Investing in the age of AI, part 2: Jeff Bussgang

0:00Starting a business comes with its share of ups and downs, which is why staying true to your vision is essential. a non-negotiable for Romeo and Milka Bregali, Capital One business customers and co-owners of Ra's plant-based restaurant in New York. Romeo and Milka took a leap of faith when starting their own restaurant, gutting an empty space and building it from the ground up. Every pipe, every wall, every detail. But building from scratch came with a heavy financial burden, which is when they turned to their Capital One business card. With the flexibility of the card's no preset spending limit, they were able to spend more and earn more rewards while bringing their vision to life.

0:36Today, Raz's success is proof that with passion and the right support, it's possible to make your dreams a reality. Learn more at CapitalOne.com slash business cards.

0:50The thing that I really hope for is that AI allows us and frees us up to be more human, to spend our time on the human things that are unique to us, like relationships and love, insight and strategy. and I think the most successful professionals in the age of AI are going to be the ones that really lean into that. I don't think AI is going to replace founders anytime soon, but founders who use AI are absolutely going to replace founders who don't. Jeff Busgang is general partner at Flybridge Capital Partners, an AI-focused fund, and he teaches entrepreneurship at the Harvard Business School. He's also the author of the new book, The Experimentation Machine, which is a playbook for how founders can harness AI in their startups.

1:42Last week, we spoke with legendary investor Vinod Khosla about how AI is disrupting every level of our lives. Now, in part two of our investor series, we're getting granular. I'm talking with Jeff about how founders can use AI to help elevate their businesses. We'll talk about what it means to be an AI-native company, how AI can make you a 10x founder, and how to scale without growing. I'm Rana El-Khalyubi, and this is Pioneers of AI, a podcast taking you behind the scenes of the AI revolution.

2:25Before we get into my conversation with Jeff, I want to talk about some headline news from the past week. The U.S. made a series of AI-related deals with countries in the Gulf that should have some pretty big implications globally. And here to help us break it down is Jeremy Kahn. He's the AI editor at Fortune. Hi, Jeremy. It's great to have you back on Pioneers of AI, two weeks in a row. Hey, Rana. It's great to be back here again. All right. So we wanted you to come on the show to break down some recent news coming out of the Gulf. And I'm particularly interested in what's happening there because I'm Egyptian.

2:57I grew up in Kuwait and Abu Dhabi. And I've just been so fascinated by how the region is investing heavily in AI. They've been doing this for a while, but it does feel like last week really solidified that. So let's first set the scene. President Trump took a three-day trip to the Middle East where he visited Saudi Arabia, Qatar, and the UAE. And he was also joined by some of the biggest tech leaders like Elon Musk, Sam Altman, and Jensen Huang. Why make this trip? Well, I think for the U.S., you know, this is an area where there is a lot of money and a lot of interest in investing, particularly in new technology.

3:39You have a lot of countries in the region that want to diversify their economies. It's particularly true of Saudi Arabia that wants to diversify away from oil. So I think, you know, the U.S. and a lot of U.S. companies are looking at the Gulf as a good region to potentially gather funds. You know what really strikes me, and I'm curious what your take on this is. So NVIDIA and OpenAI are already global companies. and they could have struck these business deals with Saudi and the UAE anyway. Why did it take the U.S. president to broker these conversations? Yeah, it's a good question. There is a strategic element and a kind of geopolitical element to what's going on here.

4:23The U.S. sees itself as in this kind of AI race with China in particular. and the Gulf, just by virtue of where it's located and a lot of its history, has, you know, kind of been in between these two, sort of the West and the East, as it were, and has kind of played that position to its advantage. So I think, you know, for the US, there's been this issue that they were concerned that some of the technology companies in the Gulf region were kind of close to China, were potentially going to use Chinese technology. The US didn't want to see this happen. So some of what was going on here, and one of the reasons that you had such governmental involvement, was the kind of geostrategic implications of this.

5:05There had already been this effort to get G42, which is this big UAE, AI, and tech company, not to use any Chinese infrastructure, not to use any Chinese chips, and to use American chips. But then, of course, there was this problem in that the US had also restricted the export of the most advanced NVIDIA chips to most countries in the world, actually, because they were worried about the potential that this technology would kind of leak from these other countries to China. And those export restrictions had initially affected the Gulf states. The Gulf states wanted to see those export restrictions lifted.

5:38They wanted to be able to obtain the most advanced NVIDIA hardware. And so sort of for all those reasons, I think there was this, you know, you had to have the government involved to make sure that not only that NVIDIA could strike this deal, but that those chips were going to be allowed to be exported to UAE and Saudi, and also to make sure that this was part of a wider kind of alliance where the U.S. is kind of trying to cement its influence and make sure that China does not have influence in that region. Okay, so there were a lot of deals that were made last week, some energy-related, some manufacturing-related, and a lot of them AI-related, particularly in the UAE and Saudi.

6:17Can you walk us through some of these kind of commitments and their importance? Sure. So Trump, when his traveling entourage of CEOs first went to Saudi Arabia, where they signed a deal with a brand new company that the Saudi government has set up called Humane. And they agreed that Humane was going to build out a data center, initially 500 megawatts, which is about the size of a lot of data centers that are now going up in the US, a 500 megawatt data center that would have access to initially 18 ,000 NVIDIA top-of-the-line GPUs. So that was sort of the first deal that was struck there. And then they went to UAE.

6:56And in UAE, they struck this kind of mega deal. The UAE is interested in creating this very large data center campus. And that deal is to supply ultimately five gigawatts of energy to this data center campus. It will eventually have, and the deal that was struck was for 500 ,000 NVIDIA GPUs. But this is one of these interesting aspects of this, again, for geostrategic reasons. A part of this agreement is that that data center infrastructure is going to be kind of managed and operated on behalf of the UAE government by U.S. tech companies. So it's kind of an interesting arrangement. Yeah. What's interesting, too, is it's reciprocal, right?

7:40So Saudi Arabia's company Data Vault will invest$20 billion in AI data centers and infrastructures in the U.S., which is also very interesting. Yeah. The U.S. tech companies are looking to build so much infrastructure in the U.S. that they really do need to find these external sources of funding. But of course, people have raised some interesting questions about that too, because if you have the Gulf nations funding the development of these data centers and potentially owning the real estate on which these things are being built, there are some issues about control and what happens if the interests of the U.S.

8:16ultimately diverge from those of these Gulf nations. Is that going to cause a problem down the line? Was AI security and AI safety part of the conversation at all? Well, so security, yes, security was definitely part of the conversation. And if you look at the rationale for why it's going to be these US companies managing this facility in the UAE, there was a lot of discussion about that was in order to make sure that these facilities were secure, that it was going to be difficult for people to hack into them and potentially steal the know-how, both either of the chips themselves, so the physical security of the facility, but also the technical know-how of how to build these large frontier areas.

8:54AI models. Security, yeah, it was definitely part of the conversation, but safety less so. I think as a whole, safety is kind of taking a backseat here. Everyone's talking about trying to push innovation and progress as far forward as quickly as possible. And so I don't think there was as much discussion about AI safety. You know, one of the things that I find really fascinating, having grown up in the Middle East, is there's a mini race between these countries too. It's not like one monolithic, you know, kind of Middle East. Actually, Saudi is in a competition with the UAE to establish itself as the AI hub.

9:31Do you have any thoughts on that? Yeah. So if you look up at the standing up of this company Humane in Saudi Arabia, which is funded by the sovereign wealth fund from the country, the public investment fund, it clearly is being positioned as a kind of rival to G42, which is the big company that the UAE has already established. And if you even look at what they're saying, you know, One of the first things Humane is going to do is train a large language model that's very good at Arabic language. Well, it turns out that G42 already has an AI model that's very good at Arabic language. So, yes, this is clearly like Saudi Arabia, you know, not wanting to be left behind by its regional rival, the UAE, and trying to say, hey, we can be big players in AI, too.

10:12We want to have a contender in this race to be at least the kind of dominant regional player when it comes to AI. Well, we will definitely be watching this space. Thank you, Jeremy, for joining us today. Thank you so much, Rana. It's great to be here again. We're going to take a short break. Stay with us.

10:47if you've spent any time building ai products or leading technical teams you know this transformation doesn't fail because of ideas. It fails because teams can't move together. Enter Atlassian's teamwork collection. It has planning in Jira, documentation in Confluence, video updates in Loom, and now AI agents in Rovo, which connects the dots across your work so nothing gets lost. It's one AI-powered teamwork platform designed for how modern teams actually build. Learn more at atlassian.com slash teamchanger. That's A-T-L-A-S-S-I-A-N dot com slash teamchanger.

11:36Hi, Jeff. Welcome to Pioneers of AI, and it's so good to see you again. Thanks. Thanks for having me. So you recently published a book, The Experimentation Machine, which I loved reading and we're going to talk about. But what inspired you to write the book? So I have two hats, as you know. One is I teach at Harvard Business School in the entrepreneurship unit, and I focus on launching tech ventures. So I get all the founder aspiring students. And then secondly, I'm an early stage venture capitalist, my firm Flybridge. All we do is focus on AI investing. And so I was seeing in those two worlds a whole new trend around using and leveraging AI to become 10x founders.

12:14And so the book tries to bring that to life. Yeah, we're going to talk about all of that in a second. But before we get into that, I wanted to hear more about your origin story. You spend a lot of time at actually a number of startups. So you're kind of an operator turned investor of sorts. So share some of the highlights of your journey and how did you end up with this investor hat? Yeah, so I was an undergraduate in computer science at Harvard. I graduated in 1991. My focus was on AI and natural language processing, which in the late 80s is hard to imagine. and then I went into a business career after also getting my MBA at Harvard as a tech founder and in the mid-90s was involved in a company that went public, Internet 1.0 company called Open Market.

12:58I was an executive team member and had an amazing ride and then in the early 2000s I co-founded a company called You Promise, also venture-backed, also tech, also an amazing ride. It sold successfully in the mid-2000s. Both companies were venture-backed by Greylock And when Greylock shifted west in the early 2000s, two of the young Greylock Boston partners spun out, and with me, we formed Flybridge. So I had this entrepreneurship background with their investing background in combination. How did your entrepreneurship background influence your lens on investing? I think what I would say is that I have a huge amount of founder empathy.

13:38and yes, I have the operating experience of taking a company public and building multiple large companies that were successful, but more at the core is founder empathy and that emotional rollercoaster and that sense of loneliness that founders undertake when they go on this journey and just being as much as I can, a advisor, conciliere, friend, companion, mentor, coach, whatever it takes to help them get through to the next level. I think that's so important because kind of drawing from my entrepreneurship experience too, it is an emotional roller coaster. Some days you feel like you're on top of the world, you got this, and then the next day, I don't know, you get a no from an investor or a customer or you lose some of your talent and you're like, oh, it's existential, right?

14:21And so being able to have investors where you can talk about this openly and feel safe to bring this up is really special. I think one of my founder friends captured it beautifully when he said, I eat no's for breakfast. and just that sense of always having to push through negativity and always focusing on that sense of belief and optimism that the next door, the next door, the next door, whether it's an investor or a customer or M &A candidate. And obviously the macro environment has a lot of turmoil and you can't control that as a founder. That can be very frustrating. So really just being an empathetic advisor, investor, and friend, as I said, is really what I center on.

15:05Talk a little bit about Flybridge's investment thesis. And also, you've been investing in AI for at least a decade, right? What's your view on the AI landscape as it relates to investing? Yeah, as you said, Flybridge has been investing in AI for 15 plus years. And our original thesis was more around data and machine learning. And that led to companies like Zest AI and MongoDB, which was an important data infrastructure company. And then later, we've been more focused on the application layer, and that's really where we're centered today. We believe that the foundation layers are getting commoditized in a very good way.

15:43Prices are coming down for AI compute and inference dramatically, and that enables this massive opportunity at the application layer. I refer to it as the AI dividend, much as we all benefited from the cloud dividend over the last few decades. We're entering into a moment where application software providers are benefiting from these incredible AI dividends that are coming our way. Do you only invest in AI companies? Exclusively. So at Flybridge, our thesis has always been around enterprise software. And AI was an important component of that over the last few decades. But in recent years, it's all we do.

16:21So we're exclusively an AI-focused investor. I'll say there's one exception, though, and that is we are big believers in AI-native founders, which I'm sure we'll get into. And if we see an AI-native founder operating in a slightly non-AI-native space, we take that exceptional opportunity to invest in them as well. So in that book, you talk about AI-native companies. Can you define what that means? An AI native company is a company that infuses AI into everything they do, every function and every process. When somebody comes to the CEO and says they want to hire someone, the first thing the CEO says at an AI native company is, could AI replace that hire?

17:05When someone describes a new process or a new workflow that they're creating, the first thing that is asked in the meeting is, how could we use AI to do that more efficiently? So an AI-native company is leveraging AI across engineering, sales, marketing, customer service, executive, functions, finance, everything, everywhere, all at once. You know, I'm starting to see companies that are basically, their product is essentially an AI employee. Like one company, tough day, they're building AI managers. There are AI healthcare administrators. Do you envision a world where an org chart will be a combination of humans and AI?

17:43Absolutely. In fact, when I talk about the contents of the book and the moment we're in, I show the standard traditional org chart of a seed stage company. You or I might invest$2 or$3 million. The team would hire 10, 12, 15 employees that go build the product and that race to get to the$1 million of ARR magic threshold. Today, it's two or three employees and hundreds, if not thousands of AI agents working in tandem to get to the magic$1 million ARR threshold. So, yeah, I think you're going to see this dramatic change in org charts. Sam Altman has this famous sort of throwaway line he uses where he says his founder friends joke about when will we see the solo entrepreneur that has achieved the milestone of building a unicorn, a billion-dollar company.

18:34as a single employee. And I looked in a recent analysis I did of unicorns. We now have a dozen unicorns with less than 100 employees. And the question is, in 2026, will they be able to do that with less than 10 employees? And in 2027 or 2028, five, three, two employees? That's really the moment we're in with the age of AI for AI native companies. All right. So AI is moving so fast. And one of the questions I have with my investor hat on is defensibility. I often tell founders, you know, if you wake up every day concerned about the next version of ChatGPT and it's going to put your company or your product out of business, then I don't want to be an investor.

19:14That's not defensible. However, if the next version of these foundation models strengthens your product and your service, then there's something there. How do you think about defensibility, especially with the pace of change and acceleration of developments? It's a great question, and it's the question in every one of our investment committee meetings. And the three things that we focus on outside of team, which I can come back to, but just in the core fundamental business and product are, one, are there sources of proprietary data? So, for example, we have an AI for legal software company called Noetica.

19:51They ingest thousands and thousands of contracts on private corporate debt transactions. and those debt transactions have proprietary terms in them. And so they now have all this insight into who Citibank lends to and what the terms are and who JPM lends to and what their terms are and all the way down the line. So that when a new lender steps to the table, they can do this incredibly rich terms analysis. That's a unique data set that only Noetica has. So the data mode is really important. Second is around workflows and that sort of last mile. The platforms that the foundation models are providing are incredible, but they're not trying to nail the interface for the last mile for the user.

20:36So for example, we have a portfolio company called Allspice, which is for hardware engineers. And it's that last mile of doing hardware design and collaboration, the GitHub for hardware, that allows us to have confidence that the engineers are in that platform at Allspice all day, every day. That's what they live in. And they're beginning to really get addicted to that workflow. And then the third area is around that human AI interface. It's that insight into how humans will interact with your AI platform. You know, if you look at the history of ChatGPT, ChatGPT was an experiment that was thrown on top of the SDK, the API that OpenAI was launching.

21:15And it turned out to be an incredible interface and the rest is history. So founders that really can understand that interface and the human AI interface to manifest these AI capabilities, but do it simply and in a way that's usable. Yeah, a few thoughts on that. I was speaking at a ServiceNow event recently, and one of the questions that came up is, and I think there are studies, there are now people looking at this, if you build personality into these bots or interfaces, they're more likely to adapt and kind of, especially when it's part of a team, if these bots have a personality, they kind of, I don't know, they fit in better into the team culture.

21:53And I think we're going to start to see more of that. I think you're totally right. I know this is an area you have a lot of expertise in, but that emotional connection that we're going to have with the AI agents that work alongside us in the organizations, I think that's going to be really powerful. And so the flip side is the software companies that can bring that emotional intelligence into the AI agents and help them work with the humans and collaboration are going to be the successful software companies. I also have a thesis around the evolution of the human-machine interface anyhow. Like if you think of a smartphone, it's not really AI native today.

22:26And so I think a combination of conversational, perceptual interfaces, we'll see a lot of innovation there. I haven't seen the thing that I think is like the iPhone equivalent, but I'm curious, do you invest in these companies? Are you keeping an eye on that? We do. We're very much keeping an eye on that. Obviously, pendants and other interfaces and hardware devices and these consumer hardware interfaces, I think, are really rich areas for innovation. Yeah. So let's next talk about the book and how we bring AI to the startup landscape. And first off, you talk about the 10x founder. What is that and how do you look for these 10x founders?

23:03So everyone knows about the mythical 10x developer. That's the developer that's so effective. They're not just 10 % or 20 % better than the average developer. They're 10 times better. And for a while, founders were sort of limited in their ability to use tools to be more productive because the tools were software productivity tools that everybody had at their disposal. But now these AI tools allow us to see founders operating like 10x founders being 10 times more productive and organizations becoming 10x organizations, leaders becoming 10x leaders, 10x salespeople, 10x marketers. And it means that founders can now lead organizations far more efficiently with far fewer humans.

23:48Yeah, I think that is so fascinating for a whole bunch of reasons. First, I think it makes this early check so much more important, right, as an early stage investor. If you're going to assume that these companies are going to be very capital efficient, then yeah, then that early check might have the most leverage. Is that the way you think about it? I think the venture industry is really about to change dramatically for that very reason. So if you look at what's happening in the marketplace, companies are building without needing capital. They can seed strap their way to a million, two million of revenue and then scale with modest capital infusion to over a million of revenue per employee.

24:30We have a number of our portfolio companies doing over a million dollars of revenue per employee at early stages. And that's because the founders and the teams are just so incredibly efficient and leveraging these AI tools. Okay. So I'll ask you one more question, and then I want to bring all of that to life with examples. So let's talk about product market fit, right? How do you define product market fit? And how does AI help accelerate getting to that product market fit? So one of the themes in the book is that we're in a moment where we're combining these timely AI tools with timeless methods for business building.

25:06And it's important not to lose track of the timeless methods, not to get too caught up in the technology and the latest tool. And so the way I describe product market fit and the framework I use, which is the hunch framework that I describe in the book, I try to really focus on those timeless methods. And it's five letters. So it begins with H, which is the hair on fire value proposition. It's not nice to have, it's beyond must have. It's your hair's on fire, you need to solve this problem. It has to be a top one or two priority. U is for usage. You wanna see a high degree of usage. As many people know, usage is the most important indicator of product market fit.

25:46N is for NPS score. NPS stands for Net Promoter Score. It's a metric used to measure customer loyalty, basically asking customers on a scale from 1 to 10 how likely they are to recommend a product. C is churn. You want the churn to be low. And then the final H is a high LTV to CAC ratio, lifetime value to customer acquisition cost. good unit economics. You can't forget the timeless importance of profitable, sustainable business building. With those five letters, the five hunch letters, then you say, okay, we are going to have different stages. In the early days, it's going to be nascent and we're going to see naturally solid performance in some, okay performance in others.