How Entrepreneurs Can Compete in the Age of AI: Henrik Werdelin & Nicholas Thorne on Their New Book Me, My Customer and AI

0:00The more that we do beyond the prompt, the more we talk to the world's leading experts about AI. it dawns on me that the questions that we are trying to get answers to are not about technology they're about human AI is such a kind of like invasive technology it's such a technology that you kind of like use as a partner that a lot of the unknowns are questions of how do we define what we want out of it and for us to be able to define it we have to figure out what do we want out of it. I got to be fun to get started. I don't know if everybody knows this, but I'm one of the privileged few who spent a lot of time working deeply with you both.

0:46And actually, strangely, especially to listeners of Beyond the Prompt, it's probably interesting to learn. It's weird to me that now Henrik and I spend more time together on Beyond the Prompt, because for a long time, Nick and I spent a lot more time together. There's a good stretch of time where Nick was more my primary collaborator, Henrik, was a peripheral one. And then the market shift happened. All sorts of things happened. Here we are. So anyway, I started, I helped start at Brown the Prong only for that reason. It was just to get Nick kind of like massaged out of the relationship. It was part of your evil plan.

1:23I was a little surprised. I'll be honest. I, in my mind, when I know y 'all are working on a book and when I see the title and things like Because like what jumps out at me most is AI. And I have to say, I was expecting an AI book, so to speak. I actually mentioned this to Henrik yesterday after we recorded a podcast episode. But what I got was not a book about AI. It's a book about me. And then probably secondarily, it's not a book about AI. It's a book about my customer. And then third, it's a book about AI. And so kudos to you, because I think there's a sense in which people are drawn to AI that actually it's a very reflective book.

2:03Do I get the emphasis right when I say it's actually not an AI book, it's a me book? And how do you react to that thought? I think that is very true when you read it, and not. The more that we do beyond the prompt, the more we talk to the world's leading experts about AI, it dawns on me that the questions that we are trying to get answers to are not about technology they're about human right like it's all about ai is such a kind of like invasive technology it's such a a technology that you kind of like use as a partner that a lot of the unknowns i think are questions of how do we define what we want out of it and for us to be able to define it we have to figure out what do we want out of it and so i think in a roundabout way it wasn't necessarily why we started that way but as i reread the book now i actually am happy about that because i think if we had like written a lot of like and here's how you prompt to kind of like understand your problem better or here's how like a lot of that would be dated like two seconds later um what i think AI does is that it allows us all now to be resourceful in a way that we were not resourceful before.

3:23It allows us to be entrepreneurs now at a period of time where I think having people who solve problems in a scalable way are much needed. And so I think you're right, it's not about AI, but I think those of us who spend a lot of time in the AI world are increasingly trying to figure out how do you create the discotomy of humanity and the more that we understand who do we want to serve and what is the problem that they have then we can use ai throughout the whole process of building to make it easier to do so nick do i do i did i talk about speak out of turn there do you agree with that i think it's the nicest thing that anyone's ever said about the book to suggest that it goes in the meme i like that it goes in order of the title not that it's an ai like you know buzzword thing so that i'm mostly just reflecting on that i i had imagined it as or there was a time when there was a in the spirit of going from a collection of essays to something there was a time where at least i had imagined that maybe we would it would it would live at the you know that the book would kind of exist along the edges of the triangle you know that you would have a section almost about you and your customer and you'd have a section about your customer and ai and you'd have a section about yourself and ai and it was it was kind of along those you know that those were the areas that at the intersections was where we might be able to think and talk and do some work and advance the conversation and so i think like that maybe it existed more at the nodes on instead of the edges i'm kind of thoughtful about because maybe to suggest the next bit of work and in particular i guess i would offer that i'm i think we need to probably do more work on the my customer to ai relationship um how much are is the average end customer of any new entrepreneurial thing what is their perception of ai how much are they bombarded by it how much did they use it and how much do they even realize they're using it uh really that really shapes a lot of all the rest than anyway.

5:32And so, I don't know, the thought process that I arrived at is like, do we need to do more work at where those different edges of the triangle are starting to interact more? You could imagine a follow-up or an expanded edition. You know, one thing that's striking me now, even as we talk, is what's not in this, you know, sequence is two things, my employer or my boss. And what strikes me now, and we've got balloons for that realization. What strikes me about that actually, which is quite interesting, I was having a conversation with a really good friend who's got a child in college who's studying computer science, which is like many families right now.

6:15And they're bemoaning the fact that internship opportunities are scarcer than ever. You can't find a job. And I found myself saying to this friend, one of my favorite expressions is an inventor is never unemployed. And one might say an entrepreneur is never unemployed. employed and what i realized is i think a lot of especially young folks paradigm is the next thing i do as i get a boss the next thing i do is i get a job and i wonder if this is an opportunity to shift the paradigm the next thing you do is get a customer and what i told him was hire yourself you should tell your son stop looking for a job and hire yourself yeah i don't know if you've thought about, you'd call it the, you know, niching down from the roughly 50 % of people who say they want to start a business to the, you know, people who actually do or from 60 to 10.

7:10What do you think about specifically young people and the moment that we're in where the AI on the bottom of the page actually is ominous, is actually scary? How do you think about them? I think it's the main why now of the book that I think we're all, whether we have stats or data or just, you know, anecdotal evidence to suggest that we think it's probably increasingly important that people come to grips with their own, like that going from the 60 % who've had an idea to the 10 % who do something about it. I think, you know, making that lead used to be a nice to have. And I think even our own work in 2010 to 2020 period is like, I definitely don't feel like I did go into work every day.

7:53I thought it was fun. I thought it was like intellectually interesting. I thought it was potentially financially rewarding, but I don't think I thought it was like important for the world that we help more people start things other than like that on the margin. I think that we see the people being feeling fulfilled and that's a good thing. But like I think we're probably in a world or heading to a world where more people will be faced with the reality that this is their primary option. So I've certainly I don't know if that's exactly what you said, but I feel that. And I think you see that. I think you see a lot of people who are un or underemployed, whether they realize it or not.

8:26I think that's true for very young people. I think that's true for people, you know, coming into the middle of their career and realizing opportunities are not quite what they thought they were. So I just think a lot of people are going to have to start going from the question, like, who's going to employ me to who's going to be my customer? Because I think the last job is maybe you can never be unemployed or whatever. That was a nice sentence that, you know, inventors are never unemployed. to Oxford and they were unemployed. And we're headed to a world where like everyone's going to have to be on, you know, not unemployed or the thing is there.

8:56I also just will add one interesting little tidbit to your, to the statistical reference and your point of like, be your own boss. You could easily, and actually the numbers I think are identical, substitute the 60 % have had an idea for a business. And we actually debated this with 60 % of people want to be their own boss. I think that the data is like legitimately identical almost. And I'm not sure the venn diagram is 100 overlapping i think they might be different motivations um but i think the be your own boss thing it fits a similar profile in terms of both like nice to have versus need to have the the way ai is shaping that and how ominous or not ai is in the mix of shaping this kind of conversation i hadn't thought about it in the way that you phrased it jeremy but i think that's very astute.

9:43The two things that I thought about in that space is increasingly I look at organizational design and we obviously write about membrane organizations as a kind of like a new concept that kind of like goes all the way through customers all the way to the supply chain. But if you look internally in our organization increasingly the way that I compute it is that everything is a resource that you prompt that gives you an output and that doesn't matter if that resource as a human one or as an AI one. And I just expect to have somewhat like of a high output when I do a prompt. And so if we are moving tactically towards a world where there is chief resource officers of companies and they are responsible for agents and humans and the way that the organizational design and communication structure is organized is, you know, basically like a computer system with APIs, inputs and outputs.

10:44I think that's one thing that would lead people to need to understand and work more entrepreneurial even when they have a boss. So I think that's one thing. The second thing is, you know, Jeremy, as we talked about yesterday after our recording of the podcast, like we had a whole podcast yesterday with a guy who runs the kind of strategy institute of BCG. and I mean in the same way it was an interview about AI but it ended up being not about AI because what we are asking people to do now that they have these Iron Man suit is to act like an entrepreneur when they have a job it is to be more of a journalist it is to have agency it is to you know like understand what data they use you know like in the work so suddenly like the job is less as part of a sausage factory, and it's more like a Navy SEAL that gets deployed to solve a problem or an entrepreneur that's trying to build a company.

11:43And so in that way, I think it's a very interesting point. It does not have a boss in the title, but that is because we're going to live in a world, I think in our optic, that requires everybody to be more acting like an entrepreneur and with AI. You know, I don't know. I've just been reflecting even myself. I don't know if it's because of my upbringing. I grew up in the middle of the country, you know, to parents who had very, call it conventional jobs. Entrepreneurship wasn't on my radar as even a possibility until business school. And so I wonder, I don't know, it's kind of an essential question, but how do we, you know, push it down the stack, so to speak, of viable career options.

12:30It wasn't, I mean, literally I was in my thirties before I even knew it was a possibility. So that's just kind of an essential thing. I want to talk about, there's kind of two frameworks that I think are worthwhile. And I've got a bunch of notes from the book we can dive into if we want to, but two things that I would love for you to talk about in terms of me and my customer, because I think anybody who's listening to this show later, especially is probably, they want to know what are some very tangible things they could take away. And so in regards to me, I'd love for you to talk about the five Ps.

12:59And then in regards to my customer, I'd love for you to talk about it sucks that. Can we start with the five Ps? If you think about this first and foremost being a self-discovery field guide, what are the five Ps and how does an individual use them to start to guide their exploration? The five Ps are powers, passions, possessions, positions, and potentials. Took me a long time to, but Nick wrote that chapter. So, Nick, this is squarely in your account. Is this like a behind-the-m? Is this inside the - I hated that part to start with, and then I became kind of like the evangelist for it. So I went like full 180.

13:38I was like, this is dumb. It was just a big job style. You know, Ron Johnson told me that he had a car ride with Steve Jobs, and he was talking about how they needed to reconfigure the entire kind of the organizing principal of the organization before the Apple store came out. And he said Steve vehemently argued with him, was yelling at him in the car, berating him in the car, all that stuff. And then they get to the store and Steve verbatim says the thing that Ron said in the car as the new strategy. And I was like, wait, what? I mean, obviously, Henrik and I have collaborated for whatever, 15 years now.

14:16So I think this is actually very interesting thing um if as it were as a quick aside like that you know insight into what it is to collaborate with someone for lunch i don't think it's specific necessarily to ours although maybe it is like you know there's um it's a funny thing when you you know someone well enough to be able to trust them with your first draft but you also like them know that the first draft's not always gonna like resonate and sometimes it takes time and sometimes you know like something i guess maybe the comment i'm going to make because for people who are as communicated frequently as we do and who i think communicate well like i generally think that i can say things and henry here's what i'm saying like you know that i'm struggling to find words but but he can kind of like parse through what i'm saying i i hope that roughly the same is true i swear that there are still times where you're saying something and the other side's not really picking up the other end of the line you know like it's just you're just saying it and and uh i'm not that like committed to ideas I guess I would, I don't think I am.

15:14I fashion myself as quite stubborn on a number of things, but ideas for one, I'm very happy to like, if it comes out and people don't like it, I'll be like, all right, whatever. So I don't get emotional about it. And this was one that was actually just very practical because we had this issue, which is like, we were kind of going along in our own process of like, we'd already made this leap to say, hey, and it's kind of leads to your next question of like, if you're trying to, if you're in the position of trying to come up with a business that you could start, if you're like i want to start a company i'm not really sure which one to start i have an idea but typically we would think that it might be not a bad idea for you to like trust test your own idea even with some other versions of your own like what else could you work on just so you have like a menu of options put yourself in a little bit of a better position of leverage well like we'd already made the step of saying you certainly give yourself a big advantage to not be anchored in a solution by the way i did this i spent four years of my life like started a company where we basically started with the solution-ish and we then kind of like had to kind of you know riff around that you just like limit your optionality a lot if you start with the solution so the problem would be a better place just like the architecture of getting going benefits and is stronger and more resilient if you have like centered the architecture around the problem you're trying to solve but very quickly you also realize well like what if I got that progress is?

16:34And like, what's my personal relationship to it? And why am I the right person to solve that problem? And you start asking a bunch of questions if you kind of abstract that, which at least one answer to is, well, like, do you give a shit, excuse me, about like the person or group of people that you are trying to help? Do you have some insight into who they are? Do you speak their language? All these different things start to become the questions as to why even pick that problem. And so that then became obviously a topic of like, you know, who are you and who is your customer and why are you even thinking about the problem to begin with why why do you have any confidence the problem is actually a real problem and so it was kind of a natural you know extension of going from solution to problem to kind of customer that we had ended up in this debate about like well how do you even frame your own personal exposure to um the question of like who should you help you know what about tourism well largely we created these different frames of reference to try to think about like a way of understanding yourself so we said powers which i think is basically like what are you good at um you know i worked on wall street i happened to be okay at the time at like financial model well like maybe one of my capabilities is financial modeling so maybe i can like help other people who i don't want to become financial modelers or who are also financial modelers i have credibility with them and so maybe i can build product or service for them so that would be like an existing of a power a potential is kind of the opposite version of that is I'm going out of order.

17:58So I apologize. This Henry knows them better than I do at this point. Potential is like, you know, what do you want to get better at? So what are you at least like intrinsically motivated to get better at? What are you studying? And if you could, if you are willing and able to study that, either and share your process, or once you arrive at the answer to how to get better at it, like you could share that with other people, that might be a good way to find other people to help is like, you know, who else wants to get better at that thing um and so that was another place where we felt you could kind of like look for uh founder customer fit you could look for a kind of relationship uh building opportunity positions is kind of probably the most traditional one which is just like look at yourself and what roles you play in life whether those be personal or professional and try to anchor your point of view on the like who your sole customer is around that so that's i'm a dad i'm a homeowner i'm a i'm a dog parent but like maybe accordingly when you think about like a dog parent um and and henrik obviously can talk to this much more like there's often like a little bit of nuance in how you view the role if you've played it you know it's like it's not i'm not a dog owner but dog parent it's a specific flavor of this thing i'm looking for other people who view themselves in that capability in that role um uh so that's position uh possessions is also kind of obvious another way of thinking about yourself is like what are the things you own and invest time effort of money in and why you know like if you're a hobbyist uh you know boat owner well like that one way to think about who you are and who other people um that might be your sole customer by what what what they kind of own or kind of curated into their like primary kind of collection of of things to own maybe the other way of thinking about dog parents i think they own a dog and they have a sense of responsibility to this dog so maybe you can help other people who are like that yeah they can passions passions yeah so similar i i personally think that positions possessions and passions are kind of the obvious ones and they all interrelate they're all just different frames of reference on the same thing like i'm a guitar player i own a guitar i'm passionate i'll learn a guitar those are all or whatever like maybe those are all just different ways of thinking about the same thing if you have a line up across all of these you know like i'm pretty good at the guitar uh i i am a guitarist but i'm also really passionate about getting better at the guitar, you know, like all those would be things that maybe you should really think hard about building something for other people who are, you know, in the guitar category.

20:20Well, I think, I think just to make it, you know, actionable for anyone to realize, okay, these are frames of reference that I can collide with my own kind of question of what could I start, right? Anybody who's talked to me for more than, you know, one single session knows my second favorite quote about creativity is Arthur Kessler's, creativity is the collision of apparently unrelated frames of reference. And what you've done with the five P's is you've given people unrelated frames of reference that they can collide with this question of what could I do? Where should I start? I feel like, by the way, this is imminently GPT-able, right?

20:57You can imagine a five P's GPT, especially if somebody's got either custom instructions or history turned on. Like, wouldn't it be amazing if the GPT just knew, just could interrogate you and then know from your, from what Chad Giputino's about you to make five recommendations based on what you already talked to me about. Here are your five piece, five potential businesses that you could start, right? And the other thing I'd add to that is I think it's also an interesting frame of reference when you are kind of evaluating one of your own ideas. And so very often, I would imagine that people would not apply that framework of working forward from the framework.

21:33It's a great thing. Probably true about the, it sucks that stuff often too. And I think Henry, I can probably talk more to this. I think, you know, our heterogeneous ideas often come in the form of like, what if we did a thing that does this? And like, I think a lot of people, I do that, but I observed it in someone I spent a lot of time with. And, you know, and even though we've written, have spent so much time not being in solution mode, the idea is first, Verbergen is often a solution. But baked into it, implicit in it, is a problem statement. And it is a founder-customer fit, is an understanding of who the customer is and why to have the problem.

22:07and then and then the solution is what kind of like guess articulated you know um and so sometimes these are also just useful frameworks for how do you take this thing that's nagging at you is in the back of your mind as an idea like hey i'm only looking at this rain boot behind me you know like we should ship mom's rain boots on a subscription because the kids grow out of them too quickly and they only wear them once because it only rains once for every six months and they get too big so what if we did that like that's a solution but i i already started to get out like who's the customer and what their problem is and whatever.

22:38And so if you work backward, especially in the AI age, and you're going to have to be clear with people as to why you're doing this, everyone's going to be able to make a copy of everything. Why you are unique. Why are you the person who's the best person to do that? It's going to become a big part of the logic. And so in the absence of having that, you know, at least just extrapolated your own connection to it, I think you'll struggle. And so at least maybe if it's not something that you can move forward from, it is something I think could be a framework for thinking about, like, why am I interested in this?

23:09Well, Nick, what you're doing is you're suggesting a second GPT. The first GPT is actually an idea generator. The other, a 5P-based idea generator. The other GPT is a 5P-based idea evaluator to give you frank and honest assessment of whether you are uniquely positioned to do the thing you're proposing based on your 5Ps. Hey, Jeremy, what you then could do is you could take all the different tasks that an entrepreneur has to do to go from, I'd like to start something, but I don't know what to start until having an actual business. You could string them together and you might call them autos. Just because it would be automagical, is that what you're saying?

23:48Exactly. Okay. What about it sucks that? So we talked about five Ps and it being a great a place for an individual to interrogate their own assets for where to start an entrepreneurial journey. Why is the next thing It Sucks That? It Sucks That was something that we've been working with all the way through the pre-hype days when we wanted to start something new. And it was fed out of the insight that problems are better to find than ideas because ideas are really just thesis statements against problems and a problem is it's it's difficult to just come up with a good idea from like initially but it's it's difficult to know if it's a good idea it's easier to know if something is a real problem or really felt problem because people are very good at articulating issues uh and i think it started all the way back when you know i was traveling around the world and And I would ask people, you know, like in sitting in the plane, I would ask like, so, you know, you ever thought about a business idea?

24:57And people would clam out and, you know, they'll give me their NDA or whatever. Right. And then you say, oh, interesting. Like, what's one of what's a big business problem you're working on? And people are like, how long time do you have? And they'll kind of like articulate all these different problems. And so we got obsessed about this kind of concept that if you could help people identify a real meaningful problem, then there's probably a good business somewhere at the end of that. If you started with a problem, not with the idea, then you had more shots on goal. Because if the first attempt of solving the problem with the first idea didn't work, then you can always go back and you wouldn't have wasted all the work and the access to the customer and stuff.

25:38because then you might just kind of like attempt it, you know, something else. And thirdly, which kind of tied in with the five Ps is that for many people, they don't realize until too late that they have to work on this thing for 10 years for it to be successful. And if you don't care about stuff, then it's just very, it's a long time. And so what I have met in my career was people who like had this probably even fine business. they just kind of hated working at their own business because the customer that would call them every day were just people they didn't really give a shit about. And so a lot of these different things came together and we got obsessed about first, can you identify the problem?

26:20And then second, can you reframe it? And then can you test yourself into figuring out which one of the ideas is the best thesis again? The underlying idea of it sucks that is that it has to be at something that requires a painkiller, not a vitamin pill. It needs to be something that people actually consider to be big enough for a problem to actually react to it. So for example, if you show the mat saying, don't you just hate when X, then it needed to be meaningful. And using the kind of like the loaded words of it sucks that instead of like, I find it to be an inconvenience when, then um then that kind of like kind of just intuitively got into that matlib format so that's the the premise of it sucks that we see we have all the time i even have it on my post-it notes right you're like we we're now as an organization just trained in seeing like when i see hear people say something i you know my eyes light up now because i'm like oh business opportunities You know, I mean, since the 1960s, I think Bob McKim started doing this, but at Stanford, he would assign early design students who wanted to launch businesses.

27:33He would encourage them to keep bug lists. What do you call bug lists, right? Which are lists of things that suck. You can say, you know, you might have an it sucks that list, right? Right. But what McKim knew, even in those very early days of the, you know, long before the D school proper came around was a problem is a necessary precondition to a solution. And if you want to be the kind of person who generates solutions, you have to be the kind of person who notices problems. And so it sucks that I think really it does a couple of things. One, it keys people in on, you know, problems as as kind of the anchor point.

28:10But then, too, to generate your point, the language itself really matters because it increases the likelihood of a worthy solution, right? Nick, I'd love for you to say why, you know, for the people who are prudish, who don't like this phrase, it sucks that, what's the case for preserving slightly offensive language here? yeah well for one i was just in bhutan and i was giving a lecture to a bunch of high school students and i was down the way autos and i got to the part where we were where autos as the first ai agent that you interact with when you're trying to start something with autos does the it sucks that thing and everyone starts giggling and i wasn't really paying attention i was like what what's everyone giggling what are we laughing about and they're like oh the you know they're all like pointing you have to word and i it's kind of a jovial version of the point i think it's visceral language and i think that's what manner of sin we're going for you're trying to create a linguistic framework in which it just would feel wrong to fill in a like nice to have like it doesn't suck that um you know my coffee's not cold enough right now right like you know like if you were trying to create a bias towards filling in the blank with something on the margin like just marginally more visceral marginally more insightful marginally more you know noise cutting and um i think that's the case part and we by the way there's all sorts of interesting i think we actually have chatted about this before but i'll just add like once you apply these frameworks and one of the very interesting parts of having written a book at the same time as we were writing prompting so roughly the same time we start writing this book is when we get access to gpt3 and we start writing prompting it so much of the the book was written almost i'm not going to say we were like super purposefully saying oh we're going to be able to load this book into a model and see what it says but we were certainly you know starting to become aware of like the prompting or like and i had a talk telling it what exercise to run telling it what these things were were going to be parts of our own ability to deliver our interest and expertise to other people.

30:26And so we certainly spent more time thinking about what words do we use. And so this is a place where we, with our agent, people will say, stop saying that. And the thing will say, it'll try to explain why it's doing that bad. But I think visceral language is probably only more important in a world where average language is commoditized. It's one of the only places where you can try to use some sentences, use some words that no matter. So the role in it sucks that it's probably only become more important as we put some AI around our own tooling, because I think it's a way to express perspective.

31:01It's a way to kind of like, you know, show what you believe to be true that maybe the average, you know, answer doesn't otherwise deliver. Yeah. Yeah, I think that's exactly right. When average is easy, what you're always trying to do, you're effectively, and going back to Henrik's comment about everything is a prompt to an intelligent system insofar as the book is a prompt to an intelligent human reading it. The elegance of it sucks that, which sounds paradoxical, but it's not. The elegance is it influences the thinking of the intelligence differently than what's mildly inconvenient. Right. And just like with as few tokens as possible.

31:39Exactly. With this perfectly set. Okay. I have two questions that I want to try to get to you before we open it up to the audience. We'll see if we can or not. But one of the things that you addressed actually in the intro was, did AI write the book? And you talked about how, hey, we tried. And I'd love for you to just talk briefly about, not that you tried, I get that and everybody gets that, but I would love for you, because you're such knowledgeable and thoughtful collaborators to AI. I know the fact that you tried means something. I'd love to know what you tried actually, why it failed, and whether you think that will continue to be the case in the future.

32:22We basically tried to make it Airtable, like a database, and kind of like turn each row of the database into a paragraph, if, and, to like almost write a script that would take whatever the immediately prior, like knew the outline of the book as we, as we kind of understood it at that time, knew the chapter. It was worth it. Like, you know, if you imagine it's like super, every row of the air table is a, is a piece of an outline. So we like pre-generate the outline. We could kind of like have like Roman numeral one section of the name, you know, small Roman numeral one, you know, part one, paragraph one.

32:58And if we could just like, you know, break it roughly down as much as possible, that maybe it could just like write the paragraph or at least the next sentence and um and so just tried to be very like you know systematic about the manufacturing of the paragraph knowing that we had a bunch of source material to at least have like helped inform both the outline and maybe sell in a bunch of the gap so i would all observe that i'd think that that would lead to a very like mechanical feeling but if you if you even achieved in a b the context windows weren't as big you know a year and a half ago as the car today so maybe you have better luck today but that was roughly a version of what we tried was to use our own way of doing paper to like you know no job is too big once you've cut it into enough pieces and and to just try to slice the number of pieces so finely that you were asking ai to do something like really pretty straightforward like i just need a paragraph about x okay in the context that i've already provided knowing what you know is what you know knowing where you are like orienting you and knowing kind of the way it is and then hope that maybe you get a little bit of like recursive kind of building as a back of that because as you start to kind of build momentum like it like it understands what to do next like almost using the fact that we created a source material to be like the introduction itch and maybe see where it would travel from there and why do you think it failed i think it failed at the time because a i'm not sure that's how a winter book really gets written you know like you can't just like go forward from one sentence and see what the next sentence is.

34:33There's some degree of like planning involved with the overall arc. I'd also just think it got lost frequently. And lastly, I think the key thing that like took the essays that we already had, and so what we were really asking the AI to do and made them a book was the humanity of figuring out like one, how do you make this like legible across the board? And two, what we were probably really shy on were case studies and i think asking a gpt for a bunch of case studies is a pretty quick way to like end up with the first google result of like you know whatever whatever you would have otherwise searched for so we were probably trying not to just deal with like you want an example of that style of entrepreneurship you need to look no further than you know joe schmoe founded coca cola and you're like oh a personal relationship with the coca-cola case then great you know okay last question and then let's over the way can you well do we have time for because hendrick said you asked a bigger question oh yeah whether i think you'll learn for and i think that's i i'm not sure i know the answer to that but i'm curious when hendrick's answer then i i wonder but please hendrick would love to hear i think we're getting close to ai being able to write a book right like the the question is not can it write it it's can it write something that is meaningful to a human and i think that is is an area that we don't really have good terminology for yet of what's happening and why that may or may not happen when i do something amazing with ai i turn around to nicholas and go like look what i just got the you and the model just did like what we just got to what i don't do is i don't go to the model like that was incredible we just did like let's just stop here for a moment and like talk about what you just did there man I just don't have that response.

36:21And so I think that there is something interesting in the human in the loop, because I think that we have been trained for so many years as humans to connect to that. So I think for most of us, that will really matter. and and i think you could have like ai writing thousands book but then i think a human would pick the one that is mostly interesting for their affinity group and they'll say you should look at what this you know you should read this book and so yes the ai might have written the words but it got its relevancies for humans because a human picked and kind of contextualized it um but you go into juno and you write a song now like there are good songs there that i could put on my playlists just fine.