79. Roman Stanek: "Assume the worst scenario and then double it, and then double it again"

0:00Welcome to The Difference Engine, the show for tech founders, investors, and innovators.

0:11Roman Stanek:This is an episode we've been waiting to recall for quite some time. We've got an extraordinary European entrepreneur, founder of three companies, Mr. Roman Stanek. He's the CEO of Good Data. Although born in Prague, he's now a resident in Marin County in California with three successful companies built under his belt. And he's a man that has an incredibly good grip on what is happening in the world of AI and what it means for your company. Absolutely. Listen up to learn how to scale a business in today's crazy AI environment, how to keep people focused and the truth about 996 and what it takes to succeed in software today.

0:55Hey, Roman.

0:56Roman Stanek:We've been waiting to do this for a long time. Welcome to The Difference Engine. How are you? Good, good. Good morning. Good morning from California. And afternoon over here in London. We thought we'd get you on because there aren't that many people we know who are European, an engineer, and have started and successfully grown three companies. And as you said, you've washed up in Northern California, not Silicon Valley. Can you take us through a little bit of that journey, please? Absolutely. I actually started my first company in Prague and completely built it in Prague in the late 90s in the first internet bubble.

1:33And it was absolutely crazy. It was absolutely crazy. It is actually, that's probably the closest to the AI bubble today. And I was able to build it on almost no funding again in Prague. Yeah, so good cost of engineers. Yeah, and it was kind of different. It was, everything was kind of super cheap. And we, you know, raised very little money, sold the company to some microsystems. The company was Java Development Tools called NetBeans. I still meet people today who tell me that they actually grew up on NetBeans and they use NetBeans. And, you know, sometimes I get, I meet people and they ask me, like, what's the future of NetBeans and so on.

2:14So it feels, that feels really good. And about a year into NetBeans, I got a call from Sun Microsystems in the U.S. from the head of M &A, and they asked me if I would like to talk to them about selling the company. So I saw the company, which was a super difficult process, you know, stories around that. And then I moved to, you know, I worked for Sun for about a year. They came to me and said, you did well. You know, they accelerated everything. they like they kind of um they they knew i wanted to start another company and i started another company in 20 uh 2001 and moved to boston for that one and you know sold that one in 2006.

2:58Roman Stanek:and people may forget but sun was an absolute leviathan in terms of like software engineering it was the place to be right sun was sun was so it was the dot in dot com um if you remember that So they were actually discussing in 2000 getting rid of their sales force because there was so much demand for their servers. They actually wanted to auction servers on eBay, which lasted about six months, and then everything crashed, and they sold the company to Oracle for peanuts and so on. But it's actually very interesting. There are two things interesting. Sun is like open AI and entropic today. Exactly.

3:45Everything what they did was unbelievable. They scaled. They didn't need to sell. Everyone that came to them, they were in the middle of everything. And the second thing is that the culture of Sun, if you look at it, Eric Schmidt, who is the CEO of Sun, became the CEO of Google. So the culture of Sun essentially became the culture of Google. So that culture still kind of lives on it. But yeah, it was absolutely crazy. And so going back to kind of my story, so first company was in Prague, the second company I started and sold in Boston, which was not, it was suboptimal. And the difference between East Coast and West Coast in the US, like people still, I think that everyone would like to see like a Silicon Valley in Boston and it's not there.

4:37and so I actually I was it was kind of impractical I was you know everything what I wanted to do was in Silicon Valley so I had to get on plane from Boston so when we saw the company I actually moved here and I started the third one you know straight here and I don't live far you position it like Silicon Valley is not Northern California I'm like you know 50 miles away from Silicon Valley and I actually believe that in today's world San Francisco is kind of the epicenter of of silicon valley as much as um you know samozek good well as soon as they

5:11Roman Stanek:clean that city up i'll be right back there but um three companies uh and um you know a move successively west we'll come back to that because we want to talk about um the differences between european and american entrepreneurs in a bit um so czech uh or czechia i do get confused to slovakia I think I've got them in the World Cup ballot, so hopefully good luck there. But this country, I don't think enough people know about it. Could you give us a sense of what it was like growing up there and the engineering ethos, et cetera? It is a very industrial, like engineering, philosophy culture, mathematics, engineering, philosophy culture.

5:50I would say to the point where the sales and marketing somehow kind of atrophied. So, you know, we have good engineers in Prague and Czech Republic, but not enough salespeople and marketing people and product managers. We are building that and training that kind of that muscle. That muscle was not there historically. And, you know, it's interesting because, you know, I had a chance to meet with Roberta Metzola, the president of European Parliament the other day. And she asked, like, you know, the audience, like, what they could actually do better. And I said, look, you know, the Czechs are producing cars.

6:35You know, we are really good at it. You know, Škoda makes more money than Porsche. The politicians shouldn't try to build Silicon Valley in Europe. That's impossible. Again, no one can build Silicon Valley in Boston. They should just stop kind of being the problem. And I told her that, you know, the European Parliament should do only three things, deregulate, deregulate, and deregulate, because that's a problem for good data, AI, as it is for Škoda regulation.

7:04Roman Stanek:Christina Cofrero was on this pod recently, and I think she would violently agree with you. And you're right about, we, Jono and I, are petrolheads. We love Tatra, a name from the past. As you mentioned, Škoda. What about the software side of things? I mean, there are a couple of players, and you obviously employ a lot of people in Prague, right? It's kind of a, it's a little bit of a puzzle because, you know, if you look at, let's say, Poland, you know, they have 11 labs, you know, they have real big visible AI success and so on. The same in France and in Germany. Again, the Czech culture is much more like an engineering culture, so I think there will be plenty of engineers, but they have not that many kind of large internet or AI big success stories, so they are not that visible or we are not that visible.

7:58But things are changing, and I think that there are more and more people who can actually build companies. Yeah, and isn't it true that with this engineering background, somewhere along the line you created some software which is now on Mars. Is that true? Yeah, it is. It's actually, again, NetBeans was very kind of an innovative company. And in a way, when we started NetBeans in 1997, it was very similar to today. There were no rules. There was nothing to guide us. There was no tech crunch. People couldn't Google NetBeans because Google didn't exist. Not even on AltaVista? Yeah, exactly. People had Alta Vista and it means, you know, I was kind of right out of college and so on.

8:49But my point is that today it's the same. People assume that they know how to do AI. People assume that they know what AI is and how it's actually going to work and so on. um internet was you know i i i sold you know i worked for a u.s company for you know a couple months in in 90 95 96 and we had like a um internal communication channel and i told the ceo we should actually put it on internet and he said he would never put like internal communication you know up internet with all the porn and so on so this everyone saw internet like it's just the place that will be used by a bunch of crazy people and so on.

9:31And it's the same with AI today. Still, I meet people who believe that AI is just some sort of a template for computers to replicate a job and so on. I don't think that people understand the impact. It's going to be more disruptive. AI will be more disruptive than internet ever was. And is there anything in particular, given your experience over the last couple of years, you see happening with AI and the economy in general? The problem with AI is that we all have our kind of experience, our lived experience with computers tells us something. You work with PC your whole life, you know what computers can do.

10:15You go to chat GPT and there is a little screen, you talk to it, and you assume that this is kind of it. And you don't understand that on the other side, there is a giant data center with NVIDIAs that are so powerful and so much compute power that no one ever experienced that. So we are dealing with a beast that we have no idea how to deal with it. And it looks very innocent. It looks like a chat GPT, chat GPT, like a screen. But I think that people who actually know the full potential are people who code with AI. I'm trying to code with AI. Sometimes I spend the whole weekend coding with AI and getting advice on that.

11:04And at some point, I'm almost building a personal relationship with my coding assistant because there's so much going on. It's like you work with someone for the whole weekend. You kind of become friends. So it's not, again, it is kind of incredible. And then you read all these kind of mathematical proofs that Entropic and Codex are now discovering and so on, and you can start to see kind of the full potential. And we are just three years into it. Imagine being like 50 years into it. So, I mean, given that, this sort of, we are right at the beginning of something we really can't even hope to understand because it's so big.

11:43what advice would you give business leaders at the moment to make sure that AI is working for them? Because there is a danger that AI doesn't work for us. It works for it. It works for its owners. I would say that AI will work for open AI and entropic. I think that you see the entropic just raised boatloads of money yesterday. day. When I meet with clients, I see essentially two types of companies. I see companies that are business leaders who take it very seriously. I have a large customer in the U.S., a major, major, you know, company that you read about in the news every day. They just did a, you know, two-week off-site where the home management was kind of sitting and strategizing how to actually compete with entropic.

12:38And that company is not in, it's a financial services company. It's not, it's not company in AI, but they absolutely understand that entropic is kind of existential threat. And then I see companies that don't do anything and believe that this is just, you know, again, it's like a, it's a, it's a, it's a, not nothing major that, you know, we have some new templates for work and people will be able to write emails better and so on. And I think that those two, you know, the companies that underestimate AI, there's probably more of them and it's going to be very dangerous. Yeah, it's interesting that the FT today has a piece on the back page that says, suddenly the software apocalypse was over.

13:23And I suspect people in that latter group are breathing what they think is a huge sigh of relief that maybe AI isn't as impactful as everybody's saying it is. I don't think that that's what they actually mean. I don't think that that's what they actually mean. What they actually mean is that SaaS can actually leverage AI to transform itself, you know, the SaaS industry. It doesn't mean it's not going to be in fact. And we are talking about SaaS industry. We are not talking about anybody else. So the SaaS industry, you know, for the last six months was like, you know, it looked like everyone would replace CRM, you know, Salesforce with Entropic and, you know, Lord David, you know, Entropic and OpenAI and so on.

14:06And there is some hope. So I would say I would see it as like hope, not as, you know, the judgment is still up.

14:12Roman Stanek:Yeah. It's the first quarter of the ice hockey game. I know you like ice hockey, so maybe. I want to take you back a little bit to the like the pre-Saspocalypse when you were scaling up the companies and obviously, you know, Boston, Czech, then the Valley. Could you give us a sense of the difference between those early days of scaling a company up and making big deals like the financial services company you mentioned? I know you've got some good stuff going on in Europe soon as well. What advice would you give people who have something hot, maybe it's an AI product, and they want to scale it to the next level?

14:51You have to appreciate complexity if you deal with big companies. There is no shortcut. There is no product-led growth, you know, trial. People will pick it up and somehow magically the whole company will use it and they will pay you millions of dollars. It's large companies. They have antibodies against new vendors. They hate new vendors. You know, onboarding, we had an example where it took us about 18 months to be onboarded by a large company. just being onboarded by their procurement and so on, because no one wants to do it. No one wants to have a new vendor. They would much rather work with existing partners than to rely on someone new.

15:35And so there is this kind of difficulty to get in that can be only overcome if you're actually super focused on complexity, solve that one complex problem, deal with it, don't get distracted by anything else. That's kind of my advice, because, again, there's no shortcut unless, again, you're like Sun or Entropic, and people come to you and ask you, all of the companies are essentially coming to you. So that's kind of where I see the analogy between OpenAI, Entropic, and Sun is that every large company now understand that they have to work with them. And for anyone else, you have to kind of work hard to get in.

16:21Roman Stanek:Yeah, we're coming up with a term called the vibe chasm to describe the problem that you won't be able to leap it now that those guys have taken all the low hanging fruit. So love to talk to you about that at some point. um you've been you have also a unique perspective in that you know the u.s and you know europe very well um we've talked about this personally many times but how would you characterize the tech adoption attitudes and uh maybe this leeches over into economics a bit the difference between u.s and european approaches to tech and tech entrepreneurship in in the u.s it's still kind of um they are still they're still pioneers you know i i had customer who um bought a million dollars of software and they wrote it off about two months later because they changed the strategy and no one was complaining about it was their problem and they just wrote over a million dollars imagine someone in europe buying a million dollars of words of software and then writing it off they would be they would be unemployable unless they were government but yeah okay take your point Yeah, exactly.

17:25But then we are talking about billions of dollars. Exactly. Exactly. But I think that's the problem. When I moved to the US in 2001, it was right after September 11th, and we were building software for integration of systems, which was a big problem of 9-11. and like six months into it, a Czech company, I just moved there, I had some cheap visa and so on and then a fax machine started printing orders from Department of Defense. They wanted to buy a software to integrate a system and we actually built half of the company on that. You cannot imagine that happening in Europe. That would never happen.

18:16And then that's one example. And the second example is that when we actually meet with clients, and we have advisory boards, you know, like large companies, they come, they share their ideas, they kind of want to talk about it, they want to show everything else. If I do this in Europe, everyone will have me sign NDAs, and they will be like, you cannot talk about me, you cannot mention my name, I'm not going to tell anyone what we do. So it's much more secretive, it's much more kind of closed. And again, failure is difficult. So the European trials take, you know, 18 months because people want to check every single box, and they never do.

18:54But that's a different problem because the technology is changing so quickly. So, yeah, it's a fundamentally different environment. And now, with kind of the speed of AI, this is becoming almost impossible, you know, for European companies to actually compete and understand.

19:10Roman Stanek:Yeah, right. I was watching, there was a debate on television last night, and they had some supposed experts, one CEO who knew what he was talking about, another guy directly contradicting him. and the guy from the government was just nodding at everything even when they said what we should not do is let a certain company in and that certain company who you can guess who it is is all over government uh so the guy was nodding like we should stop them but it's already he doesn't know what's in what's out but that's you know because very few politicians have any background in business let alone technology so they're ill equipped to make decisions not only that, but they also want to kind of regulate it.

19:47So they want to be behind the steering wheel. Like this is where AI will go under our leadership. So that's a problem. So the point is, is that those of us that exist in the technology environment are constantly evolving what we do and the technologies that we do it with. And I think you've got a good instance of that because you've evolved good data from BI to AI. Can you talk us through that? Because that's always an exercise in evolving a company to survive and go to the next stage. And survival is the right word. I think that as many or most SaaS companies are dealing with this kind of notion that anyone can wipe code anything.

20:35It's anyone can build, you know, features overnight. The AI models are much more, you know, more and more capable, like the AI models double their kind of cognitive functions every 200 days. So when we, you know, every time we talk, you know, things double. And so for us, and for not just for good data, but for everyone, you have to look at it and say, well, if I don't do anything, we will be out of business in 18 months. This is really, really dangerous. And yet, you deal with engineers who don't want to change. You deal with engineers who have certain experience and they believe that they actually know what we should be doing and so on.

21:22So getting a sense of urgency, getting that alignment, all the signals are against you. You know, everyone's saying, oh, I don't see that. I don't see that problem. Nothing is happening. Customers are still buying. Customers are still telling us, you know, we don't need this and so on. So it actually takes like real efforts to actually change the company. And, you know, I don't like aizoki. I actually like American football much more. But my kids played soccer. And one thing I learned from Saka, from the best coach on the local team, he called it one voice. He always, you know, when the parents were actually screaming from the sidelines, he kind of invited them, and he was screaming at parents and everyone, like, I'm the coach here, I'm telling everyone what to do.

22:15No one else will actually, it is one voice. So I think that every company in this environment needs one voice. as the CEO voice and everything everyone else needs to follow.

22:25Roman Stanek:Yeah, I totally agree. So we are constantly saying this to clients, and we could probably name some very big AI names where they've boasted to us, we're going to be the open AI of Europe. They've talked about 500 million euros as a phone, and we've just literally told and straight, that ain't going to be enough and you ain't going to win. But unless you've got someone at the top who sees it, you can't go. Yeah, exactly. And then there's always some dissenders, There's always people who want to go in a different direction, who don't believe that the change is necessary and so on. Yeah, that's life.

22:57Most people don't like change. That's what makes entrepreneurs entrepreneurs. We like change. Most people don't like it. Most people don't like being in the gray zone when the old doesn't work and the new doesn't work yet and so on. So that's very uncomfortable for most people. And that's kind of where I live my life, in that gray zone.

23:16Roman Stanek:Now, speaking of gray zones, recently we had a very nice dinner with you and some esteemed guests from oil companies and large investment funds, etc., etc. Fabulous meeting of minds. One of the things that we discussed was AI adoption and the speed of it, whether it's going to be carrying on at the pace it is, whether it accelerates, whether some people will not adopt and therefore die. What do you honestly see out there from when you're selling your good data AI products? How does that all work? What happens when the river hits the road? And again, it's different in the US and it's different in Europe.

23:57The stakes are much higher in US. The competitive environment is much stronger. There's always new company being funded and so on. So there is a few exceptions. And government is obviously one of them. But with few exceptions, there is this real sense of, again, existential threat. In Europe, it's not the same. You have national monopolies and companies with guaranteed business, and they own the national markets, and they don't need to move and so on. So I would say it's going to be... I'm actually getting to the point where I believe that Europe didn't want to go into tech full speed 10 years ago.

24:40So I actually think now it's maybe too late. Maybe it's actually too late. Maybe it's impossible. I mean, all the national interest and kind of competing priorities and so on, maybe it's too late. Maybe tech is not good for Europe.

24:54Roman Stanek:And in terms of rounding up a little bit where you see the future, and obviously you're on this journey, but an exciting set of products coming out that are being adopted, as you say, Gray Zone, getting there. how do you think business leaders so the people you sell to what is your advice for those guys as they think about AI strategy etc don't underestimate it assume like the worst scenario and then double it and then double it again okay but look at it look at it again it's doubling the cognitive speed is doubling every 200 years and the funny thing is that the leaders from entropic and open AI They were telling everyone how AI will kill every single job in 18 months or 12 months and so on.