← The Growth Podcast
Loading summary
Transcript87 lines
- [00:00]
A
Is Sam Altman just a bad CEO or is he trapped in a bad structure?
- [00:04]
B
The focus on Sam the person obscures the structural problems that need to get fixed, not just for OpenAI, but for all of these companies.
- [00:13]
A
Eric Ries is the lean startup guy. He created the most successful movement in tech ever.
- [00:18]
B
For me, the worst scandal is the time that they put asbestos in the baby powder and covered it up.
- [00:27]
A
What?
- [00:28]
B
Yeah, yeah. In baby powder of all things. The killer that's about to kill you is already inside your house. Most of the time when companies collapse, the problem is internally.
- [00:37]
A
Can I copy paste these governance features? And will I have an incorruptible company?
- [00:41]
B
No, absolutely not.
- [00:42]
A
So you write about this concept of financial gravity, where success is actually pulling companies away from their mission.
- [00:49]
B
In order to understand and build an incorruptible company, you have to understand the deeper forces that act on organizations. You can say, well, I don't believe in gravity, but guess what? Gravity believes in you.
- [00:59]
A
So what were the choices you and Jeremy made? I believe you actually have answer AI as a for profit. So how did you structure it properly so that it's incorruptible?
- [01:07]
B
I just want you to ask yourself one simple question before you agree to what they say.
- [01:14]
A
Before we go any further, do me a favor and check that you are subscribed on YouTube and following on Apple and Spotify podcasts. And if you want to get access to amazing AI tools, check out my bundle, where if you become an anal subscriber to my newsletter, you get a full year free of the paid plans of Mobin, Arise, Relay App, Dovetail, Linear Magic Patterns, Deep Sky, Reforge, Build, Descript, and Speechify. So be sure to check that out@buildle.akashg.com and now into today's episode, we have a different episode for you guys today. You, you come to this podcast for a lot of aipm tool breakdowns and AIPM courses. One of the things that people don't talk about enough is governance. And in aipm, it really matters. AI safety and ethics is one of the top things that AIPMs are evaluated on in interviews and on the job. And governance is one of the most important angles. Today we have on one of my personal heroes and legends, Eric Ries. He's kind of, as I told him, 20 years ahead of me in his career. He's created the amazing book, the amazing movement, the amazing brand. He's run a bunch of companies through it, some of which you'll get to hear about in today's episode and in today's episode, he's gonna break down the famous OpenAI versus anthropic structural battle. He's gonna walk you through exactly how companies get corrupted. He's gonna help you understand Enron versus Costco, and he's gonna help you sell. See how you can structure your company, how you can make decisions on a day to day level that actually contribute, that actually are incorruptible. So if you want to get better at the craft of business, if you want to get better at AI, this episode is for you, Eric. I think the structure that is on the top of Everybody's mind is OpenAI. OpenAI started as a nonprofit, it created a capped profit arm, then they just converted to a for profit pbc.
- [03:09]
B
Yeah.
- [03:09]
A
So based on everything you just showed us, what does that structural trajectory tell you?
- [03:15]
B
Yeah, OpenAI is a really funky example and it's kind of hard to know what lessons to draw from it because it's such a singular and strange story. I mean, if you've read that giant New Yorker profile, Sam Altman, it's like mind bending how complicated the whole thing has been. And obviously there's big personality people involved, Elon and Sam, and is complicated. And OpenAI has been anything but stable. I mean, it's really depending on how you count like two or three crises that have almost ended the company. Like, it's a recurring theme. But if you look at this diagram, OpenAI, for all of the attempts they made to protect the mission, it never had this structure. It only ever had one board. And I think that was a big mistake. Now, could they have known that all the crazy things that were going to happen were to happen? No, I think they were caught very much by surprise. So I'm sympathetic and I'm not here to throw stones. But I think it's really interesting that the number one lesson of OpenAI to me is that people think that governance means whatever it says on the piece of paper. So the board of OpenAI very naturally, like, look, it's our responsibility to hire and fire the CEO. Right. We have that power. And in fact, Sam used to brag about that. If you look at his congressional testimony from before the crisis, he would say stuff like, the reason you can trust me is I'm accountable to a non profit board who can fire me anytime. But when they actually tried to fire him, they did it, by the way, in the middle of trying to close a tender offer that he had personally negotiated. So, so pro tip, maybe don't do that. But that created tremendous financial gravity that was aligned with Sam because all the employees were about to make millions of dollars personally from this tender offer. Microsoft, their number one investor and their biggest supplier, was livid because they hadn't been given any kind of notice. And the other investors were all tied to Sam. He had raised all the money. So this confluence of events meant although the board had power on paper, the actual power was held by the employees, the supplier, and the investors. That's not an uncommon situation. So part of the goal of studying governance, like part of the reason why it creates career acceleration for people, whether they're founders or not, is governance gets you a seat at the big boy table. Okay? This is where the grownups meet to discuss the really consequential things. And governance is a study of power relationships. Who actually holds the power to do what? And it has that political character to it. The question of, like, if this company, this organization was like an ancient Greek polis, right, a city state, what kind of. What kind of culture will we have? Are we more like Athens? Are we more like Sparta? Well, every organization really reflects somebody's vision of utopia. It's somebody's idea. This is the proper power relationship between people. And organizations have this space, special quality that when they grow, they turn the outside world into themselves. They metabolize resources, and they make the utopia bubble get bigger. Someone once described. It was. Benedict Evans described Amazon once as a machine for creating more Amazon, turning more of the world into Amazon. Like, that's what it does. And I think that's true of every company. Like, that's what it does. If it has integrity, it grows. It makes more of the world the way that it thinks it should be. So. So if you ask yourself from that perspective, based on everything we've heard, like, what was the ethos, what was the integrity of OpenAI? I think we're left with a lot of questions.
- [06:24]
A
Earlier, you had mentioned how a lot of founders give you a call when they're creating a company. One of the craziest stories you tell in the book is that Dario Amode gave you a call in the early years of Anthropic. So when you analyze their governance structure, are they implementing things in a better way?
- [06:42]
B
I think so. You know, of course I put a role in doing it, so of course I think it's great. For the record, okay? Like, people. People misunderstand this all the time. I am not taking credit for anthropic success, okay? All credit to Dario and that exceptional team for all that they have done. They are crushing it. Amazing. Okay? Well done. Nor am I saying that they are perfect and have never made a mistake. I can tell you some mistakes. Okay, no problem. But I think it's really important to understand that what makes them different. And I've been called by a lot of companies over the years, they were already like I met them right after they had left OpenAI. Okay, so they were still smarting from that whole split. They were still trying to figure out what the seed rounds were. They were trying to figure out what is our very first investment going to be, what should the team be. And they were very concerned with their ethos of AI safety and say what you will about that, whether they have it right or whatever, they sincerely believe it and they have really put it. Their actions speak louder than words, as you know, as my father used to say, right. Like they have put their money where their mouth is and they have shown that commitment over and over again. So to me they said, look, we want to make sure that we are going to be able to protect this ethos. They didn't use that word, but that's the word I would use for it, protect it as we scale. They understood that this was going to be a multi trillion dollar company or more if it worked. Which meant that the temptation to steal the technology, to use it for some evil purpose was going to be immense. And they needed the tools to resist. Now they were very lucky. They had aligned investors who were on board, or so it seemed. And we talked about this exact structure, the need for a mission locked vehicle. Now to do that required them to def this idea for two years as they raised their series A, series B. And I think it was in the series C they finally enacted what they call the long term Benefit Trust. The LTBT is the formal mechanism to have these trustees who are AI safety experts who have us who are able to appoint directors to the for profit board. And they have. It's a very complicated thing but as the company achieves milestones, the trust gains in power so it's able to appoint more and more circumstances. It's a very clever solution to a very serious problem. And if you've ever wondered how come anthropic just seems like a little bit more courageous than the other AI companies. Now you know why it's not just Dario personally. I mean after all he's a first time founder, technical founder, he's great. I love him a lot. You know, I love him to death. But like this is his first rodeo. Most founders who are doing it for the first time when the stakes get High chicken out because they don't have the appropriate protections to support their conviction. And, you know, I think the difference, you know, partly is about Dario's character. I admire him greatly, but part of it is about the proper structure that Anthropic has surrounded him with.
- [09:16]
A
Your new book has this framework at the center of it. I myself was lucky enough to get this copy that your team sent me. So as I was studying it, there was something on page 91, I believe, which is where I wanted everybody to start.
- [09:31]
B
Okay, you got it.
- [09:33]
A
So in this, you're talking through the blueprint, and you talk about it as the financial blueprint. So what are the four responsibilities that each company needs to get right?
- [09:45]
B
Okay, so let's take a step back, because when. When we talk about being an incorruptible company, people want to know, well, what's that? What does that mean? And they hope we're going to talk about something exciting and fun like, ooh, is it about strategy, business model, culture, stakeholders, whatever? No. Wrong. In order to understand and build an incorruptible company, you have to understand the deeper forces that act on organizations that are invisible. You don't see them in the same way that you can say, well, I don't believe in gravity, but guess what? Gravity believes in you. So when you jump up in the air, you come right back down to earth. The same thing happens to organizations. If there are cracks in the foundation, if they're not strong enough, they get pulled down by financial gravity. So for most founders, most leaders that I work with, they don't even know what the word governance means. They've literally never read their corporate charter. They think it sounds like a building permit, like some kind of thing for lawyers to fuss over. But actually, corporate governance is the foundation, the constitutional structure that determines if an organization is strong or weak. And so if we want to build an incorruptible organization, we need to do two really important things. We need to make sure that the internal stuff is strong. Everyone is aligned around common metrics, common goals, common purpose. And we have to have that strong exoskeleton that if people try to tempt us or pressure us to abandon the mission, we have the strength to resist. So, yeah, if you want, I'll share my screen. If you want, I can show you the diagram.
- [11:10]
A
Yeah. You have listed four responsibilities. If you can walk us through the four. And then I think you said most companies only get one of these, Right? So why is that?
- [11:20]
B
Yeah, So I call it the new governance. Like, it's A new idea about what corporate governance can and should be has four dimensions. So our goal is to create something worth protecting, okay? To build a trustworthy organization. Trustworthiness is the most underrated asset on the planet right now. And in order to stockpile it, we have to really get these inner dimensions right. But then if you stockpile a super valuable asset, guess what? People are going to try to steal it from you.
- [11:47]
A
What do you expect?
- [11:48]
B
So one of the things I think I really didn't understand earlier in my career is that to the extent an organization is successful, that just makes it a more valuable target for someone to try to capture. So instead we want to make sure we build with structural integrity. So the foundation of all governance things is compliance. Today, pretty much every board is focused only on compliance. That means making sure we follow the rules, making sure no laws are broken, making sure there's no self dealing. That stuff's important but well covered elsewhere. And then purpose is the second dimension and maybe the most important dimension. Today we live in the era of what's called shareholder primacy, which holds that the purpose of a corporation is just to enrich its shareholders, and often just its short term shareholders. This is a pervasive legal doctrine that has spread all over the world and even metastasized out of business into almost every kind of organization today. So that's pretty much it for modern boards. They're pretty much focused on compliance. And then to the extent they talk about purpose, they talking about enriching the shareholders. I think we should add two more dimensions to that. First of all, purpose can be a lot more rich than just financial purpose. We'll talk about that. The second, the third dimension is coherence. What I was saying before, to what degree is everything internally aligned around that purpose? So for example, is it possible that the organization might be tempted to make money by betraying the mission? This happens all the time in modern companies. We've all had that experience of a favorite brand, you know, that loses its way and then all of a sudden starts using cheaper products or the food starts tasting bad. I feel like I talk to so many people who like if a private equity firm takes over their favorite restaurant, they say they can taste it in. The food ownership structure of a company has a certain taste to it. That's how pervasive this problem is. And then the last dimension is what I call integrity. And integrity means both as it would with a person, the ability to make and keep a promise, but also structural integrity. To what degree can an organization be bullied or pushed off course by outside forces. When we have this complete package together, we can build an incorruptible company.
- [13:56]
A
Which of these is the hardest to achieve? Which do companies have the most trouble with?
- [14:00]
B
Well, hardest to achieve, they're kind of from easy to hard. So most companies, like, do okay on compliance. They maybe take an attempt at purpose, like they have a mission statement or something like that. I don't think that counts, but at least it's an effort. Such companies are not mission driven, they're mission hopeful. And then as we get further and further away from the things that we as leaders directly control, it gets harder. So coherence is harder because purpose is something we fundamentally choose. It's a matter of our intention, what are we trying to do? And generally, when we're starting a company, we can write whatever purpose we want directly into the legal documents. Coherence is an operational discipline. We have to get our employees and all the other people we touch aligned. That is not the easiest thing to do. But integrity is certainly the one people struggle the most with because that relates to boards of directors, to our relationship with investors, and our relationship with the broader financial system.
- [14:51]
A
When it comes to doing some of this coherence work in product building, what is the role of the executive team, the product leadership, and then the actual engineering team on the ground? How does that split up? Who's responsible for what parts?
- [15:07]
B
Yeah, so product is usually in a product led company, Especially product is where the rubber meets the road. So a lot of companies, a lot of leaders tell me, I believe in product quality. I'm like, oh, do you? Is that your corporate purpose or just, you know, a slogan you want to use? For a lot of people, it's just a slogan. When push comes to shove, when the pressure comes down, they will compromise on product quality, they'll compromise on design or whatever, on efficiency or whatever it is that they hold most dear. But in mission driven companies, it's not like that. Everybody's aligned around that choice. So I'll give you a funny example. People hear the word purpose and they're like, oh, no, it's some kind of ESG thing, right? It's just like, sounds vague and like, oh, what are we going to talk about? And I remember reading an article about a very famous big time public market investor who was criticizing Unilever, the big industrial giant. Unilever had said it wanted to infuse purpose into all of its products. And the hedge fund guy was like, oh, come on. At the point that we're talking about the purpose of Hellman's mayonnaise. I think we've lost the plot. And I was like, that's a, that's a good line. Like, I totally hear you, man. But interesting to me is like, funny you picked that particular example because actually Hellman's mayonnaise, humble though it may be, is full of food. So its purpose is actually very clear. It's to nourish and delight the people that eat it. So people think, well, what does it matter what the product manager thinks the purpose of Hellman's mayo is? But I actually think it matters a great deal. For example, imagine an efficiency consultant shows up and it's like, hey, product manager, I just realized we could save like 5% of the cost by using this carcinogenic ingredient. Or hey, why don't we lace it with us the tiniest little bit of heroin. No one will notice, but it will make the product super addictive. What do you say? They might be like, well, I don't know, man. In the long run, that seems like it could really blow up in our face. They're like, yeah, that's true. But your stock options will be long vested by then. No one will notice. If you think the goal of the company is to enrich its shareholders, it's very easy to be tempted by these kinds of decisions. Of course, a purist, you know, the Steve Jobs type, will never ever make such a trade off. Outrageous to suggest such a thing because we love the craft. It's itself. What most people do not understand is that if you have a belief that your organization is meant to do something specific other than make money, whether that's as humble as you want to make a high quality mayonnaise, or as complicated as you want to solve climate change, you are a business revolutionary, whether you know it or not. Because our dominant financial system only accepts the idea that the purpose of a company is to increase shareholder value. So when push comes to shove, the product team is where these conflicts ultimately get borne out. And those of us who've been in organizations that get a little bit larger know this feeling well. In product is usually an engineering is often where you find the people I call the torchbearers. These are the people who not maybe don't have the fanciest title, maybe don't have all the formal power. They're probably not the highest paid people in the org, but they're the people with the moral authority who really believe in the company's mission. And you know, they're the people you go to when you need help? Like, hey, some product manager is trying to get me to. To make this feature worse for customers to make a little bit of money. Can you help me? And they'll be like, ah, I got you. I got you.
- [18:17]
A
Right?
- [18:17]
B
You know who I'm talking about. Everyone's met these people. If you've ever had the job of being a torchbearer, though, in a modern company, you live your life under siege. Because every day of your life, some person is in your office with a spreadsheet being like, excuse me, but what is the ROI of doing the right thing here? Wouldn't it be a little bit more efficient if we just, you know, like, you can't argue those points because the way that we've constructed most companies today, the ROI of doing the right thing is negative by definition because the costs are tangible, but the returns are intangible. So this is why product and engineering leaders especially are essential for fighting back against this consensus we have to have. Those are the people have to stand up and say, listen, I want to work somewhere where our commitment to quality or whatever the value is is not just skin deep. I want to know that. It's written in the corporate charter. I want to know that even if we get pressured, we're going to have the freedom to do the best work. And if you look at the best companies, the ones that have really been able to last for a long time, the ones that get a consistent talent advantage, they're the ones who have done this hard work upfront.
- [19:22]
A
Talk about forces kind of pulling us away from the right decision. One of the most interesting forces you wrote about is financial. So you write about this concept of financial gravity where success is actually pulling companies away from their mission.
- [19:38]
B
Yeah.
- [19:38]
A
Can you explain how that works and how to combat it?
- [19:40]
B
So financial gravity is this force I have been doing battle with my whole career. You know, I have witnessed it in the companies I've helped start, and I've faced it in the companies I have started myself. Whether I've been a board member or a CEO or cto, whatever my role is, I see it all the time. And of course, the privilege of being the lean startup guy is I get to help lots and lots and lots of people start their companies. People call me all the time for advice. It's maybe the most common phone call I get. I'm thinking about starting a company. I have a company. I want to raise a Series A. We've raised a bunch of money. I want to prepare for an ipo. Those are the kind of calls I get. So I've watched this process as companies lose what made them special. And I eventually started to realize that there was this force. It was like, I call it the force that no one controls, but everyone obeys. That's like seeping into all the decisions. And it's not necessarily conscious. People say they still value quality or they say they want to put the customer first. When push comes to shove, they just feel like, you know, it'll be a little easier to compromise just this one time, and next thing you know, it's death by a thousand paper cuts. Jim Senegal, the founder of Costco, called it literally, like taking heroin. You do it one time, you raise prices a little bit, you get away with it, your stock price goes up, you're like, ooh, I don't know if I want to do that again. But you kind of have to because now that's baked into the forecast. And think about how many startups have had this problem. You want to raise money, you need to show consistent growth. You want to show a hockey stick shaped growth curve up into the right. Well, the compromise you made in one month, you got to make it again the next month. Else you risk, God forbid the numbers go down. Oh, no. So it's a very easy trap to fall into. And in the book, I really work through the psychology of it. Like, why does this force operate? What are its laws? Just like Newton's laws, of course, gravity. We need to know the laws of financial gravity. And one of the most pernicious parts about it is that gravity is about the feeling you have unconscious, that if you comply with the values of those who have more than you, you will get ahead. So it's about your imagination of future transactions. And if you've ever been around a person who's meeting a celebrity or a billionaire for the first time. You ever seen that? It's so gross. They're like all of a sudden. The strongest people I know will like all of a sudden become super obsequious and like, they just desperately want to please that person. It's really gross. Why does it happen? If you ask them about it would say, was that your plan to turn into blubbering jello in front of, you know, whoever? No, I just, it just came over me. It's an unconscious reflex. You can't turn it off any more than you can stop your eyes from dilating in the dark.
- [22:11]
A
I've been building a lot of air products lately. My job search OS has 16 different agents. My newsletter has a recommendation engine, and I kept Running into the same problem. I'd ship something, it would work in my testing, and then I'd get messages from users saying it's hallucinating or picking the wrong tool. The issue wasn't the prompts or the tools. It was that I wasn't actually evaluating anything. I didn't have a way to see what my agent was actually doing. Step by step, every tool call, every decision. That's where Arise comes in. Let me show you. I'm going to open Claude Code and install Arise with just one command. NPX skills. Add Arise. AI Arise Skill. Skill, yes. Now, Claude Code already knows how to instrument my agent. I tell it set up tracing to Arise, and it automatically analyzes my code base, figures out where the LLM and tool calls are, and adds instrumentation automatically. Now I can see everything. Every trace, every span, every decision, and more importantly, I can evaluate it. That's the shift trace what's happening, evaluate where it fails, then fix it. This trace right here, my resume feedback agent was supposed to pull the company's tech stack from the job posting, but instead it hallucinated that they use React when the posting said Python. But instead it hallucinated that they use React when the posting said Python. I never would have caught that without seeing the trace. And here's the part that blew my mind. I asked Claude Code to look at these traces and tell me what I should be evaluating. It came back with four eval criteria I hadn't written. Things like picking the right tool and staying grounded in the input. I wrote the evals, ran them, and found that my agent was making the same kind of mistake about 12% of the time. Claude pushed a fix, I reran the evals and it dropped to under 2%. That whole loop, trace, evaluate, fix took me about 20 minutes. And now it runs automatically. If you're building AI products and not evaluating them, you're shipping blind. Try arise free@arise.com and get a year free a 12 cent $60 value with my bundle Arise. Check it out. It's one of the top AI evals platforms used by all of the top AI teams for a reason. Okay, so this financial gravity is always pulling us. There are these forces always pulling us, especially product managers. I think one of the coolest cost case studies you talked about was Costco.
- [24:30]
B
Yeah.
- [24:30]
A
Where they built this governance fortress. Can you walk us through how Costco has built an incorruptible company?
- [24:37]
B
Yeah. Costco is like the classic exception to almost every rule. When you talk to people about corruption, especially, people will be like, Ah, you know, I was an economist, was giving this lecture the other day and they were talking about how only family run companies can sustain a mission over multiple generations because every other company succumbs to cost cutting and quarterly pressures and all this stuff. And they're listing off the family run companies that have had all this longevity. At the end of their lecture, they were like, oh yeah, also Costco. For some reason, Costco's not family run, but they also seem to be the exception. They can do this thing anyway, moving on. And that's so many times people say like, oh, once a company gets big, of course it's gonna fall to corruption. That's just human nature. And then you're like, but do you like shopping at Costco? Oh yeah, they're the. That hasn't happened to them. But Costco is a 400 billion dollar public company. Surely if it was inevitable, it would have happened to Costco by now, right? So to understand Costco, you have to understand a guy named Saul Price. He's the father of modern retail. If you want to know how influential he is. Walmart is called Walmart because Sam Walton was paying homage to Saul Price's company, fedmart. So when Saul started fedmart, Saul was a lawyer originally. And when he was a lawyer, he learned in his training that you have to have a fiduciary duty to your client, meaning you put the client's interest before yourself. And when he became a retailer creating fedmart in the 1950s, he thought, well, who's my client? Oh, the customer is my client. So just like Costco today, he believed in what he called a fiduciary duty to the customer. That meant that when competitors would try to undercut him on price, he would literally put up signs inside his own store saying, don't buy this product from me. You can get it cheaper down the street. Think how unthinkable that is even now. Like, how many DTC brands would you wish would just do that? Just tell me where the lowest price is. Why don't I have to be the one to find it? As a result, customers loved fedmart. They would travel miles out of their way to shop there. Company was very successful. It went public. And as a public company, he was frustrated by this financial gravity, constantly trying to get him to conform to the retail best practices. So to avoid that, he took the company private. He found really awesome investors. So he thought they became the 51% owner, he became the 49% owner, and he operated as a private company. Once again, this solved exactly zero of his problems. Because the new board was just as captured by financial gravity as the public markets had been. They wanted faster growth. They wanted higher prices. Higher prices and lower wages, not the reverse. But Saul was a completely uncompromising person. He had an ethos, a character. He believed in his business philosophy, and he understood that it was an engine that was powering FedMark's growth. He wouldn't compromise. So what happened? Well, one day in 1975, Sol Price came to work and found the locks on his door had been changed. He could not get into his office because he doesn't work at fedmart anymore. The board had fired him. So the board got what they wanted. They got Saul Price out of their way, and they instantly turned Fedmart into conventional retail practice again. By 1982, within just seven years, they completely destroyed what Sol Price had built over more than 20. The company was in liquidation by 1982. There are no Fedmart stores anymore anywhere in the world. That was a pretty sad story, but very typical. If you've been around this business, you will know lots of stories like this of a founder, a naive founder pushed out of his own company. But Saul, classic founder, he did not give up. He didn't go retire. He took two weeks off. This is my favorite detail in the story. He took two weeks off. And after two weeks, he leased the office upstairs from fedmart and he was back at work. He created a new company called Price Club, which when I was growing up in San Diego, was like a local fixture, even at the point that Fed Mart was a distant memory. Now, Price Club, you have probably never heard of, because one of the people that left fedmart to go to Price Club with Saul was a guy named Jim Senegal. Senegal had started as a stock boy in Fedmar and worked his way up to executive. He was very loyal to Saul Price. But after a few years of working at Price Club, he decided it was time for him to strike out on his own. And he created a new company. And a few years after that, his company and Saul's company merged to form the company you now know as Costco. And here's what makes Costco different from fedmart. It's not that financial gravity likes Costco any better than it liked fedmart. Believe me. Every few years, Wall street makes an attempt to try to damage Costco's governance, try to prevent it from being different. We see it all the time. I have a whole bunch of quotes in the book. One of my favorites is a Wall street analyst. Let's see if I can Find the quote here. The analyst wrote, costco is spending what could have been shareholders profit on making a better experience for customers. Now most product people hear that and they're like, is he complimenting them? No, this is a criticism. Costco is wasting money that it doesn't have to spend on customer experience. It could just be giving that money to the shareholders. And here you really see the essence of how our modern finance driven economy sees value creation. Customers are a resource to be mined for the benefit of investors, not the other way around. So the reason why Costco has endured for 40 years and is one of the best performing stocks in the whole S&P 500, I include this little stat in the book right here. If you had invested $10,000 in the S&P 500 in 1985, you would have $151,000 today. If you'd taken that same money and put it in the Costco IPO, you would have $8.7 million today. Costco is an incredible performer. And this is the paradox or the confusion that we have. People think the way to create shareholder value is through shareholder primacy. But no, Saul Price understood you have to have a fiduciary hierarchy that puts shareholders last, not because shareholders are not important, but because shareholder value is the exhaust from the engine. When you put the exhaust in the intake, when you make finance the goal rather than the result, you get deeply confused. So the reason why this has worked is that Costco was built with this governance fortress. The fortress prevents outsiders from pressuring the company to abandon the ethos that Saul Price created. And that really is the blueprint. In a nutshell, you want to have the ethos of Saul Price and the integrity of Jim Sinegal.
- [30:48]
A
So can I copy paste these governance features, and will I have an incorruptible company?
- [30:52]
B
No, absolutely not. These features make sense for Costco's unique structure, but every company has to have the structure that makes sense for it. And I want to be super clear about this. It's like just because you have a structure that resists pressure doesn't make you incorruptible. Because remember, the. The killer that's about to kill you is already inside your house. Most of the time when companies collapse, the problem is internal. It's an internal misalignment. And founders and leaders are routinely naive about this point. Can I tell you one more story?
- [31:24]
A
Yeah.
- [31:25]
B
All right. So in 1910, a guy named Robert Wood Johnson II went to work at his father's company called Johnson and Johnson. Okay. And he rose to the ranks of the company. And in 1932, in the depths of the Depression, he took control. I think his father died right in the middle of the Depression. Now, Robert Wood Johnson was a real Sol Price type character. An absolute giant of industry, completely uncompromising. He opened factories when other people were closing them. He raised wages during the Depression. He understood that the depression was temporary and that if he strengthened the company during that time, his employer, the employee loyalty, the customer loyalty that that would generate would drive the company to explosive growth. When the depression finally ended, that's exactly what happened. So a decade later, he's preparing for the company's IPO during World War II. Okay, so can you imagine, people think, oh gosh, there's geopolitical uncertainty today. But he had to plan this IPO during World War II. Okay, just really think that through. And he was worried even then that the company might lose its way once it became a public company. So unlike Saul, who I think went into this relatively naive, Robert Wood Johnson had a plan. The plan was to create something called the Johnson and Johnson, our credo. This is a very famous document in business history. You can see it here on the screen. Basically, patients, doctors and nurses first, employees second, communities third, shareholders last. Notice the pattern. So he wanted to make sure that everyone who worked at Johnson and Johnson would embody this credo. So he had it carved into 8 foot high limestone blocks and had them inscribed on the company's walls in the headquarters where they exist to this day. The theory of change was very simple. His idea was that if you want to work at Johnson and Johnson every day on your way to work, you're going to walk by the credo. You're going to see it carved in stone. Doesn't that sound good? Doesn't that seem like that will help remind people what this is all about? But it didn't work. While he was alive, it worked great. He was an absolutely uncompromising person. As I mentioned, he fired his own cousin, he fired his own son for violating the credo. Like he was really strict about this. But he eventually died. And Stone doesn't enforce itself. As the company grew and as the eras went by, the company eventually came to be captured by financial gravity as it came under the sway of our modern best practices, including shareholder primacy. So much so that Starting in the 80s, 90s and early 2000s, it started to be rocked by scandal after scandal after scandal where the company put its own self interest ahead of customers and patients. I think you can pick like which of these Scandals is the worst. For me, the worst scandal is the time that they put asbestos in the baby powder and covered it up.
- [34:10]
A
What?
- [34:11]
B
Yeah, yeah. In baby powder of all things. It's kind of shocking that the company that had this credo would do this terrible thing. We know that they covered it up, by the way, because eventually these things always come out in litigation and then the documents get made public. So they knew and they've had to pay unbelievable settlements. It'll most likely be like a $10 billion settlement by the time this thing is done because they actually gave a bunch of people cancer. And you ask yourself, how is it possible that at Johnson and Johnson of all companies, they could do this? The people who put asbestos in the baby powder walked by the credo on their way into work every single day. Yet that wasn't enough. So a good corporate structure can protect a pre existing ethos, but it doesn't cause it to come into being. We gotta have both for this to work.
- [35:02]
A
Mmm. And I guess the Enron example is another famous one people can probably relate to. They had the missions printed at their headquarters. It didn't make a difference. So we can't print out a mission statement. One thing you did identify for folks was a mission lock. And I really like this. I think this is going to be on page 261 for you.
- [35:26]
B
Here's the Enron slide, by the way. Oh yeah, this is from the, this is from the Netflix culture code deck, which is just, it's just awesome. Yes. Enron's values were integrity, communication, respect and excellence. And I, this, I didn't know this until I did the research for this book. Enron's board was named one of the five best boards in America right before it blew up. Had a lot of independent directors followed all the best practices.
- [35:48]
A
So you can't just print them out like Enron. You can't just carve them into stone like Johnson and Johnson. But you can do something you talk about which is a mission lock vehicle. What is that?
- [35:59]
B
So before we get into the details of this, I want to warn your listeners that it's going to sound weird, okay? Because there are a lot more kinds of organizations out there in the world than most people have ever been trained to in or taught to encounter. Even if you've gone to business school, you may never have heard of this structure. Certainly most VCs have never heard about it. Most founders are never told that there's any option other than the standard Delaware C Corp. Venture backed company like that. We live in a Very much a corporate monoculture. But just because this idea is new to you, doesn't make it new. In fact, the German optics company Zeiss, the ones that made like my and probably everybody else's lenses who are watching this right now, they had this structure in 1885.
- [36:40]
A
Wow.
- [36:41]
B
So it's actually a very old idea. The idea is that just like in political science, we understand that having only one decision making body, you know, Right. Like one branch of government is actually intrinsically unstable. You can have checks and balances between multiple branches of government and create stability even when people disagree. So the Mission Lock vehicle is the idea that having a separate entity, that is a set of trustees or stewards who oversee the performance, they hold accountable the board of directors of the for profit company. This structure is more common than you realize. If you've ever eaten a Hershey chocolate bar or shopped at Ikea or you've been to Patagonia, if you've shopped at John Lewis Partnership, if you've ever been to the uk, if you have a Vanguard mutual fund, if you have taken a Novo Nordisk medication, you have been a customer of this structure. And in fact this structure is so common that we have enough data to be able to answer questions. There's a lot of rigorous academic research on these structures and it turns out they perform better than you'd expect compared to conventional structures. So for example, the Novo Nordisk Zeiss called industrial foundation structure, where the Mission Lock vehicle is an actual non profit foundation, companies with that structure are six times more likely to live to year 50 compared to conventional companies. We're talking about 10% versus 60%. So it's not an inconsiderable difference. And they also overperform in almost every financial metric you can think of because they just are capable of sustaining long term plans and making long term investments in a way that traditional normal companies are not able to do.
- [38:20]
A
So should everybody consider doing this? Are there people who shouldn't adopt this?
- [38:24]
B
I think it's kind of a no brainer. And in fact I'll tell you one of my favorite stories about the development of this structure. It goes back to 1920, when a woman named Marie Krogh was living in Denmark. She was one of the first Danish doctors, actually very interesting woman in her own right. But she contracted diabetes, which at that time had no known cure, so it was a fatal disease. But her husband August had just received the Nobel Prize. And so he asked her, in spite of her illness, would she accompany him on a lecture tour of North America. She said yes. And they came to North America to travel around and meet other scientists and have his lectures. Now, at one of those lectures, she was sitting next to another scientist at the dinner table and he told her about a new breakthrough that had happened in Canada. He said that scientists there had managed to finally isolate insulin, which could potentially be a cure for diabetes. So she convinces August that the two of them should extend the trip, travel to Canada and see this research breakthrough for themselves. They do. They see it, they realize its immense potential immediately. And they ask the Canadians if they can have a commercial license to bring this cure back to Denmark with them. Not just to save Marie's life, but to save thousands and millions of others. The Canadians agree. But everybody involved is very uncomfortable with the plan because. Because decades before Martin Shkreli, they all had this instinctive fear. They said, wait a minute, if we are a for profit company and we make a life saving medicine, it's no problem to charge a fair price for that. Of course we have to pay for the laboratory, after all. But what if our future selves, our future progenitors, our descendants in this company, what if they feel the temptation to say, wait a minute, if I have your life saving medicine, couldn't I charge you whatever I want? Won't you have to pay? See, they saw science as a public trust. And they, just like that product manager we were talking about, where quality is number one to them, scientific integrity was number one. And they wanted to build a structure that would embody that principle for all time. So when they went home to Denmark, they created the Nordisk Insulin Laboratorium. They created it as a for profit company with a non profit parent company. And that foundation has guarded, has acted as the mission guardian or the mission lock vehicle for more than 100 years. The company of course is not called Nordisk. Today it's called Novo Nordisk. It is one of the largest companies in Europe and for, for a time it had a larger market cap than the GDP of Denmark. It is a behemoth of a company and this structure has worked incredibly well. Just like the the governance fortress at Costco. When people have come at Novo Nordis to try to get it to do the wrong thing, this structure has resisted. But it has an extra feature. Not only has it helped them resist external pressure, it has also helped them resist internal betrayals. So in the book I tell the story and this is going to sound like a ludicrous exaggeration, but read the story and you'll find out that this is 100% the truth. There's actually a great acquired podcast episode about it. If you don't want to read the book, you can just go watch the podcast episode. That's fine too. There was a time in the, I think it was in the early 2000s when the board, the trustees of the nonprofit foundation, had to make a decision to stop the for profit subsidiary from doing something really stupid. And they did. It was really difficult. It cost everybody a ton of money and it pissed everybody off. But as a result of their foresight, the nonprofit trustees created for the shareholders of Novo Nordisk more than $500 billion of shareholder value. Because just having the long term perspective allowed them to protect the research program that ultimately enabled them to invent GLP1. So the next time someone tells you that you have to do the conventional structure. I talked to so many founders and they're like, well, but my lawyer said, my investors say, I talked to a banker. I talked to this person. I'm like, I'm sure whoever you talk to is very smart, but I just want you to ask yourself one simple question before you agree to what they say. Are you sure that they're smarter than a Nobel laureate? Because August and Marie Crow worked this out 100 years ago, and it worked. And I feel like when I learned these stories, I was frankly pissed. I was like, wait a minute, this is like my birthright as an entrepreneur. How come no one ever told me these stories? How come I never know these? Like, how come this, this history has been erased from business education? It's been erased from entrepreneurial education. And my goal of writing this new book is to bring it back and to say, you know what? These structures work, we should use them.
- [42:46]
A
Yeah. I feel like everybody these days, they just pick up 0 to 1 from Peter Thiel. He says, Delaware C Corp. They never think about these structures. So let's talk a little bit more about structures. Okay, guys, so we have walked you through how to create an incorruptible company in like 35 minutes. We've given you a little bit of a crash course. Eric is a man of many talents, and when I was researching this episode, I learned that he's not just the lean startup guy. Actually, he's co founded Answer AI with Jeremy Howard. With nine people, no managers. You also built Solve it, an AI powered platform you use for everything. You're living in cloud code and shipping things every day. Maybe. Can you start with what does a typical workday look like inside your tooling stack? How are you using the different tools and Putting them together.
- [43:38]
B
Oh, typical day. Are you kidding me? My life is so complicated and that's just like. Yeah, it's just such. It's so different. Like now I'm doing this with you, but like, if you'd asked me the question six months ago, I would have been heads down writing and had absolutely no time for anything. So, you know, it's different. It's different every day. And like I said before, the true privilege of my life is that people like Jeremy or Luke at Virgil or so many people over the years have entrusted me to be on the entrepreneurial journey with them. So I've been a founder, I've been a CEO, but I've been a co founder too. I kind of like that. I like both jobs. They're both really interesting and I learned a ton of. So yeah, I like to be behind the scenes. I like to be on the board. I like to be having entrepreneurs back. That's kind of how I conceive my job. When I'm helping someone else build their dream, their vision. My job is to make the structure that supports them, like I did for Dario, incorruptible. That's the goal. So, yeah, Answer AI is someplace I spend a fair bit of time. You know, I also put the long term stock exchange. So I've got a lot of things going on. And depending on what's up for which company, you know, that determines what I'm doing that day. But like, if you want, I can share a little bit about what it was like to use Solve It. Solve it as the platform we make it Answer AI. And just to maybe give a little bit of background on Answer AI. Jeremy Howard is like a machine learning legend for many of the people who are listening who got their, like, people who got their start in machine learning as a programmer. Like, it's not uncommon that they learned deep learning from Jeremy on one of his fast AI courses. Okay. He's like for many years been my friend and the person I call when I have a machine learning question, like need to understand something, I call him. And when he was ready to take his research and turn it into a research lab, he was concerned. He was like, I don't want to do it if I'm going to wind up like everybody else. So actually like the. Our ability to wrap him in an incorruptible structure unlocked the investment opportunity for his VCs that made it possible for this company to exist. It wouldn't have existed otherwise because Jeremy's a person of tremendous integrity. And if he Couldn't do it in a way that protected his principles. I don't think he would have done it at all. That's partly why I admire him so much. So, yeah. So one of the products we make at Answer as called Solve It. And it is a way for us to embody our philosophy about AI, which is that AI should never be used to replace people. It should not be used as a replacement for human creativity. It should be used to augment human creativity. That's the idea. So, for example, when I first started writing this book, there was no generative AI. I've been working on this for a long time. But over the whole time I've been writing and researching this book. While I had a human research team and human editors, I also benefited from GPT3 and then GPT4, and then Opus and Sonnet. And all these different models have gotten better over the course of time. And when they first started getting good, I was like, this is going to be great. I was excited to use them for writing, but they really did not work very well for writing. They still really don't. Because everything you write winds up sounding like garbage. Makes it sound like AI. Like AI. Just like private equity has a flavor, AI has a sound. It makes a certain sound. If you understand how LLMs work, you'll understand why it actually makes sense. It's confined conforming you to the training distribution. It is actually the way that most of these chatbots are organized. They are memetic conformity machines. So that's actually one of their problems. So when people are so pumped to use AI to replace humans, I think they're making a huge mistake. First of all, it's like profoundly antisocial to be gleeful about such a thing. But more importantly, they're sanding all the rough edges out of their intelligence. And it's funny, like when someone uses the phrase AI, by the way, they are literally taking sides in an age old debate about the nature of intelligence. And if you really think a generative LLM is quote unquote intelligence, you're siding with Jacques Derrida and Foucault. Okay, like you're siding with the postmodernists who claim that intelligence is just language. If you understand LLMs, they're just language. So like, you really have to understand what they're, what they're good for. So I couldn't really use AI effectively in my writing or editing process until we invented Solve It. And that was the major unlock for me. It made this book so much better because it allowed Human in the loop, AI interactions where we can blend what the human is doing and what the AI is doing in such a way that we have true shared context between both parties. So you can see everything that AI can see and it can see everything that you can see. And not only can you correct it by giving it chat commands, you can just edit its responses. So if it does something wrong, you can just be like, oh, let me just replace what it said with what it should have said. And because at the end of the day, LLMs are autoregressive, the more wrong answers there are in a chat history, even if you tell it immediately afterwards, no, don't do it like that. The more it learns the pattern of wrong answer. No, don't do it like that. Right answer. So then it starts to mimic its past behavior. And not to mention all the problems with sycophancy and hallucinations and all the other things. So having. Having the end user have a lot more capability and control over what's happening helps a lot. It also allows us to focus. And you can do this in any tool I use, cloud code and other tools too. Whatever tool I use, I try to really focus on capabilities, not artifacts. This is the Lean Startup 101 for those that are paying attention, right? Like the. The unit of progress in a highly uncertain situation is what's learned, not what is made. So as long as you're having the system teach you how to make an artifact, it's okay. Then your capabilities increase, your sense of agency increases, you learn how to use it better. But as soon as you're just like, just make me an artifact and then tell me when it's done, or, hey, like, I caught myself in some of the codes when I've done vibe coding. It's a bit like a slot machine. You know, you give it a design for something and then it's like, do you want me to suggest improvements to the design? Yeah, sure. You want me to make some adjustments? Yeah, why not? Hey, you know what would be better is if we kind of went over here. Next thing you know, the product it built for you is not really what you had in mind. It was just the most convenient thing where you're like, yeah, sure, why not? Why not?
- [49:25]
A
Why not?
- [49:26]
B
You kind of get into this soporific state. I think it's going to look a lot like people who are playing slot machines look when we put them in the MRI machine. It's like you're not really in control of what happens. And now going back to what LLMs really are. As a result, you wind up making mediocre stuff. What does mediocre mean? People think mediocre means, like, lacking in quality, but it really just means in the central part of the distribution of outcomes. So you're making the same stuff as everybody else. If that's what you want, go for it. There's a lot of things where you don't care. It doesn't matter. You just need something very basic. You don't need originality, no problem. Use it for that, no problem. But the second you want to do something that will give you a competitive advantage, that will have some level of artistic integrity that will be reflecting who you are as a person. You have to be very careful how you use these tools.
- [50:13]
A
So how to Solve It. Maybe you can walk us through how you wrote some specific book chapters with your actual artistic integrity and voice.
- [50:21]
B
Let me see what I can pull up for you. I'll show you. This is Solve It. This is my Solve it instance where I have just tons and tons and tons of these dialogues going back months and months and months and months. You can see how long I've been working on this book. Months and months and months. They go back and back and back and back and back and back and back and every single one. The way I think about Solve it is it's like a canvas. I mount my work onto the canvas, some part of my work, and then I work on it myself. Solve it's job is to help me, the human being, do good work. Not to tell me what to do, but it's extremely helpful for keeping track of things, for researching things. See here. Here's just a copy editing pass. You know, let's see what I have here. May have to click on a few of these before I get something that's actually interesting because copy editing can be boring. So this is what Solve it looks like. There are three kinds of messages in Solve It. These green ones are called notes. This is just raw markdown. The orange ones here, the kind of reddish ones, are prompts. This is where I asked the LLM to do something. And then blue are code so I can write Python code natively. So we, we mix and match any of these things. And when you're working in a certain message, the LLM only sees the messages that come before it. So here you can say I'm giving it some context. Here's a. Here's its kind of overall prompt here. We developed a Eric Reese writing style guide so, you know, understand like the kinds of things I'm Trying to do it has a list of things to avoid. This is all the kind of AIE stuff that is really toxic. So I like, even when it's talking to me, I want it to know, please don't use that stuff. It drives me crazy. Special rule about EM Dash is this dialogue's from a while ago. So you can see that was when I was still an issue. And here is even a briefing about what is this book about. This dialogue's from late in the process where the book already has a really defined structure and defined outline. You can see all the chapters. Just a chance for it to get a sense of what are we accomplishing together. Here's a bunch of Python code for stuff we need to pull in various pieces of the book. And I'll skip over this stuff. Code, code, code, bunch of code. So, like, one way to think about this is like, it's my own custom personal rig where I have combined like I have written some of the software myself. I've used the LM to help me write some of that software. I've used my teammates at answer. I'd help me write some of it so that everything is highly customized for what I want to do. So here, yeah, like we're. This is, this is. We've got a to do list. We've got reader comments from test readers. So here's like, this enables new kinds of writing that was never possible before. So this is. Maybe I'll just focus on the reader comments part of this because that's the most interesting part of this dialogue. I had more than 600 people be test readers of this book. I'm a big believer in feedback, haven't you heard? And we use this product called Help this Book, which is like an online reading platform where people can leave structured comments as they go through. So I had a CSV with more than 10,000 reader comments in it from all these 600 readers. Now that's two I read as they came in. I read every single comment, like every day. My daily ritual for months was I started my day with what are the comments that came in overnight?
- [53:32]
A
Okay, what?
- [53:32]
B
I want to read every comment. And I would really take it really seriously. But at a certain point it gets overwhelming. Like, oh my God, I'm trying to work on this one paragraph. How many comments have I gotten about that paragraph? I don't know. I can't even remember right. The context is more than my wetwear can understand. So here I wrote some software to help me look at the comments. Let's see if I can find some comments, comments, comments, comments. This is all instructions about how it likes to work. Here we go. Okay, this is mapping chapter IDs. Oh, here we go.
- [54:04]
A
Yes.
- [54:04]
B
So we're going to be working today on chapter four. Great. That's the sixth chapter in the comment structure. Because of interstitials and stuff, there are. At that time I did this particular Dialogue, there were 9,546 comments total, of which 470 of them are from recent versions that relate to this chapter. So even that, that's too many comments. I can't be like, okay, how do I incorporate this number of comments? But I can have it summarized for me. Okay, and here you can see the tool calls that it's calling in Python to understand. It's like, great. Here's what people like about this chapter. Here are the things they especially like. Here are problems that they have flagged. Here are some suggestions. So it's like, okay, great, now we can get to work. Here's the actual chapter itself from my copy editor. And now we start going through all the different things that we're going to work on. This dialogue is a little boring because we're working on copy edit. This is a lot of technical stuff in the writing. Let me see if I can find something that's a little more substantive. Just give you some sense of the variety of it and stop me if this is getting boring.
- [55:08]
A
No, this is really interesting. So you've basically created like a agentic harness within. Solve it with various code blocks, various context blocks. A lot of people build these out as like cloud code operating systems. You have build this out and solve it in a visual way.
- [55:24]
B
Yeah. Okay, so this is a much older. This is a much older version of it from when I. Oh God, this is really old from when I was still working in Google Docs. So this is like, this is maybe from a year ago, but you can still see, like, here is a summary of what the chapter was at that time, what's good about it. Okay, here's the. Here are all the different sections. Some of these sections are still in the book, but a bunch of this. Oh, unfortunately master using it and then you can have. This is gone from the manuscript. I like this one a lot. So, you know, yeah, where we're going, we don't need roads, I think was also cut. So, you know, not every, not every meme and pop culture reference made it into the final manuscript script. So here is a typical. This was like a typical day for me a year ago was like, okay, Great, here are 27 things I need to do for chapter five. And just being able to write them all down is already such a relief. Like, okay to do list. And like my editor and anyone I work with can like add to do's to this list is like, here's something that needs to happen. Read this Costco story and make sure it actually makes sense. And we made this, like, make sense. So anyway, so. And we would just go through one thing at a time. You know, here's the repetitive language that we were using in this story. Here's what we need to work on. Now let me see if I can show you one more thing. This is a lot of going back and forth, back and forth here. We're seeing like, okay, here's the diffs. This is more cumbersome version than what I have now. So it's a little clunky. But what you're not seeing here, the one thing that is not present is it looks like the, the AI is doing all this writing. Because I'm just saying, next, next, next, next. But what you can't see is very frequently if it generated text that I didn't like, instead of saying, no, that's not right, I could just pop in here and I can edit this markdown directly and just be like, actually, this is what you should have said. Even what I would frequently do, these are all been deleted. But usually what I would do if I was getting into a kind of a difficult section, this stuff is all practical, easy that we have here. See if I can find a more difficult one. Yeah, here we go. So like, here's a more complicated paragraph where we're trying to figure out, like, how do we adjust? So here's all these like, oh, this is good. This is actually good to show you. So here's the cultural significance of the Costco $50 hot dog research report. So this is, this is actually a deep research report pulled from another LLM, probably from OpenAI. Deep research, if I had to guess, which is just all the links and research we have on this topic. So just all the things you need to know about this topic. Okay, great. So now, given this research, what could we consider doing in the manuscript itself? So again, this is like not doing the work for me. It's helping me keep track of all the work that I've already done. Sometimes I would pull in deep research. Sometimes this would be research done by a human person. Sometimes it would just be my private notes. I'd accumulated thousands of pages of notes in the course of researching this book. So then we would say, okay, let's make a revision incorporating this plan. And it would give me a new paragraph here. Here you can see we're turning it into diffs, and then we're actually adding it to the manuscript. But what you can't see is in almost every one of these instances, if I wanted to, I would edit it directly. Or I might say, okay, I actually don't like that. Let's figure out what to do. And we would go back and forth, back and forth, analyzing, analyzing. And the most common thing I would do is I would write something and then ask it to analyze it for me. Is this any good? What's good about this? What's not good about it? You can't just believe everything it says. You have to be very careful with this feedback because it can ruin your authorial voice. Believe me, it took me a while to get the hang of it, but once you figure it out, it's just so nice. Like, writing is the loneliest, saddest activity in the world because it's just you and the page. And it's just like, it's very nice every once in a while, even to get sycophantic reviews of your work, you know, like, even if you know at some level that, like, it's kind of telling you what you want to hear, it can help you keep going. It's very nice. So anyway, so we would go back and forth. When I would figure out what I liked, I would just put it back in here as if it was the LLMs first suggestion. And so what you'll see, if you read these dialogues carefully, the LLM is actually getting smarter as we go as it figures out what good looks like to me. Because it's not all the kind of context bloat you see with something like Claude code is removed. All that code is. Everything has been deleted from context. Only the information that it really needs is present. So anyway, that's solve it in a nutshell. And yeah, I use it for writing and research, but of course, the team mostly uses it to write software. So it also can do all the things pretty much that Claude code can do.
- [59:49]
A
Okay, guys, I don't think anywhere else you're gonna be able to find out how he actually wrote the book. There you get to see the actual chats of it. Fascinating. So I want to bring this back to our discussion in the first part of the episode, which is the structure. Yeah, because a lot of PMs, I'd say probably like 50, 60% of them, they eventually graduate to founder. So what were the choices you and Jeremy made? I believe you actually have answer AI as a for profit. So how did you structure it properly so that it's incorruptible? What were the specific decisions you made?
- [60:20]
B
Sure, yeah. So answer AI has certain advantages. First of all, Jeremy is a legend, so that's always nice. And therefore investors were very keen to invest. So we were able to really partner with them and be thoughtful. We sat down with our investors and it wasn't just like, oh, please give me money, give me a term sheet. No, we had a real conversation about what is the right long term structure to protect the integrity of this research. Now, we didn't do the works because at the end of the day, answer I was a couple years old. But it's a relatively young company. We've raised only a Series A and that's it. So it is a public benefit Corp. So yeah. So answer AI the first thing we did was we converted to what's called a public Benefit Corp, or pbc. There's a lot of confusion out there about this because it sounds very similar to the B labs or B Corp certification. People talk about being a B corp. That's the little B with a circle in it you've seen at the farmer's market. That is not the same thing. That's a certification and that's good. I have a whole chapter about that. You can read chapter 13 if you want about certifications. But PBC is something much more fundamental. PBC is the opportunity to write your corporate mission directly into your corporate charter. And that gives you and your directors on your board permission to pursue that public benefit even if it conflicts with shareholder value in any given time period. So it's a very important tool and it requires you to be serious about what it requires. It allows you to fix this very serious problem we see in so many companies. Like for example, remember Silicon Valley Bank?
- [61:48]
A
Of course.
- [61:48]
B
Remember when it imploded? Yeah, that was a very stressful weekend for me too. I had at the time you counted up all my companies. I had probably like, I don't know how many, maybe more than $100 million of various companies invested at Sol Devali bank on the day that that thing went south. So yeah, it was very stressful. And if you. In the aftermath, I remember reading about why the company had, why the bank had blown up. And it had a lot to do with this like very arcane interest rate maneuver that they had done. And what's so interesting is that five Years before, the bank had lobbied Congress. They had actually gone to Washington to lobby Congress for permission to do this thing so that they could make more money actually would have been illegal. And they got the laws changed that allowed them to shoot themselves in the foot so that then they could collapse. And I was like, why did they do that? Like, it's actually like kind of crazy how like they had such a good thing going. Talk about killing the golden goose. They had such a good thing going. They were a beloved institution. They had such trust with our customers, you know, so much so that like many of us were not properly diversified because like we couldn't even imagine. Surely Silicon Valley bank would take care of us. Right? And I remember reading at the time their mission. Let's see if I can find this in the book. Actually, since I have it up. I love this format where we get to really look at the, look at the screen share and stuff like that.
- [63:07]
A
Yay.
- [63:09]
B
Okay, yeah, here's. Here was the official mission statement of Silicon Valley bank to help innovative companies move bold ideas forward. Doesn't that sound cool? I love that mission statement. To help innovative companies move bold ideas forward. But that was not their legal purpose. That was just the mission statement. Their actual. Let's see if I can find it. Here it is. Because SVB is a bank, their documents are all super easy to find in public. Whereas for most Delaware C Corp, you have to ask a lawyer to pull it for. But you can pull it. If you want to read anybody's corporate charter, your lawyer can pull it for you. Oh, not a secret. Yep. Anyway, since they were a Delaware corp, their actual charter said as the purpose of the corporation is to engage in any lawful act or activity for which corporations may now or hereafter be organized under the general corporation law of Delaware. You're like, what? What about moving bold ideas forward? Where'd that go? No, Any lawful act. Now unfortunately people hear that, they're like, that doesn't sound so bad. Any lawful act seems pretty open ended. Wrong. Big X wrong. Today most governance experts interpret this phrase, any lawful act or activity as meaning maximize shareholder value. So at the time that SVB was doing this lobbying, it would have been seen as good governance. They're doing everything they can to maximize shareholder returns. Even though the bets that they made, the risky interest rate bets that they made, they didn't help the mission at all. They were a purely self interested cash grab and as a result they killed the golden goose. So the Public Benefit Corp. Is a way to fix that problem. So the Other thing about the PPC that I want people to know is that it is not an. Although it's a relatively new idea in America like that, you should have something called a public benefit court. The idea that corporations should exist to pursue a public benefit is not a new idea at all. In fact, for the vast majority of the time there have been joint stock corporations. This was the default. It was only allowed to pursue a specific purpose. In the 19th century, if you tried to take over someone else's company, first of all, they were allowed to resist you. Not like today's boards which have to roll over and play dead. They were allowed to resist you vehemently. But if you succeeded in taking it over, imagine you took over a company that was designed, the public charter was to make a railroad and you said, you know what? I'm going to change its corporate purpose from make a railroad to enrich shareholders. That would have been a crime. The courts would void your charter as beyond its authority. That is the natural state of corporations. The modern idea that corporations are just financial instruments, instruments to benefit shareholders is younger than most of the trees in your local park. So act accordingly. So that's an important part of the answer AI Governance Fortress. But another element that is really important, especially in the early days, is you have to have somebody act as the mission guardian. That's what I call it in the book. The mission guardian is who has the responsibility to make sure that no matter how much pressure is brought to bear, the company can stay according to its purpose, stay on mission. Let me see. Even the LLM is having trouble because there's so many Public Benefit Corp charters in my inbox. It's like, do you mean this one? Do you mean this one? No, I don't want that. None of those, please. Because the Google search API for it, just like it, only it's only as good as the Gmail search API. It's really frustrating. Somebody please solve this problem. Okay, so the mission guardian. So in this case it was easy. Jeremy and I are the mission guardians. We're the founders. He is the mission majority founder. So he really is the mission guardian and he we vested voting and board control in him because at the end of the day, an AI research lab reflects the research agenda and vision and priority of the founder. Duh. So if you can't trust him, then you have no business investing in this company or even working in this company. But when we are trying to recruit people, we have several people who've worked for the company who have said they're like I wouldn't have taken any other job but for the chance to do this. I'll take a job, I'll come work here. But in order to recruit such high caliber talent, we had to make them a promise. They had to be like, how can I be sure this technology will be used in an appropriate way? How do I know that this vision will in fact be the reality of our day to day life? And we could say instead of having to be like trust the nebulous future investors who may or may not invest in this company, we could just say trust Jeremy. If you trust Jeremy, you're going to be alright. And that was enough for most people. Now I hope as the company grows we will expand to what we call constitutional governance and to things like the mission lock vehicle. But you know, we're a young company so it hasn't really been necessary to do that. I would expect by the time we raise more money then we will have implemented those features.
- [67:53]
A
Okay, so you don't always have to do everything from the get go, which I think might scare some people away. And I think that was a detail you included in the anthropic timeline too which was like they did some major stuff in. Did you say there's series C?
- [68:05]
B
Yeah, yeah, yeah. This is really important. I will read it to you. So Here is answer AI's defined public benefit. The specific public benefit to be promoted by the corporation. This is, I'm reading to you now from our legal charter filed in Delaware is to develop and increase society's understanding and awareness of the benefits and risks of artificial intelligence. I love how they put AI in quotation mark parentheses by researching, teaching and discussing the proper uses and benefits of AI as well as developing useful and practical applications of AI that could be ethical and beneficial to see society. This may include participating in discussions concerning AI as well as its various applications and uses, teaching and or developing AI technology and or applications, as well as engaging in foundational research related to all aspects of AI. In addition, given the rapid development of AI technology, a further benefit to be promoted by the corporation is to do such work as the corporation hopes would be effective for helping develop ethical and benefit beneficial uses of AI for society. Can you tell that an engineer helped draft this? It's very comprehensive. Most public benefit corp statements are much shorter and much loftier. But Jeremy and I were very worried that this would be misinterpreted. So we tried to be extremely exhaustive to say look, this is what we're about. Everyone who invests in the company, you need to know this is what we're about. And all you have to do to invest in this company and believe it's a good investment, you just have to understand that if we accomplish this, do you think we're going to be worth something? It's like, yeah, if you could accomplish this, if you actually do the foundational research that establishes how AI could be beneficial to society, like, that's probably going to be probably pretty valuable. And that was the same bet people made, you know, made with Dario. Like, you don't have to necessarily know exactly how the money is going to be made. Nobody could have anticipated Claude Code in particular, but they were like, look, this is foundational research. It's in the right area, it's being done by the right team. That's a pretty good bet.
- [69:49]
A
So that brings it full circle for everybody. We've covered governance. You've seen now how Eric is building with governance, sequencing it over time. I have a couple more questions because I have the Lean Startup guy here with me. Build, Measure Learn. I think it is the most famous business framework probably ever now. You created it, it's informed everything. We've had guests like Dan Olson on the podcast who created the Lean Product playbook entirely on top of that idea. I'm curious if Build, Measure learn breaks for AI, and specifically foundational AI, we're talking about like $10 billion training run takes like two years. Does it apply to all cases?
- [70:35]
B
So in order to answer that question, you have to understand that Build Measure Learn is not a set of tactics. Like, the reason people have been able to build so many consulting frameworks and product frameworks and books and all. Like, there's like a whole series of people who've made their careers on top of it is because it is a general purpose statement of principle which says that the value in an entrepreneurial situation, a situation of high uncertainty, the value is in the learning, the reflection on each idea. So whoever has the faster velocity through that loop has the advantage. So yes, that's true, whether the thing takes a month or a year or 10 years. I've used Build, Measure Learn on combined cycle power plants that take decades to build and construct and it cost like a billion dollars. I've used it on cancer therapies and all kinds of super long term projects. Heck, I built the long term stock exchange. It's right there in the name of the company. It took me like 10 years to get it off the ground. So what matters is not how fast you go relative to dating apps or whatever you use as your Fast cycle time, you know, imagination. It's just, can you go faster than the convention in that industry? So if you look at, you know, the AI foundation model wars, the labs that are able to release more quickly, both on the product side and on the research side, have a decisive advantage than the ones who have, like, put out one supermodel and then go dark. Have you noticed that?
- [71:59]
A
Yeah.
- [72:00]
B
Like, they're the ones who customers come to rely on. And it's not just that you learn more. That's awesome. It's also a matter of trustworthiness. If a company's on fire, if they're cooking like Anthropic's been cooking lately, you start to feel like, well, I should build my app on top of Anthropic because my app will just magically get better all the time. And so now you're starting to create these network effects that are really powerful. We see that whenever we see a platform war people love and the pressure to standardize on the leading vendor is super high. And the fact that Claude Code is emerging as that standard, I think it's one of their huge competitive advantages, mostly being powered by the fact that their product organization has unparalleled velocity. So, yeah, I think it's build, measure, learn a.
- [72:47]
A
So this is the part of the podcast where I'm trying to get you in trouble, you know, so is Sam Altman just a bad CEO or is he trapped in a bad structure?
- [72:56]
B
You know, it's funny, like, everyone always wants me to comment on Sam and, like, Sam has been nothing but lovely to me. So, like, I have no beef with Sam. I think, you know, from my interactions with him, he's done nothing but be high integrity and kind to me and a really, like, really nice guy, really supportive person. I've read all the same stuff you've read, so I don't know. I honestly don't know. I think that the focus on Sam the person obscures the structural problems that need to get fixed, not just for OpenAI, but for all of these companies. I feel like AI governance is one of those crazy, critically important things that's probably as important as getting the technology right if we want to see this technology be used in beneficial ways down the road. So, yeah, I would rather the focus be rather than on Sam v elon, which I know drives a lot of clicks. I get it. But I would much rather the focus be on what is the appropriate governance structure, who should have insight, who should have control over these models. And I personally am a believer that we should have the equivalent of a sovereign wealth fund like the Alaska Permanent Fund that oversees oil wealth in Alaska or the Norway Fund. Like that is a common structure that you use industry wide when you have a natural resource explosion like this. I would love to see in exchange for all these regulations that the AI companies want waived for them, including, by the way, like, training on all copyrighted material without compensating the authors. Hey, I'm an author. I'm in the training set. You know, like I always tell people it's if you think being in the training set is bad, you should try being excluded from the training set. Okay, that's even worse. So, like, everyone's in a no win situation right now. It'd be so much smarter if we had what's called compulsory licensing, meaning we just established the rule that they're allowed to train on everything. But some of the profits that get made from that have to be contributed back to the people who contributed all that cultural worth. And what's so crazy about LLMs is like, yes, they were trained in all the greatest works of literature that have ever been written, but there's also been traded on every mean tweet, every Reddit post, every, like, random scrap of cultural output is in there. So like, to me, this is like humanity itself created this technology as a collective act. The people who did the matrix multiplication of course played an important role. I didn't mean to say they did nothing, but like, the raw material required the silicon chips, it required the RLHF people working in Africa and other low wage countries and it required all this cultural output. To me, it's the most obvious thing in the world that we should have a shared governance structure that includes broadly shared prosperity. Justin Rosenstein, the former co founder of Asana, has an op ed the other day in Fortune magazine about democratic councils that could oversee. Like, there's a bunch of proposals floating around. I would wish, I wish that was getting like one tenth the social media and TV time that Sam V elon was getting personally.
- [75:39]
A
Oh man, I literally have a list of like 90 more questions I want to ask you, but we'll have to save that for the next episode.
- [75:46]
B
I look forward to it.
- [75:47]
A
Eric, there are so many things you have going on, but now is the time to let everybody know about them. What are all the things you have going on? Where should they go to find them?
- [75:56]
B
Everything. How much time do you have? Good Lord. The most important thing is to go to Incorruptible Co and buy this book. Okay? And I know people hear about A book. And it's like, okay, look, here's the book. But first of all, it's very sweet cover, designed by my longtime collaborator, Marcus Garrett Gosling. You can see it's shiny, it's nice. He did a really nice job. This book is beautiful. The reason you want to get a copy is not just because you're going to really enjoy it, because logically speaking, I get it. People like, well, I'm just going to wait and see if the book becomes a thing. And then if it becomes a thing, I'll get it then, right? If everyone's talking about it, then, okay, then I'll have to read it. Okay, whatever. But if you do that, if everybody does that, then it doesn't become a thing. So if you actually like an author, it's not just about me. If you have anyone in your life who's an author and you actually like them, you need to pre order their book. You need to get it right away when it first comes out. So if you want to do me a personal favor, you really want to make my day and the day of someone who matters in your community, rather than just going on Amazon and clicking on the book, which you can do, you go to Incorruptible Co. We have lists of all the independent bookstores all across this country that are carrying the book. Or even better, don't do it online. Walk into the store or call them, tell them you want 10 copies of Incorruptible on launch day, May 26th. You will make their day and you will make my day. So, yes, please pre order the book. Of course, if you don't want it in hardcover, you can get it in ebook, you can get it in audiobook. I am the narrator of the audiobook, for Better or for Worse. So that has a bunch of cool bonus content and there's all kinds of cool bonuses on the website for your readers. I think there's two things they might be really interested in. One bonus is we have implementation guides. So if you've heard me talk about some of this stuff, like Public Benefit Corp and Mission Lock via going to, like, how do I actually do it? Like, down to the level of what are the docs and how does it work? We have those guides available for those who pre order and we also have reader's guides. So if you're like, hey, I'm not a founder yet, or I'm thinking maybe being a founder one day, or I'm a board member, I'm an employee, I'm a consumer, I want to think through these ideas from the perspective of my unique role. I've partnered with a whole bunch of people who have those jobs to help develop reader's guides that will help you kind of master the material in a way that's relevant for your career. So that's incorruptible. That's definitely job number one. If you want to take your company public and not lose its soul, you can learn more about ltse.com that's long term Stock Exchange. If you're an early stage company and you want legal representation that won't charge you by the hour, that will work on the kind of things we're talking about, but it will also handle all your other really boring administrative back office for a flat fee you want. Try virgil.com if you want if you're a company that wants to do lean startup trainings, you can go to leanstartup co. I could do this all day. So I'm involved with a lot of companies is the short long and short of it. But yes, if you want to follow me, the best way is to get on my mailing list, which you can do at Incorruptible Co or follow me. Obviously in all the usual places on
- [78:33]
A
social media, people say I have a lot of businesses and a lot of things going on, but now they know that the people I look up to like Eric, they're doing a lot more. If you want to see a glimpse into hopefully the future of Akash, that's Eric. You have had such a huge impact on the tech industry, on the product world. It's been an honor to get a chance to speak with you. Thanks Eric.
- [78:54]
B
Take care. Thanks a lot. Thanks for the kind words. That's really very nice of you.
- [78:58]
A
I hope you enjoyed that episode. If you could take a moment to double check that you have followed on Apple and Spotify podcasts, subscribed on YouTube, left a rating or review on Apple or Spotify and commented on YouTube. All these things will help the algorithm distribute the show to more and more people. As we distribute the show to more people, we can grow the show, improve the quality of the content in the production to get you better insights to stay ahead in your career. Finally, do check out my bundle@bundle.akashti.com to get access to nine AI products for an entire year for free. This includes Dovetail, Mobin, Linear, Reforge, Build, Descript, and many other amazing tools that will help you as an AI product manager or builder succeed. I'll see you in the next episode.