff out. There are people like my neighbor who are heavy users of it. I would consider myself a light to medium user of the technology, but everybody's got everybody's got a different take on it. So I think what's different today than three years ago is we've been exposed to this technology that is seemingly magical. And now we're trying to figure out how does it exactly fit into our lives beyond just answering questions or giving me recipes or telling me what to wear.

in our book, you know, Rob kind of astutely predicted a mass adoption moment, but we were we were picturing enterprise grabbing the reins of this technology and doing all sorts of stuff right away and and it's been quite the opposite and and what you're saying alludes to that, Your neighbor's using it probably more than a lot of enterprises in the world are using it. and that that has kind of created a strange effect where, you have this this looming specter of outbound AI in the hands of consumers that's a threat to businesses that they might not be seeing. And then also, you know, there's there's individual use happening across these large organizations, whether they are sanctioning it or not, which creates risk, but also potential opportunity if you can get an architecture in place to capture some of the innovation that might be happening at the ground level.

I retired from Gartner recently, but when I was there, I was leading a line of research that we called Autonomous Business, which is the successor to Gartner's research on digital business, which I was also involved in. And part of that idea is that AI is just a tool, but what is the impact or change to the business model, which is the bigger question that most people want answered. yeah, we've got tactical implementations of this stuff, like writing software. But what does it mean for the business model? What does it mean for competition? What does it mean for people? So we basically put out a definition. We did a bunch of research, or Gartner did a bunch of research, and say, look, you know, this technology is going to create new types of value that you haven't imagined before. It's a little bit like what you talk about in the book, which is beyond human experience. That is the way that we try to talk. Differently about autonomous business than digital business, that it has to create new types of value. And yes, AI is part of the equation. But I think what's happening is that organizations, at least from my experience, are also still just learning about the implications of this technology beyond software development. So when we think about, you know, how does this change our business model? They're not there yet at all. They're just trying to figure out, okay, I'm getting pressure from my board. to lower costs, my CEO says, Hey, you should be able to cut thirty percent of headcount 'cause I read it from McKinsey. That's the dynamic that's happening right now. and that to me is is what's really interesting.

Yeah, I think recently too, we've been hearing a lot about token maxing, right? for a for a minute, for a hot minute, I guess, it seemed like that was some sort of flex, right? There was the news. I think it was in Meta that they had sort of a leaderboard in terms of who was token maxing.

yeah. Mm-hmm.

You know, and and that could signal in some ways like, we're leaning into AI, but I I think in a lot of cases it it's sort of moving in the wrong direction, token maxing should probably be more about maximizing the value of a token rather than using as many as you possibly can.

Yeah. I mean, isn't it kinda interesting if you if you think of tokens as a alternative to labor and then you imagine people bragging about their labor costs going up. man, we're just cranking. We're getting

Mm-hmm.

our labor costs are going up like crazy. Everyone's gotten an assistant. And their assistants have assistants. And then no one's talking about like, well what are they producing? well yeah, revenues are, you know, marginally going up.

Yeah. I I mean we all know that tokens are significantly undervalued right now. I mean the the companies are giving them to you or selling to me at a fraction of their true costs because they want you to use them. And there's definitely going to be a reckoning of that as well. but I think I you know, I want to go back to something you all said in your book, which is without systemic change, none of the stuff works. So yeah, token maxing that's a leaderboard, that's that's no different than Who is selling the most in a quarter, right? It's a, it's a what I would call a lazy metric. To me, the more interesting metric is where are you being more productive? Where have processes been reinvented? Yes, where have you deployed talent? Where have you eliminated talent? Where have you hired talent? There's a whole bunch of metrics out there that are really, you know, and Gartner's doing some great work in this area that are really getting underneath the impact of this change What are you doing with that investment?

there seems to be a lot of pressure on CEOs right now to be more operational minded, less spreadsheet jockeys and now like understanding the mechanics of their business and how AI seems seems to or could fit into improving you know, the overall business model. And it seems like especially with PE firms, putting a lot of pressure, replacing CEOs that they think are are not operational with folks that are. And and it seems like what it's coming down to is reconciling with the fact that it isn't that AI's not there yet, it's that organizations ability to leverage it is not there. and and if you kinda lay out what we were saying is it's not it's not the individuals, because they're leveraging it, clearly, as you pointed out, with how many tokens are getting burned by individuals, it seems to be management teams. So the management of companies, when we say companies aren't adopting, I think that's false. They're adopting like crazy. It's just a bottom-up adoption. it's kind of like Hogan's Heroes in the day, like I see nothing, you know, like just because just because it wasn't sanctioned,

Colonel Klink

yeah, exactly.

I'm old enough to remember that.

I know, I dated myself big time. yeah, I just think I think they're adopting like crazy. just bottom up and and the top is the it's like the tail wagging the dog, right? It's not that they're not adopting, it's that it's they're adopting in a really unconventional and weird way. And I like how you have put it, because that's gonna have implications on who the buyer is, who a business sells to. It if you don't believe you need to fundamentally change What's one thing that can help you as a company come to that realization is that you're gonna be selling to a machine.

Mm-hmm. Mm-hmm. Yeah.

Like, okay. How do you not change then?

Yeah, I mean that's one of many changes that that we're going to be seeing. and going back to the topic of CEOs, I also worked on Gartner's CEO research when I was there. And CEOs are totally and raptured by AI. I mean, it's like eighty percent of them see it as the number one technology that's gonna transform their business and their industry. Every other technology pales in comparison. But at the same token, you know, Yes, they want efficiency, but what they want is business model disruption and transformation, which is a much, much harder task to do. It's easy to get efficiencies from the technology, and you're right, people are adopting them. But without that strategy, which is what Gartner's been trying to offer with autonomous business, then it's just going to be a collection of projects. what's getting caught in the crossfire with CEOs are people. Because on the one hand, They say that people are very important. We need the right talent, the other type of thing. And on the other hand, they want to whack the majority of middle managers with AI. So it's a really interesting dichotomy. And one last thing that I'll mention Gartner's board of directors survey research says that boards are looking for CEOs who have more technology background, not less. So I think you're seeing a bunch of factors kind of swirling together in this transition period

Yeah, what is the implications of that? you know, more technology. I mean, they're there not because they weren't good at their jobs, they were there because they were really good at their jobs and and now they're saying I'm oversimplifying, but now you're not good at your job. and and that's because you're not technical enough. What exactly do they mean? You know, what What is technical enough for a CEO? What what is it that they're missing? What is I I'm not trying to like create a self help book here, but I mean like really at the end of the day, if you were talking to one and you're like, go read a book, you know, like is that it? Just read a frickin' book. Like what is it they're missing?

No, I I think that I mean a lot of over ninety percent of CEOs actually use AI in their day-to-day jobs. So they're using the technology. That's not the issue. But do like do any of us fully understand exactly how it works, exactly what the limitations are, exactly what the risks are? That's the that's the what I would call part of being AI savvy. And when we asked CEOs at Gartner, you know, who is the most AI savvy person on your team, it was only the chief The CISO and the CIO, but that was only forty, forty-five percent. Everyone else, like the last one is the chief HR officer, had the lowest level of AI savviness. So the thing that that that Gartner's talked about, where I talk about is you gotta use the technology, I've heard stories from clients of doing workshops where they build their own AI agents. I think it's that roll up your sleeves. Use the technology, ask good questions, show your people that you're using the technology, just don't tell them to use it. That to me is part of adoption. I think people at the rank and file need to see the leaders use the technology, not just talk

Yeah, that makes sense. And it kind of comes to our whole philosophy on if you're the head of AI, use case zero isn't going around and coming up with solutions for different departments on how to use AI. It's training everybody. It's educating everyone in the company on AI, that job number one is to educate, not to do. this whole idea that AI teams should fish instead of teach fishing, I think is might be some of the biggest issue is they want to go in there and play. I'm I'm hired to do AI, I want to play in Claude code all day. When in reality they need to be teaching other people to play in Claude code. and and maybe that's like, yeah, a piece of it is blind leading the blind, who's teaching your organization, how to wield it. The other thing I want to bring up is that is even more important, I think more of a puzzle to me. anybody who's creative understands that boundaries help creativity, constraints and limitations. Blank canvases are very hard to be creative within and it it's always nicer when you got a box, right? And with technology of the past, It constraints were a gift to all of us. Like this technology every technology was so constrained that we were always given this little tiny box. And and that helps you be creative in it. Like what can it do? What can't it do? Was a much easier question to answer. with this technology, every single person is finding new ways to use it that no one else knows about. Like you said, there is no way to draw a box around what it can do. There's just a way to to get your hands on it and start to feel it and get a feeling for what it can do, you know, and how it does things. But but there's no like standard box we used to have that says, you know, a database can, you know, do this really well, can't do this really well.

Yeah, I would say that my own experience at Gartner, we were given the tools, and a lot of tools, actually, a lot good stuff to work with. But we relied on our own curiosity to figure stuff out. And I I have to tell you, I learned more from my peers than I did from any formal training program about how to use this stuff. I have one peer in particular who developed a an unbelievable prompt series. To help sharpen your position. So at Gartner, we take positions, we might call them predictions. and we want to make sure, Gartner wants to make sure that it's as as clear and actionable as possible. So, yes, we in the past we use on peer review and internal judgment to do that, but now you've got this tool that can actually maybe shave a little bit of time, not completely replace you as a creative thinker, but help you along further and faster. And I I think that is the way to think about these tools, always as an augmentation of the work that you're trying to do, not replacement

I'm imagining a set of markdown files, right? And I'm like, cool. And then I read them and I think, well this is this is just his intelligence. Right? This isn't AI's intelligence. This is his intelligence. What and and and by writing this all down in markdown files and and then every other you know analyst being able to use these markdown files to compare what they've created against, well that's not a new idea either. And and probably something Gartner should have been doing all along anyway. And

Mm-hmm.

so what we've did is made it more convenient to do the things that we should have been doing all along but weren't.

Mm-hmm.

And and maybe that's like part of the problem is there's all these things we should have been doing that we weren't doing. And now We haven't been doing them for so long as a company and many companies that we forget you're supposed to be doing things.

Yeah. Yeah, I mean I think that I think that's a fair point. I mean, I would say that the Gartner business model has been very successful in terms of using human intelligence to advise clients, take positions, predict the future. The work that I did on machine customers was part of that environment that I worked in. But yeah, there's probably in any type of business, there's going to be things that you can't get to for whatever reason, probably maybe you're traveling too much or you've got too much work on your plate And you take shortcuts, we're all guilty of it. So yeah, I mean, if AI can help us deal with some of those shortcuts more effectively, but the end product is the same. I mean what I tell people is I can use AI all I want for the work that I publish at Gartner, but my name still goes on it. It's my name. and I have to be the backstop. You know, one of the things that Gartner talks about, Nate Suda, who's one of our one of the analysts there, came up with this productivity zone, which basically says there are jobs with low complexity and high complexity, and there are people with low experience and high experience. And the best way you're going to get productivity is give a highly experienced person with a highly complex job those AI tools because they know what good looks like. And they know how to evaluate it and can use it effectively. And on the other hand, if you have a new call center operator, low experience, but also low complexity, AI can actually help them get up to speed faster than normal training because they don't know what good looks like. They need to be trained. So part of this is situational, it's not a blanket statement. It depends on where you are in the complexity of your job and where you are in your experience level. but but I think, you know, in in the time, you know, that I had access to those tools about two years before I retired, and they were getting more and more powerful. And yeah, a tingle was going up and down my spine about what does this mean. But my my thought was I'm the human face of this knowledge to the clients and to the people that attend the conferences, and I cannot lose sight of that. I can't hide behind AI because I'm front and center. So I think in those situations where the human really has to be front and center, you cannot hide behind AI. It has to be a it has to be a tool.

Yeah, I think analysts have a slight advantage over writers in that at least the verb that explains their job is truthful. Like you analyze. Okay, so so now we know that AI is not replacing you because it's not replacing analyzing. but when you're a writer and we see AI write, well your job's over because it writes. But honestly, writers aren't really writers. They're analysts. they're storytellers and curators and they synthesize and decide and they have taste and they use that taste. And and yeah we sometimes just the the name of a job confuses the market Coding is like twenty percent if or not or ten percent of what a coder does, but then you're like, Well AI can code. We're like, yeah, so that what about the rest of the eighty percent? We should have never named it that and we wouldn't be so confused. so I think yeah, you what you're saying is you still analyze and this helps you analyze, it helps you to expose yourself to more analysis, but at the end of the day, you're compressing this for somebody who's busy. And you're compressing it to what you think is gonna be relevant to them and you're looking for novelty, you're trying to share information that isn't common in the world, and the AI is doing the exact opposite. It is looking at what's common in the world and synthesizing it to people. I like AI as context and analyst as ideas.

I totally agree. I totally agree. you know, I have faith in the power of human intelligence, our ability to create combinatorial insight, to be able to take different pieces and weave it together in something that nobody ever saw before. You know, that's how the machine customers work. It started with a question from Chris Howard, who was our chief of research at the time. He wanted me to do a presentation in a TED-like talk for one of our conferences. This was gosh 10 years ago. He said, What if an IoT device was a customer? And that's it. And then he just walked away. And he actually wrote the forward to the book.

I love it.

Yeah. So it's a I mean, he dropped this gift in my lap. And I'm like, okay, well, let's work with it. and this was 10 years ago, so no AI, human ingenuity, research, talking, testing ideas. Iterating, iterating, iterating, and then finally we get a book out of it. That was humans that did that, not AI. Now, the third edition, I'll be completely honest, I did use AI to do to write parts of the book. I would actually

Mm-hmm.

take new research that was published after we did the second edition. I said to Notebook LLM I said, okay, I want to integrate insights from this research note into the book. Tell me what I should say and what chapter to put it in and where I should put it. And I'll be damned that it did that. So what that did is it saved me several hours, but also I still had to review it. I still had to make sure it was worded correctly and stuff. And then

Mm-hmm.

lastly, my son, who is an English major, he did some editing. And he said, Dad, I can tell when AI wrote this stuff. You need to change a couple things around. And that's the other thing that we're seeing right now is people can tell when AI writes something because it's too perfect. So I think you're gonna see that backlash. that you know, obviously they call it AI slop. Not to say that I anything I did was AI slop, but that's the type of dynamic. And this is, you know, we would never even two years ago we wouldn't be talking about this, but now we're talking about it.

Yeah, it's really wild. I I actually had an experience as a writer, I recently finished a novel, and there was no AI tools really involved in the composition of it at all. But I'm pitching it as literary fiction, and so I gave it chapter one of my book, and I said, Read this and tell me where it is succeeding as literary fiction, and then tell me where it is. Clearly trying to be literary fiction and failing. And it it gave me such an incredible breakdown of like lines that I kind of already had a sneaking suspicion about and really a powerful moment, but also a weird one to internalize, right? Like how do these tools fit into something that's typically viewed by me and just by the broader public as like such a personal. sort of lone wolf activity. It's like, no, there's actually a new type of collaboration that just saved me a lot of time and really helped me improve my human output.

And I I think you use the right word, Josh, which is collaboration. It's a collaborator.

Mm-hmm.

It's a coworker, whatever you want to call it, not a replacement. my wife is she's written a few children's books and she's using AI to do the same thing that you're doing, Josh. It doesn't change she

Mm-hmm.

still write the writes the words, but she's getting some coaching that is helping her think differently about it. to me, that is the right way to use this technology, and that that's pretty exciting. And then obviously you can use when it comes time to promote your book, you can have, you know, Claude or Gemini, whatever, create the promotional campaign. It can create your Instagram reels, all that stuff. I mean, that's there's nothing wrong with that. It's still your words, it's still your content. But you're using these tools to, you know, expand the reach of it. To me, that's totally fine. And I I I love that example because you just say, Hey, what if we tried this? And you tried it.

Mm-hmm.

And you were pleased, you could you knew it could look like, so you could evaluate it. You just didn't take it carte blanche. That's a type of discernment that we all have to build. And and I think going back to one of my earlier comments about what's different now than three years ago, is I think we're all building our levels of discernment around the output of these tools.

Yeah. I think that sort of connects in a way to something Robb and I were talking about right before we got on this call. we were thinking about hype cycles.

Mm-hmm.

And you know, we were talking about the trough of disillusionment, which is part of the hype cycle. does the nature of this technology, which is so different from even things like like RPA, for example, like does it change the nature of the hype cycle? Does the hype cycle become more compressed and is the overlap between hype cycles more severe?

Yeah. That's a common question at Gartner. There's always people trying to punch holes in the hype cycle. I mean, the original

Mm-hmm.

intent of the hype cycle was really to track essentially promises versus reality, right? So vendors

Mm-hmm.

make promises, then we have the real results, and when it's things aren't happening the way the vendors promise, it slides into the trough and in some place, it stabilizes and it goes back up again. I think that with AI we are definitely seeing an acceleration in you know, going up to the peak and then dropping down in the trough and then coming back out. But what's interesting, given the dynamic nature of this technology, you could ha literally have a you know, peak, trough, go back to peak. I mean that type of loop, conceivably. which is interesting. I don't know if anyone at Gartner's thinking about that, but You know, that that to me is very realistic. And that's because this technology changes so rapidly. We have never seen anything like this before.

The hyper hype cycle.

Hyper hype cycle, yeah. I like it.

So yeah, I I was thinking about what you were saying, you know, Josh, writers, we all we're not we're not forcing AI into our work. this isn't forced. you're not trying to use AI on your latest book. you're using it in a in a very like natural way. And and and clearly you don't need to be an expert on AI to have used it. There's no

Mm-hmm.

expertise in AI. that was particularly required, like years of training so that you could use it on the book. so clearly that's that's not the thing in the way. and I I kinda go to the bull whip thing that Joshua Gans talks a lot about. And and I'm

Mm-hmm.

I'm very much stuck on this. It's it's as individuals we don't have to get anyone on board with how much we're using it, when we're using it. We don't have to get permission to use it. We don't need someone to review it afterwards. we don't have to agree with other people on which one to use. you just used it. And

Mm-hmm.

all of a sudden in an organization, you've got this bullwhip effect, which is now we we have to share the same one. It's it's used to buying software that everyone uses, right? It's industries share this solution and all of our employees need to share this solution and and everyone needs to do it the same way and we all have to agree on which one to get. And and the whole thing just bogs itself down because it is such a flexible technology and it can do so many things that all that really makes sense is for you to just have access to it and use it organically. But it doesn't work with this groupthink. And this bull whip is if one organization, one portion of your organization leans in, it creates a bull whip effect where the rest of lash back, you know?

I think what you're seeing is just fear. I think people are afraid. You know, it's human nature. Especially a lot of people in the workforce did not grow up with this technology. there's a level of hesitancy, myself included. You know, I I resisted using AI for a long time. And then I started hearing my friends use it at work and say, Hey, do you know you could do this? I'm like, No. So I tried it and then I found something else and I told somebody else, Did you know you could do that? So I think a lot of What we're seeing is the technology is pretty robust, but there's the human fear factor, if you want to call it that, is involved. That's holding people back. That's the systemic change that you all talk about. It's almost like the elimination of fear. But how do you do that in a big company? How do you you know, the nature of big companies is to minimize risk? And what you're describing is, hey, we're gonna take risk by using this technology. That to me is an interesting conundrum. It's very, you know, companies talk out of both sides of their mouth. You know, they want the

Mm-hmm.

growth, but they're not willing to take the risk to get there. I think that's what you're seeing with this.

Yeah, we've been kind of obsessed with this idea of crisis engineering. it's a it's a book that was co-authored by Marina Nitsa. And it it it really is a powerful idea, right? When you're we're thinking about systemic change, that in a crisis there's these brief moments, these little windows of consensus where, the thing that really has to happen is that you have to have a plan lying around, right? Because in a crisis, whatever's lying around gets grabbed. So if you're I guess a true crisis engineer, you're waiting for a crisis, but what you're really doing is having a plan ready. So that when they're looking for that tool, you're like, okay, well here here is actually a way that we can use this technology to help us now, but also create a a foundation for doing big things down the road.

Yeah, definitely. I love that idea, Josh, of using AI for risk management and risk mitigation and alternate futures and alternate plans. I mean, that's a very, very powerful tool. And I'm sure I I don't I'm not I don't have personal experience in this, but I'm sure a lot of risk managers are doing that and using it for scenario planning. I mean the the the bottom line is that there's gonna be risk. I I wanna I've been looking for a quote from the book that I I just love. And you wrote in the book, I'm starting to wonder if rather than AI in the hands of companies eliminating jobs, it might be AI in the hands of people eliminating companies. To me, that's the risk. That's the risk. And the way that like Claude works is if you can imagine it, you can build it. And we've never had that technology before. That to me is a dynamic. I don't even think we've seen the beginnings of it yet.

Yeah, that's what happens if if you think the world's static around you and it's gonna wait for you. I guess that's part of just all these mental models that we carry with us. One is that our current competitors are our future competitors, right? And so if they're not moving, you're okay and safe. And then maybe you think, well the only alternative is a startup and you're like, well, that certainly is an alternative, but not seeing the adjacent companies that were never competitors, we're seeing more that happening in at incredible rates where where everyone thought they were in a swim lane and then someone just pulled all the ropes out. And and people are still sorta swimming in a line, but you can see they're like drifting, you know, like Salesforce and ServiceNow like way everyone's getting out of their lane now.

Yeah. Well, what's interesting is when we first started working on digital business at Gartner back in twenty fourteen, people thought we were crazy. People thought, Hey, well, aren't you just talking about e-commerce? And we said, No, no, we're talking about people, business, and the Internet of Things and how they will change business models. And the biggest thing, one of the biggest outcomes of that was disintermediation, cutting people out of the The supply chain cutting people out of the relationships, cutting brokers out, things like that. And that happened to an extent, but I don't I think you're gonna see a lot more of it. A lot more of it. People take out whole sectors of intermediaries with a Claude program. Because at the end of the day, people want easy experiences, like you said in your book, they want easy experiences that are intuitive, seamless, frictionless. And right now, a lot of there's a lot of processes, there's a lot of intermediaries in place that cause friction. You know, here here's one example, a recent one. So I brought my personal phone number to the carrier, which will remain nameless because I don't even want to get you guys into trouble. But that carrier had my number, my personal number for 15 years. And now it was time for me to get it back. I swear to you. That carrier made it impossible. Impossible. Hours on the phone, hours on the website, hours in the stores having to prove my identity. I thought I was in hell. and supposedly I had an intermediary to help me with that, but they ended up being useless. So what did I do? I gave I gave up. I actually got a brand new phone number. But that experience could easily, easily be handled. by some simple program. But what do you have? You have a legacy company with over a hundred years of legacy systems that are simply not capable of offering a frictionless experience. So what I see happening is the companies that were not dealt with during that first wave of digitalization are going to be dealt with in this age of AI because people have simply had it. And they add cost, you know, and cost big deal right now, whether you want to say it or not. So anyway, it was one of those things where it just made me realize there's so much opportunity out there.

Well, it's tangential, I think, to something we talked about last time, which is there's this side of AI, Where if you're using it really well inside of an organization and you really are driving towards efficiency and better experiences for people, your activities might actually cut into profitability, part of like making people more efficient would be them consuming less and not buying things they don't need and not using resources they don't need. And that sort of runs counter To what a lot of, you know, business exists to do. So at some point it seems like that has to be reconciled, right? And it's hard to know if that will come at the hands of consumers using outbound AI to go after a company that just irritated them and they're

Mm-hmm.

you know, they have two hours of free time and they're just gonna build something that floods a call center. or if it comes through some sort of coordinated effort between a forward thinking CEO and a a board that's rethinking their relationship to profitability.

I would say it's it's a concept we explored in the book, which is supply and demand are often out of balance in many categories. So the example that I like to use is produce. I don't know about you, but I buy way more produce than I consume, and I end up throwing some of it away.

Mm-hmm.

And I think the idea of the machine customer who's acting in your best interests is saying, Hey Don Know I've noticed that you buy five apples, but you only eat three and you end up throwing two away. Why don't we just order three next time? that type of gentle coaching. that I think will help even the scales out. The converse might be true. hey, we noticed that your bandwidth, you're using you know X gigabytes when you really need X plus Y. And let me recommend a way to do that. So that that would be an example of consuming more. But I think what it what we're looking at is how does the machine help you make those decisions? because a

Mm-hmm.

lot of us are just simply not attuned to it enough to be able to make that. And the same thing might be true of businesses. So I see it that way, not wiping out industries it could lead to a short-term decline. in in sales, but over the longer term, you'll get more satisfied customers if I'm consuming what I need when I need it.

Yeah. last time this prompted a question from you, Robb It was like who who pays for the AI that makes you buy less?

Ha ha.

Yeah, I if if you were to try to look through one lens, AI requires companies to go to first principle thinking. And they're so in the weeds and they're so bottom up and they're so how do I improve this process at the bottom instead of like should this process exist? This is just everywhere. How do we automate this process? Not should this process exist? How do we eliminate this process entirely? And when I go to first principles thinking on why does a business exist? So you go one layer up, why do why do we exist as a business? You get into some pretty deep thinking. from folks like Gartner, which is transaction cost, right? and and understanding that the fundamental reason large businesses and many businesses exist is because they lower transaction costs. part of that is just trust, right? Part is the cost of building trust for a consumer is is the cognitive weight of a transaction. Like I know what I'm gonna get. whether it's the best or not, I know what I'm gonna get. as I think about AI, I begin to think of what you're saying, which is AI advocacy or or AI as what did you call it?

Think it was AI activists, maybe?

Activists, thank you. if now I don't have to trust the brand anymore, I just need to trust my AI. And as you put it in our last conversation, not the AI someone else programmed for me, the one I programmed, by just talking to it. And and it's out there and I trust it, then then that whole sort of moat of transaction cost that most businesses ride on suddenly goes away and and trust becomes proxied to this AI. and I think that's sort of part of the existential crisis of companies versus people and jobs.

let's let's talk about the first thing, which is why businesses exist. I also agree it's to lower transaction costs, but also create value for stakeholders, whether it's profits or services, things like that. I mean I I fundamentally still believe that. I think that with the introduction of more AI tools into business processes and operating models, The risk is people automate a bad process. And we see that we've seen that time and time again. the question I've got about that is do you have the right people to reinvent a process? So let's imagine you have a department of people, they've been doing the same thing for 10 years, and you say, Hey, you know what, we got these AI tools. I want you to reinvent the process. They're gonna look at you and say, Well, I don't know, I can do that. I mean, I I'm so steeped, I am so I have so much vested in this whole process. I don't know if I can think differently about it. So that's that to me is a is a very interesting issue that you've got with talent is do you have the talent that can reimagine or reinvent processes? know, if I think about the upcoming generation of talent, that's gotta be one of their skills, which is I see this process, I think we can do it differently, we can use these AI tools to help us. So that that to me would be a really, really interesting skill to have. Now, when it comes to machines. And customers, and what we talked about in the past is right now they're just being used for very simple transactions, nothing very complex right now. And there's still a lot of risk. There's trust issues. And I think it's gonna take time for people to work through that. One of the the things that I came across, I think I mentioned this on our last call, was Target made the announcement that if you use their shopping assistant, which is run by Gemini and it makes a mistake, you're on the hook for that. You have to pay for it. And I really had to scratch my head on that one. I I mean Target, do you really need to do that right now with everything you've got going on? But they did it. And I'm like, so what's the incentive for me to use that shopping bot? None. There's zero. There's zero incentive for that.

And what's their incentive to make sure you don't buy produce you're gonna throw away?

Yeah. they don't have an incentive. They they want you to buy as much as possible. Now a company I could see a company like Sprouts or Whole Foods saying, Hey, you know what? We want to promote responsible buying of produce. So we're going to ask you these questions and we're going to suggest how much you buy. So you don't overbuy. I mean that to me would be probably a temporary decrease in sales for Whole Foods, but a major plus in terms of the relationship and the affinity for the brand. I mean big companies will hopefully balance the two.

You can imagine too, if like people are buying the right amount of produce, that would trickle up to Sprouts or Whole Foods produce section buying the right amount of produce to sell.

Exactly, right, right. Right. So maybe they they wouldn't overbuy, have to market down. You feel compelled to buy a dozen apples because hey, the price is just too good to pass up. You end up throwing half of them away and nobody wins. So yeah, I think you're absolutely right, Josh. It it goes it can go up and down the supply chain in terms of really being precise about what's needed. And I think going back to this idea of trust is and we talked about this last time. It took us a good what ten years to trust Google Maps and

Mm-hmm.

Apple Maps. It wasn't immediate because there were a lot of screw ups in the beginning and and a lot of people swore it off. Most people, I think three of us probably stuck with it until finally we don't go against what Apple Maps or Google Maps says because we know that when we do we get we get in trouble. That took ten years. I would say closer to fifteen years. And I see the same kind of trajectory. You see, you know, seventy percent of of people are interested in using AI agents for shopping. This was from a study that I think Deloitte did last year. but only less than ten percent are actually using the full technology. And some people would say it's even less than that. So and trust is the big issue. I'm interested, but I'm not sure I can trust you just yet. That to me is the big challenge we're gonna see.

Yeah, the other side of it is I I'm finding it interesting because it's seeming like these retailers are finding ways to block UI from helping their consumers. And I wonder if they'll succeed. it's sort of like using their you know, sort of power of wouldn't call it monopoly, but their their sort of convenience power, whatever you want to call it. to stop us from using our AIs to make smarter decisions by not allowing AIs to access their system

Mm-hmm.

and find prices and find products. I see Amazon is inaccessible directly

Mm-hmm.

to these systems to be able to like go buy it here. I guess that's like the fear of of letting these these companies get too big is they they now sort of wield this this power I mean I nobody wins. It's always a delay, right? we'll win eventually.

Yeah. Mm-hmm.

and it's it's simple, you just put open claw or something on your computer and they can't tell. But but it's it is interesting, right? That we're already seeing it. We're already seeing Both sides. We're seeing people using it and then we're seeing retailers try to stop the use of it, which means they must be afraid of it.

I think so. I mean the eBay was the other platform that said no AI agents are allowed into into the platform. But yet Amazon ironically has its own AI agent that will go outside the Amazon system to get what you want. They call it Amazon's Buy For Me. So they're they're protecting their markets. This is not a new behavior at all. one of the things we talked about in the book was this is never gonna scale if you're gonna have a series of walled gardens. You know, my HP printer can only buy ink from HP. you can also buy HP ink from Amazon, but it can't buy refurbed cartridges, it can't buy from other types of sellers, things like that, because it'll brick the printer. So yeah, you're gonna have a good sized market, but not the big market that you want. So what to me to me, what Mark and I talked about in the book is in order for this to truly scale, we have to have these big platforms, commerce platforms, where machines and humans meet. without restrictions, free and fair and open trade of goods and services. and it's not happening. You know, we we've got some small examples, but it's until that happens, you're gonna see a lot of things like the Target or the or the Amazon or the eBay's out there.

Yeah, it's so interesting to me. You know, the the printer ink is a great example. they're sort of preying on the cognitive load of life for most people, right? you walk in, you see a printer for half the price of the other ones, you buy it, and you don't have time to research the implications of the you know, annual cost of ownership for that printer against how much ink you're probably gonna use and so they sort of sell the the printer at a at a loss so that they can later catch up with the cost of ink. But AI, it doesn't have this cognitive limit that we have. It it can totally study that in in seconds and and provide that help to us. And so a lot of these tricks aren't gonna be useful in in the future with AI. and and so you're just gonna have to provide value, I think. I mean is that is that the truth

Yeah, you know, and I I'm not picking on HP because they we interviewed them for the book. but their instanting program is essentially a printing as a service or subscription. It actually is quite cost effective because you only pay for what you use. and it it's an automatic payment plan and well as as you replace ink, you basically pay for a certain number of pages that you're gonna print. So that in my in that my opinion, that's the right balance. of a service but also being cost effective. so that to me is is kind of the way to go is we're gonna put you in control of this, but we're gonna automate and make it easy for you. And that's not even AI. That's just that's just rules. That's deterministic. the the thing that that I keep going back to is that these AI assists don't work for us. They work for the people that made them. And I think I mentioned to this to you on our last call is I'm waiting for the announcement where somebody is gonna say, Hey, you can buy or rent your own AI agent that you train, you give it your data, you take responsibility for it, frankly, which means you might have to buy insurance, but it's yours and it acts in your best interest. So until somebody does that, what I tell people is that you know, Alexa not working for you. Siri's not working for you. Amazon Buy for me is not working for you. They're working for the companies that made it. And that's a really, really important distinction. I want to open people's eyes up to say, hey, these technology's really cool, but let's let's be clear about where the incentive

Yeah, and especially with like OpenAI saying they're gonna introduce advertising and things like that, that debases a lot of what's trying to be accomplished. And I think it's interesting too, because if that if that moment you're describing happens and takes off, like if people really do start buying subscriptions to these kind of agents, you know, if organizations haven't really moved the needle much they're then they have to recalibrate again. Now they're no longer trying to implement AI and figure out how to reach humans. Now they have to figure out how do I restructure my company now that most of my customers don't come to my website, don't look at me directly, just interact with me through this gatekeeper that is basically can see right through all any marketing. Like marketing is is noise to it. It doesn't,

All the way.

you know, like it it just wants data. Like like what does that do to the makeup of of a company then?

Well, it definitely is, I think, a substantial you know, change. So I have there's two people out there who are also working on machine customers, Katja Forbes in Singapore and Sirte Pihlaja in Finland, and they talk about this stuff is that a company and we talk about it, which is you have to fundamentally change the way you go to market, how you sell. Because if you're selling to a machine, it's a very different dynamic. emotions are not involved in that. And that's a very, it's a huge structural change that I don't think people are really kind of conceptualizing right now. That that to me is is a major, major stumbling block. The other thing too that I was challenged on was I I grew up as a marketer. I didn't I didn't work at Gartner my entire career. I was a marketer at Coke and Quaker, and I learned all about the fundamentals of brand building and brand management, et cetera. And I understand the human element for building a brand and why it's important decision making. So when one of the early reviewers of the book said, Hey, well, you're saying brands don't matter. And I I admit it's a fact that yeah, they probably don't. But then I I pivoted and said, Well, what if there are two brands? What if there is a human brand and what if there's a machine brand? And if you believe that the fundamental role of a brand is to keep promises, then you could have two brands. So Coca-Cola, for example, could have two brands: the human brand, which we're all familiar with, and a machine brand, which is all about service commitments and pricing information and consumer insight, all that kind of stuff. And they can exist side by side, but they serve two different types of customers. That is a big change, which I don't think many people are even thinking about right now.

Yeah, and in some ways that machine brand sort of the the information all exists, but it's not cohered together, right?

Right, no, it's not it's not accessible.

I feel like right now there are a lot of companies, there that gulf is wide, right? Like

Mm-hmm.

we we say do no harm, but like we but we really just want to make money. And here's here's all the things that are happening behind the scenes, but then Yeah, as that a lot of that data becomes consumable to machines, it would have to have an effect on the human facing brand too, right? Like that gulf can't just sit that wide forever, I wouldn't think.

Yeah, I mean, Robb was talking about cognitive load. It's very real and even more real today with all the distractions that we have. Wouldn't it be nice if someone could sort through that? So here's a quick example. I I, you know, having retired from Gartner, I had to kind of rebuild my computing infrastructure, if you want to call it that. So I I I'm a big fan of Google. So I did I did Gemini Plus, so that 80 bucks, now forty-nine ninety-nine a year. They just lowered the price. Now I have Gemini giving me a daily briefing in my email. It reads all my emails and is giving me advice about what to do. To me, that is one of the closest things I've seen to an AI agent that works for me. And I'm thrilled about it. And I actually do believe, and I've said this before, I think Google could be one of those companies that would basically rent or buy your own agent from them. but that that was a it's a small thing, but like For example, I had a conflict tomorrow. I had a haircut and I had an appointment, another appointment, and I haven't resolved it yet. So Gemini is saying, Hey, you need to resolve this. So I did. I saw that as a very, very positive development. But who else is doing that? And who else is gonna span the entirety of your activities? It's not gonna be many.

It's exactly what you're saying. It's easy to see the connection from something as simple as as sorting my inbox. Like which which let's face it, email clients have been trying to do for years and just failing at. And now with this technology it's doing a a much better job of Sorting your inbox from what matters, what doesn't, what's spam, what's not spam. it's it's like finally a spam filter that works in in some ways, right? You're like, finally a spam filter that works. and then going from do I work for you or do you work for me? Like don't tell me to go and resolve this. You resolve this. you know whether you're looking for a job, whether you're looking to hire somebody. I think a lot of people have a trouble anchoring this part of the book, but it's what we call core patterns. And there's these core patterns which is, you know, compare this thing to this objective.

Mm-hmm.

Right, and then and then weight it. Just very simple. Compare this email content to my objective and give it a weight as how relevant it is to meeting my objective. Now that core pattern can apply to inboxes, text messages, documents, articles. Like it it doesn't stop and that pattern is i is algorithmically the same.

Mm-hmm.

And and so it's so easy to see how why just your inbox? Why not go pull all the articles relevant to machines as customers that were recently published on the internet and also throw that into the mix? And and and that's why I think it's all gonna happen so fast and and and why the bottom, like you said, why customers will adopt faster because it's It it's a small step from what you just saw.

Yeah, I mean, you had a several scenarios in your book. I've done the scenarios about all the things that machines can do for you, but they're just not there yet. you're talking about multi step, complex transactions. And I think those pose inherently more risk than just, hey, order me more toilet paper or buy me three apples instead of five. So that to me is the journey that we're on and that is where We will likely be in the trough when it comes to machine customers because all these promises of not just giving me information but doing things on my behalf are just not materializing at this point. You're right though. You know, Gemini should have said, What would you like me to do to resolve this conflict? Your book focuses really on conversation. I would just talk to it and say, Here's what I need you to do.

And then it would send an email back. I mean, that's it. send an email back to them and then how long until they say, I'm tired of talking to robots, I'm gonna get a robot.

Right. Yeah. So but I I'm encouraged. Like I was really encouraged by the Gemini briefing because what it also did is it also referenced some of my other conversations. So I'm in the process of building, you know, I have an LLC, I'd like to do some select advisory work doing podcasts and stuff like that. So I have to build the infrastructure for that. So Gemini was reminding me, it said, Hey, you talked about this. Here's some things you should be thinking about over the next couple of weeks. it's a little bit like Josh, the writing assistant. I have a collaborator now. So Gemini is kind of like a collaborator for me, just like Claude is. Actually Claude helped me plan my retirement activities, which was fantastic by the way. So these are collaborators and and that's the way to use these tools.

W earlier you you had mentioned that boards are kind of looking now for CEOs with a heavier tech background. I was wondering if if focusing too deeply on that is a blind spot also, in sort of your penultimate Gartner presentation you were saying, like don't forget moments of humanity, right? I think a lot of this technology and we we've kind of been alluding to this all along too, like so much of it is not mapped out and not understood and there's no box for it. So while it is helpful to have technical understanding of how these things work, a lot of what might be unearthed could just come from human to human conversations, like people within your organization communicating better, finding things to automate where it like creates more opportunities for people to interact with one another. is it possible that there is maybe too much focus on tech savvy sometimes in leadership roles and not enough focus on how do I ignite passion and interest in my workforce to be talking about this and interacting with one another and helping surface some of the stuff they've been working on so that we can use it at at more of a high level.

Yeah, I would say so. I mean, I think the vast majority of CEOs that that I've been exposed to are very focused on growth, no matter what it takes. I mean, that's whether it's people or technology or acquisitions or divesters, whatever that looks like. So that to me is undeniable. It's always been the number one thing. things like workforce, customer experience are always further down the list. They, you know, on a list of fifteen different things, it it can range from number six to number ten. So they're just not that high. So and I don't know what you're gonna do about that, honestly. You know, I I worked on CX for many years at Gartner and trying to say this is important, but at the end of the day, a lot of CEOs are very focused on that share price and and doing whatever it takes to get that. So, you know, yes, people are important in that moments of humanity presentation that you're referring to on YouTube, you know, I talk about the fact that some people don't want to use the technology all the time. They want to talk to people. They want to have somebody empathize with them. they don't want everything automated. And I I think you're gonna see a bit more of a backlash, especially when it comes to some of these IVRs, these automated call center systems, infuriating. so I I think you know, if you ask me, there's there's always going to be a room for service workers who are able to empathize and and be able to not just say, I know what you're going through, but say let me help you fix it. I can help you, I can help you solve all that problem. And I don't know about you, but I am relieved when I can get out of the automated IVR trap and actually talk to a live, competent person. I feel like this huge burden is off my shoulders. I can actually get something done. I think that's that's one of the things we're gonna be wrestling with the next few years is people think they know what customers want. Hey, we're gonna make you completely self sufficient, but in fact that's not what everybody wants.

Yeah, because there are certain things I call in and I want I'd be totally happy for a an automated system to tell me the location of a of a retail store and I don't have to talk to anyone. But then there's all sorts of other stuff where if the automation is doing that, the representative has more time to walk me through something that is sort of higher stakes and more complex.

Yeah, if there's ever a technology that didn't work, it was IVRs. You know, it just didn't work.

Mm. Mm-hmm.

And it was a promise that deserved the trough of disillusionment and never got it. it's a great idea and it it just failed for decades and and now it has a chance to work. And I think w working means at the very top knowing when it should try to handle it and not. Let's start at the very top of it knowing this is this is someone who's gonna need a human and this is not. Like let's just begin with no one even tries that.

Yeah, I I think I think what it what you're seeing is the kind of industrialization of business processes. You've got people that are not customers trying to reimagine processes and taking humans out of it. And I think in some circumstances, Josh, as you mentioned, that works. I think in the vast majority of them they don't. but I think there is such a push on efficiency right now. I mean, look at look at fast food restaurants. Now you have to go up to a board you type in your order. and I don't know about you, it takes me a lot longer to do that than it does is to tell somebody my order. When is that gonna

Mm-hmm.

change? you know, Walmart taking self-checkout out of their stores. Yeah, there's probably, yes, there's probably a theft issue, but there's also a human interaction issue. Because the checker at a Walmart is a major point of the experience.

From a UX perspective, it seems obvious. Like there's someone who was trained to use the interface that is efficient on it, and then there's you who's being trained on the fly. Like I didn't come here to be a checker and to train to be a checker at Walmart. I came here to buy some groceries.

Well, we've all been so so trained on self service now that, you know, we're just not surprised by it anymore. We just kinda sigh and do what we have to do 'cause we don't have a choice anymore. And, you know, like everything else, I think it'll swing back. Right now we're just on this whole automation, industrialization of experiences. At some point it's gonna switch back, I think, or c find some reasonable medium.

Yeah, I I think we've always advocated for the frontline get stronger with AI. That's what happens. If you if you look at it through the frontline eyes, which I like to do, and this is with the help of Joshua Gans again, another call out to him. you know, the middle manager's job was to reduce friction on the front line. That's that's their job number one. They were sort of a aggregator of information to the top and and a reducer of friction to the bottom. And if you talk to most people on the front line, they would say, yeah, a bad one, like an IVR. Like sounds great in theory, but in in their ability to reduce friction was super limited. And AI can If you use it right, you can really really push that middle and and you not not to say replace the middle managers, just remove that part of their job, which is to reduce friction, reduce friction on the front line, and then let them do the things that AI can't do, which is team building and cooperation and all of the human things. and AI can provide better transparency to the top than they were getting and sort of sort of flattens it. and if that's probably it doing its job, you have more frontline, you have better frontline, enhanced experiences with humans, you still get those efficiencies and you have a you know, people at the helm that actually know what's happening.

Mm-hmm. Mm-hmm. I worked for a number of years for the Coca-Cola company in the food service division. It was called Coca-Cola Fountain at the time. And one thing that came up consistently in our research was that a high performing restaurant had high performing frontline managers, period. That was the deciding factor, even at even at regular retail. And I would say you might need fewer frontline managers, but the ones that you have have got to be really, really good. You know, that they are they are the linchpin. Qualtrics has done some wonderful research on the front line. they they did some work for a footwear manufacturer and they figured out that the high performing stores had two distinguishing factors. One that they were adequately trained in all the products, and two, their managers cared about them. It was that simple. That simple. And I think a lot of the things that we're talking about today are surprisingly simple, but we are overcomplicating them. We are over automating. And what you said, tools in search of a solution. The solutions are there. It's,

Mm-hmm.

you know, making sure that we remember that the people that we're dealing with are people and not machines.

Yeah, and the opportunities seem more ripe when you think of it in terms of like augmenting what a a frontline worker is able to do instead of like trying to figure out ways to replace people, right? Like what you're mentioning. Like you can you could have someone working in a shoe store who has access to all sorts of helpful tutoring and real time help so that they can answer questions better and spend more time with people, you know, driving them towards The perfect sneaker.

yeah, service design. We're we're big fans of of this, but let me explain why why the sudden surge in excitement from our standpoint. one thing is that we talk about like AI allows companies not just to automate what they do, but they really allow them to do the stuff they should have been doing all along. planning I think is in my opinion the biggest area if you code with AI, if you do anything with AI today, you understand that the execution is becoming invisible. and the planning is becoming center. So we don't have coding now, we have planning. And now we have the kind of planning we should have been doing instead of saying we should do and not do. and we have this concept of token burn as an objective, but then no way to reconcile that against company value. so you need these assets, these plans like a service design that carries a token budget and And and can get passed around to say, hey, we want to allocate this amount of effort,

Mm-hmm.

AI effort, if you want to call it tokens. and we need we need something for people to sign off on. What is this asset? What is this this this article or artifact that before we just give people unlimited access to token use? And service design seems like one of those really useful. ways from a UX standpoint of s of a human that can look at something in the how-we-should-be-doing-things versus

Mm-hmm.

the what-we're-doing and and the cost of automating that, the uplift of value, being able to see like the simulated return that you can now do and then having humans sign off on on this. So looking at different assets, but service design's one of those ones we're sort of big fans out of and I'd love to get your thoughts around service design.

I love service design. I mean, I think it's just an amazingly complex set of problems to solve. And I think a lot of people like complex problems. you know, Gartner, I learned from Gartner there's a metric you might be familiar with called the customer effort score, which is simply a question, which is how easy did we make it for you to complete your task today? Period. And then maybe a follow-up, which is why did you give us that score? There's also an equivalent one called employee effort score. And when I think about service design, that to me is the ultimate metric, which is how easy did you did we make it for you? Because I believe firmly that it's that simple. That if people feel that you're easy to do business with, that you're easy to engage with, to talk to, they'll keep coming back because there are so many companies that are not. So service design for me, as long as it's in the service, no pun intended, of reducing effort, reducing friction. then that's great. I just don't see a lot, I see a lot of people doing this without consumer insight or customer feedback, even employee feedback. They're just they're just holding up in a room saying, hey, let's redo this process. And they get minimal input and then they wonder why things fail. So done right. I think good service design really leads to low effort and being easy to do business with.

I think there's always been obstacles to service design. You said it complexity, so that's one of them. The other one is we have so much to do on our plate. This is just adding more on our plate. I think it's a A great opportunity to crack it because AI can do a lot of the minutae and lifting.

I absolutely believe that AI can help because it can see patterns that humans miss. if and if you're using the right generative AI tools, it literally has the sum knowledge of thousands or hundreds of thousands of experiences. So just like we talked about earlier about being a collaborator, a companion in solving a problem, I absolutely do believe that. But the other thing which I would say, and you say in your book, and I agree with is that a human's got to sign off on it. they have to sign off. So as long as whether you call it human in the loop or human on the loop, there's a lot of different terms. And yes, it takes a bit of extra work and nobody wants to be an AI babysitter. But at the same token, we have to make sure we're doing the right thing. So I I do believe that AI can get us a lot of the way there.

you talked about people making service design maps with like minimal input, but the good ones are ones where there's like a a good journey map, and a lot of that is often someone physically getting up from a chair and going and having conversations with people in other departments and really understanding how work flows through an organization. I think once you have that information and then you combine that with what AI brings to the table, then you can you can have really big impact.

Yeah, to me it's so exciting. So I'm I'm a big Star Trek fan and I'm also a Star Wars fan. And if you look at the relationship with AI and those two universes, they're very different. You know, in the Star Trek universe, AI is very much at the service of the humans. you know, they they even down to the fact is computer is the is the term used to talk to them, right? It's very much in a service capacity. You do what I tell you to do. Yes, you can give me ideas, but I'm gonna make the decision. In the Star Wars universe, the droids are often partners. You know, you've got you've got some that are servile, but a lot of them are partners. They have opinions, they have their own needs. it's a very, very interesting contrast. And to me, how that evolves, our relationship with these machines, is it going to be mostly in a servile capacity? Is it gonna be a partner capacity? I think that is what is gonna evolve over time. I hope we have both, because I think you need both. I I'm a big science fiction fan and you know, I've been reading about this stuff for fifty years and to see it start to come to fruition is super, super exciting.

I've recently talked to a number of analysts. I tend to get into s conversations kind of like this. Analysts are smart

Mm-hmm.

and so I have fun and they have fun, I think. but there seems to be a theme emerging and that is this idea that companies are kind of either underestimating the complexity of this stuff and sort of jumping in thinking they can just build it and make it quickly and easily or or overestimating themselves. I don't know, maybe both. but there seems to be like a lot of analysts trying to say like guys this is not like you know you know purchasing a a piece of SAAS software here. This is super complex. It has lots of sharp edges. and it seems like that's really at like there was a lot of symptoms, but root cause, one of the root causes is this expectation that this stuff is is gonna be much easier. Like, I you know, I I used Claude Code and I made an agent. I think we should have it start trading our stock now. and and and maybe CTOs that are worried about being relevant and feel like, we we should build this stuff.

Well, I think it's human nature to oversimplify what you don't understand. So I I think and you can also we talked about the hype cycle earlier. They are just replaying the hype, the the marketing messages, that what they read in the press. And that's normal. This is normal human behavior. Part of the role that Gartner analysts play is to cut through the hype and say, here's what's really happening, and here's how you should be thinking about it. Are we right all the time? Of course not. We're right most of the time. but a lot of what we do recommend is caution, not caution so you're paralyzed, but saying, hey, you know, experiment, try these different things, see what works, see what doesn't work. some recent research that some of my colleagues did saying that the companies that are very successful at using AI know when to quit. So whether it's a project or an initiative or a technology, if it's not working, they just cut it off and and retool. Versus many people tend to hang on to it. So I I think that The key here is you can't sit on the sidelines. you have to be responsible of what you're sending. What we tell clients all the time is you have to know what your objectives are, you have to know what you're trying to solve. If you don't know that, then any solution you buy could potentially work or not work. It's it's not that different than the digitalization wave we saw 10 years ago. If you don't know what you want to accomplish, then you can buy any type of technology. So what what I see happening is unfortunately a lot of lazy thinking is the only way to describe it. You know, I don't understand this technology, so I'm just gonna oversimplify it and hope for the best. And I I don't think, given how expensive this technology is, and it is expensive, you can't do that anymore. It has to be much more strategic.

Yeah, it's it's interesting how much tolerance people have. there seems to be something. I don't know if this is part of this hype cycle, you know, characteristics, but it seems like there's a there's something about the early stages of it. I I remember in mobile I sat down with Martha Stewart and this was early days mobile. So I worked on one of the first iPad apps for Apple, like the first eight, and And so Martha Stewart called and said, You know, hey, I wanna build an app to plan your wedding. And we sat down with her and we kinda said, Okay, this is you know, it's gonna take months. I mean, planning an app

Mm-hmm.

to to plan your wedding is no small app. and the end of the meeting she said, Well, I could get a college kid to do this for $15K and I was like, Okay, you know, I'll get

Okay. Good luck with that.

You called the wrong person. I can't. if you could give me the name of that college kid. and and i that was kinda normal back then. It was like no one wanted to spend much on it. They you know, fifteen, $30K there were companies like Bottle Rocket that would come out like, we'll do mobile apps for $25K you know, and and my company was doing mobile like the first, you know, FDA approved mobile app, which was a million dollar app. and then eventually as it matured, you know, people understood this is complicated and this costs money. and so the failure was due to a to whatever that is, maybe they you you're saying it, they just simplified it. and and when you simplify it, maybe you simplify the c cost of it, right?

Potentially. I mean CIOs and CTOs know how complex this technology is, period. They know it. It's convincing everybody else it's complex So sometimes they are forced to compromise in order to get stuff done. And that that is, you know, it's it happens. that's why going back to one of my earlier comments, I think it's incumbent upon everyone, myself included, to try to understand as much about how this technology works as possible. 'Cause if you don't, then how will you know what you could use it for?

Hundred percent.

Cool. This was great. As always. Love having you on.

Thank you. It's it's always a pleasure. I l I I miss having these conversations. That's one of the things I'm getting used to in retirement is is doing this. So yeah, I love doing these podcasts. It's it's a great way to talk and

And the book's awesome. I I think everyone should be thinking about selling to machines. it's an impossible idea to me that that anyone that has a business wouldn't be thinking about this.

Yeah, you know, it's it's you know, Gartner gets things right, but we we call it too early. We call it very early. So even 2023, when we published the book, it was arguably early, but thank God,

Mm-hmm.

you know, OpenAI launched ChatGPT a few months before we published the book. And then people said, okay, this makes sense. We should be thinking about this. But that was three years ago. And, you know, I I think part of it is, you know, the actions that our people are taking. What are you doing about? this and that to me is the next wave of research for Gartner and others is the implementation of this. You know, what do you what technology platform do you have to build? what are what ethics or safeguards do you have to put in place? That that to me is the next big challenge. Great. Thank you so much.

Yeah. Thanks, Don.

I really enjoyed this. It's always always a lot of fun. And again, congratulations on your book. Thank you for citing our work as well as the the the comments from the interview that we did. I can't believe it's been three years ago. So time flies.