Jeff Bussgang — venture capitalist, Harvard Business School professor, and author of The Experimentation Machine: Finding Product-Market Fit in the Age of AI — joins me to explore how artificial intelligence is transforming the startup playbook.
We discuss how AI compresses the cost and time of learning, why execution velocity now matters more than moats, and what defines a true 10X founder and an equally valuable 10X joiner. Jeff shares lessons from MongoDB, HubSpot, and ClassPass, insights on avoiding the PLG trap, and why discernment, judgment, and taste may become the most valuable human skills in the age of automation.
We also touch on Jeff’s HUNCH framework for testing product-market fit, how founders can design kill criteria to avoid zombie projects, and what it means to create win-win outcomes in a world moving at AI speed.
I hope you enjoy this conversation as much as I did. We’ve shared some highlights below, together with links & a full transcript. As always, if you like what you hear/read, please leave a comment or drop us a review on your provider of choice.
— Jim
Links
Our Substack is growing! Subscribe below for more brain-tickling content designed to make you go, "Hmm that’s interesting!
Highlights
The Founder’s Clock Speed
Jeff Bussgang: And we have two ways that we define clock speed. And this is a term we picked up from Bill Gates from his early writing about the Internet 1.0 days. But you know, there’s one aspect of clock speed, which is their ability to make decisions rapidly and then the other aspect of clock speed which is just raw horsepower, just raw intelligence. And so, we really look for both. There are a lot of very smart people who are, you know, frozen analysis paralysis and they don’t make quick decisions. And there may be some people who make very quick decisions but exhibit very bad judgment because they don’t have strategic insight. And so, we really look for both aspects of clock speed when we evaluate founders.
The Revenge of the Liberal Arts
Jeff Bussgang: And now if the AI is doing it for our young people, how are they actually going to know what excellent looks like? And so really being good at discernment and taste and judgment, I think is going to be really important. And for young people, how to develop that. I think it’s a moment where it’s like the Revenge of the Liberal Arts, meaning, like, go read Shakespeare and go read Homer and see the best movies in the world and, you know, watch the best TV shows and be strong at interpersonal skills and leadership skills and communication skills and really understand human motivation and understand what excellence looks like, and understand taste and study design and study art, because the technical skills are all going to just be there at our fingertips for all the reasons you said, Jim.
Escaping the PLG Trap
Jeff Bussgang: …people think it’s very easy to shift from PLG to SLG just by changing pricing and by saying, hey, before you used to buy this thing on a seat by seat, one off basis, now let’s just do an enterprise deal and we’re done. What they don’t realize is that you, that transition from product led growth to sales led growth is an organizational transition and a cultural transition and a product and technology transition. So, if you want to avoid the PLG trap and break out from not only that 1 to 10 million and 10 to 100 million, but get to as MongoDB now has 2 billion of revenue, you’ve got to really go through that entire cultural, organizational and operational transition.
Humans and AI: The New Org Chart
Jim O’Shaughnessy: what does the ideal human team look like when the company is treating AI as a true and real partner? It’s a cultural shift where the humans need to be very comfortable directing the AI agents, in some cases overruling the AI agents, putting the guardrails up for the AI agents. We’re all going to operate in organizations where there are more AI agents working for us than humans. That’s just a fact. It’s going to be the case across business, across government, and across nonprofits. And so, the humans in those organizations need to be more and more comfortable orchestrating, managing, controlling, overruling, or being overruled by the AI agents. And they need to sign up for that.
🤖 Machine-Generated Transcript
Jim O’Shaughnessy: Well, hello everyone, it’s Jim O’ Shaughnessy with another Infinite Loops. Today I’m very excited because we’re going to be diving into the operating systems of startups in the age of AI. My guest is Jeff Bussgang, entrepreneur, venture capitalist, Harvard Business School professor. Wow, you got quite the CV, Jeff. Jeff is the co-founder of Flybridge Capital, a seed firm backing AI-forward community-driven companies. He’s the author of three books that map very cleanly to a founder’s arc. The first is Mastering The VC Game, the next Entering Startup Land. And the one we’re going to be talking most about today is the most recent The Experimentation Machine: Finding Product Market Fit in the Age of AI. Jeff, welcome.
Jeff Bussgang: Thanks Jim. Great to be here.
Jim O’Shaughnessy: I love your stuff. When I was going through it, we share quite a bit of outlook on where we see VC going. In the most recent book, you argue that great startups are not static companies, but rather systems for running disciplined experiments where AI will massively compress the time and cost to learn. Why don’t we start there? Tell us more.
Jeff Bussgang: Broadly speaking, in the age of AI, we are seeing founders being able to execute far faster than ever before. I refer to this notion of the 10x founder in the book where similar to what we used to talk about developers that were exceptionally talented and able to use the modern tools to develop software faster than anyone, we would say they’re not 10% better, they’re not 20% better than the average developer. They’re 10x developers, they’re 10 times better. At Flybridge and across the early-stage VC industry, we’re seeing 10x founders emerge and those founders are leveraging the modern AI tools to execute 10x faster than they might have five or 10 years ago.
There’s a kind of general view right now, and maybe we’ll talk about this in more detail, that technology moats are becoming harder and harder to establish because the barriers to building technology are so low with these modern development tools, but the one thing that is very hard to replicate is execution velocity. And so, the founders that are really dialed into the new tools and leveraging those new tools to execute faster than others, those are the founders that are set up to win.
Jim O’Shaughnessy: Yeah. And what was the moment when you stopped thinking about potential companies to invest in as companies and you started thinking, no, no, no, these are experimentation machines?
Jeff Bussgang: Well, I have to go back a little bit in time. People may remember The Lean Startup, the book by Eric Ries, which was really, I think, the first piece of work, along with Steve Blank’s work around lean startups and running experiments and thinking about minimum viable products and things of that nature, but what we’re seeing. And so, I started teaching at Harvard Business School 15 years ago, and I brought that methodology to HBS, into my classroom and in my teaching, where I teach about entrepreneurship and what have you. And so that notion of experimentation was very ingrained in my work, but the tools for running those experiments were now looking back quite rudimentary.
And over the last three or four years, both in my work at HBS with founders coming out of the HBS community and in my work at Flybridge, investing in AI forward founders and many 10x founders, I was seeing this totally step change in execution and leveraging these tools. And that’s when the light bulb went off for me. And I thought, oh my gosh, this whole notion of the lean methodology and this experimentation mindset is just being turbocharged with AI. And that’s what really inspired me to write the book and to write about, speak about and work with my founders and work with my students in that vein.
Jim O’Shaughnessy: And you have a framework which you dubbed the Hunch Framework. Hair on Fire Values, Usage High, Net Promoter, Churn Low, High Lifetime Value to Customer Acquisition. Can we go through these kind of step by step and tell our listeners and viewers why this framework is especially rewarding?
Jeff Bussgang: Yeah. Stepping back. Product market fit is a dynamic, emerging arena and it varies with maturity of your product and with your market development. And I have to give a little bit of credit to Fred Wilson on this. I had him as a guest in my class for a number of years, and as we were riffing together, he came up with this notion of having a hunch in the early days and having data in the later days when it comes to judging product market fit. So, I’ve been sort of stewing on that and developed this methodology and this framework tied to that moniker HUNCH as the acronym and the reason by the way I go into this detail is in marry it with the world of AI is because I really believe in this sort of notion of combining the timeless methods with the timely tools.
So we talked a little bit about the timely tools. We’ll talk more about them perhaps for software development and running experiments, but the timeless methods of finding product market fit and building a valuable, enduring, profitable company. You know, that’s there are decades and decades of methodology that we can apply to that. And so, this framework is to try to marry those two things. And so, one of the things that is a timeless concern that many founders skip over is having not just a nice-to-have value prop, not just a must-have value prop, but a hair-on-fire value prop, a value proposition that for your customers is going to be a top one or two priority. And so that’s how I kind of frame the H and the Hair On Fire value proposition in hunch.
And then, you know, going through each of the other elements in the acronym, you know, usage being high, it’s not just what customers say they’ll do. It’s not just the downloads and the vanity metrics, but actually using the product day in, day out, week in, week out. As a little sidebar, this week, one of the most extraordinary pieces of research dropped. You may have seen this by one of my colleagues at Harvard with a group of OpenAI researchers where they deconstructed ChatGPT usage. And they stated that they are seeing over 700 million weekly average users for chatGPT. 10% of the planet is using ChatGPT on a weekly basis. So, I mean, that’s a stunning number on so many levels. But it really points to the fact of usage.
It’s not just did you download or experiment or play with or try to do a poem for a birthday party or anniversary, but are you using it week in, week out. So that’s the U usage. N is NPS score, which we could talk about more Net Promoter Score. It’s all about customer love and having a strong Net Promoter Score and strong customer love. The C is churn, having low churn. So, usage patterns and then also seeing that usage endure over time and having these cohorts continue to endure over time. And then the final H is high unit economics, you know, strong Lifetime Value versus Customer Acquisition Cost. Those are all timeless. If you don’t have good unit economics, if you don’t have good continued, you know, cohort behavior, good continued customer love, good continued usage and good continued value proposition.
It doesn’t matter how clever your AI tool is, you’re not going to build a great business.
Jim O’Shaughnessy: Yeah. And as a former quant who used algorithmic methods to select securities, people often would say to me, hey, aren’t you kind of straitjacketing yourself by having this highly articulated algorithmic framework? You know, Jim, aren’t you maybe optimizing local hills and missing the big mountain over here? How would you answer that critique?
Jeff Bussgang: Well, I would say that the thing about product market fit is that, and I talk about this in the book, there are these stages. There’s, you know, the nascent product market fit when you’re not really sure if you’ve got the right thing and you’re looking for early signs and you have a hunch that maybe you’ve got it and some of the markers might point you in the right direction all the way through radical product market fit, where it’s just absolutely ripping. And what I would say is that, yes, you want to have a global view, but you have to have that global view informed with data and you have to make sure you’re measuring the right things. There’s a long history of startup founders being overconfident in their value prop.
Nice-to-have versus must-have versus hair-on-fire and falling prey to vanity metrics, downloads and clicks and, you know, fake ARR and doing all these, you know, trials and everything else. When the real rubber meets the road, you want to make sure you’ve got something that’s enduring and that’s creating, you know, a profitable business model. And, and, you know, shame on us at Harvard Business School. We actually care about enduring, profitable business models.
Jim O’Shaughnessy: So do I. How weird that there’s two of us.
Jeff Bussgang: There’s, there’s a, you know, there’s a great thing about in, in a, in a private market transaction, it only takes one fool. The public markets are the ultimate weighing machine, as Benjamin Graham said. And over the long run, the public markets are ruthless. And we’ve seen time and time again even hot companies go public. And, and I’ve been an executive at one of those that once it went public, the ruthless weighing machine just really puts a, shines a very bright light on the quality of your business model.
Jim O’Shaughnessy: You know, that is something near and dear to my heart because as you might know, most of my career was in public markets. We started in private markets maybe 2005, 2006, and I kept looking for that. I kept looking for how do I create the ruthlessness that I am used to in public markets and port it over to private markets. Have you got any insights for me?
Jeff Bussgang: I’ll try to answer the question, but I’ll tell you a story first. I was an executive at a Internet 1.0 company called Open Market. We went public in May of 1996 and I was one of the executive team members. And at a very young age, I was in my late 20s at the time and the CEO sent me to a conference at Goldman Sachs to talk about the company and pump up the stock. And I remember this is now 1997. At this conference, I do a briefing and I go to a breakout session and I meet with some of the analysts and they ask me all these tough questions in the breakout session.
And I go back home that night from New York, fly home to Boston, go online late at night and I see on the Yahoo message board the transcript of my remarks and the analysis of that transcript. And this is 1997. Okay. This is pre-AI, pre recorded, you know, AI recordings. And, and I realized that they didn’t give a crap about my story. All they cared about were my numbers. That’s all that mattered. And all they were doing was reporting the data and the numbers and feeding that into their models. So, to answer your question, now as a venture capitalist sitting on many private boards, I do care about the story, I care about the strategy, I care about the people, naturally, of course.
But I also try to put the mirror in front of my teams and say all that said, you know, we missed the quarter and we said we’re going to do this and we did that and if we were a public company da da da. So that’s really the, you know, way that I try to combine the nuance of private companies, strategic thinking and long-term thinking and equity value creation with the short termism and the ruthlessness and analytical, you know, kind of clear mindedness of the public markets.
Jim O’Shaughnessy: Yeah, I think that’s a great idea. And could it also potentially be a wonderful tell for you if when you bring data to them they get all hand wavy and they’re like, yeah, but Jeff, you got to think of this this. And then at least in my case when I do similar things, I smell bullshit. Is the same true for you?
Jeff Bussgang: Yeah, I mean, look, venture capitalists are professional bullshit detector machines. And at the same time, we do want to work with founders who are visionary, who are creative and who are a little bit out there. And you know, the phrase we use at Flybridge is head-in-the-clouds, feet-on-the-ground. I love that it’s okay to have founders who are a little creative and visionary and have a hypothesis about the future, but you also want them to have their feet on the ground and be very execution-minded and results-oriented.
Jim O’Shaughnessy: Yeah. Well, let’s design a minimum viable thesis sample, time-box it and then tell our audience and those listening in and watching what would the kill criteria be in this? We’re just going to make a small little experiment.
Jeff Bussgang: Sure. So, let’s say you have an idea and let’s say you have a hypothesis that a certain customer segment is underserved. So, give me, Jim, give me a customer segment you want to focus on. Just as an example.
Jim O’Shaughnessy: Okay. Let’s focus on the people who are absolutely frustrated that they are going to miss the AI boat entirely and let’s create a product that makes them feel like, oh my God, I’m going to become, they’re not going to pay me. I’m going to be an ambassador for this because all of my friends are in a similar boat. They don’t get AI at all. And I now get it. And so, I’m going to tell everybody I know that they need to go to this company.
Jeff Bussgang: Amazing. So what you’re touching on is there’s a huge information gap, education gap, and maybe there’s a way of addressing that education gap and maybe there’s a way of empowering individuals who can cross that education chasm to be actually make money and be productive. So, then the next thing you do is, okay, let’s really lean into who is our ideal customer profile or the ICP. And I generally tend to believe you should be more narrow to start while still keeping the big vision in mind. But to initially start, you want to go narrower than you even think you may want to. So, you may talk about this audience. Let’s go more narrow. What’s the demographic or the profile of these individuals that you want to reach and that you want to help educate on the age of AI?
Jim O’Shaughnessy: Let’s make them my former, I guess, competitors on Wall Street. Let’s make them asset managers who’ve been doing things a certain way. Listen to a lot of stories. Don’t. Are not. Yeah. As deeply, you know, the ones who always called me Poindexter for being such a data guy. Let’s, let’s focus on them.
Jeff Bussgang: I love that. And that’s a quite narrow group. Maybe we would bound it by age group maybe because perhaps the, if I may say it as a 56.
Jim O’Shaughnessy: Yeah, yeah, we olds, let’s focus on anyone above 55, young at heart.
Jeff Bussgang: So yes, you know, folks in, you know, 50s, 60s who are of that profile but still very active, still have a lot of assets still perhaps in the game, in the arena. I like the focus on asset managers. You can imagine what that looks like and then you might say, okay, normally a startup might say, okay, I’ve got my hypothesis of my value prop, I’ve got my ICP, my ideal customer profile, let me go do some customer discovery. But what a modern AI founder might do is they say, you know what, let me just create an AI persona to do the customer discovery first before I run around Wall Street and try to interview a bunch of folks who are very busy.
Let me feed a notebook LM the Google tool, or let me feed ChatGPT, or let me create some artefacts and anthropic and let me build an AI Persona for my ideal customer profile. And then once you’ve built that persona and fed it various documents and maybe even shared a few archetypes with it, then interviews with those archetypes so it can get inside its head, you can throw some YouTube links in there and what have you, then you can begin to query that AI customer and say, hey, look, you know, would you like this? Would this matter to you or would that matter to you? What would be a real hair-on-fire issue that you have? How much would you pay? What’s your willingness to pay for something like that? What would that product look like?
And then you envision the product and then again, the normal way might be you got to go hire an engineer and you got to go build that thing and go away for a couple months and then come back with a prototype that you would then show these customer prospects in your ICP. In the AI era, you would fire up Lovable or Replit or some of these other vibe coding tools and in an hour or so you would build a prototype. You would use English as the primary programming language of the world. You know, all the hot computer programming languages are now distilling down to English. And you would build a prototype.
And then instead of going and trying to schedule 50 interviews to show people that prototype in your ICP and your customer discovery, you might spin up an AI agent to scour LinkedIn and scour your personal network and the network of three of your friends who are in this cohort and generate hundreds of outbound emails with a link to the prototype, with a pitch. You might A/B test your pitch. And again, you could use the AI to help you with that A/B test and then you could record a video of yourself walking through the prototype, explaining the thing, and then have that be customized 100 times and individualized for each of the individual customers. I just articulated for you what the startup process looks like in the modern AI era. In a day or two, you could do all of those things.
That would have taken months and months beforehand.
Jim O’Shaughnessy: Yeah. And the camera was on you because were talking, but I was smiling throughout because that’s what we do here at OSV. I felt that AI was literally so important that when we launched this company, it was based on everything has to be AI-first. And we have to have these in silica cohorts. We have to have all of that. And what’s interesting is we have other verticals as well. It’s equally applicable over there as well. So, we have a book publishing vertical and guess what? We spin up the archetype reader that we’re trying to hit. We do exactly as you have just outlined. And it’s really amazing how quickly you can say, go, no-go, but let’s cover the kill criteria. So, we’ve done all of that. And thank you, by the way.
That was very elegant and eloquent description for people who might not be as into this stuff as we are. But now let’s have the executioner’s block show up. How do. How do. What happens when we’ve got all of our results? What. What kills our interest investing in that company?
Jeff Bussgang: Well, taking the perspective of an entrepreneur, before I take the perspective of an investor, for an entrepreneur, you have to decide, is this an area I want to dedicate the next 8 to 10 to 12 years of my life? Do I have enough passion for it and do I have the belief and conviction that I’m on to something? You don’t have to get it right. In the 0.1 or the 0.2 version, you can be very early in your journey through what Chris Dixon refers to as the idea maze, and, you know, hitting a dead end and you need to back up and go left and go right.
But as long as you’re enjoying the journey, have conviction that at some point you’re going to get to an outcome that you’re excited about and have passion for serving that, that customer, then you’re going to be, I think you should keep going. One of my entrepreneurs, I, I was saying to her that I was so impressed with her passion for the company and the, and the idea that she was working on. And she said, you know, I actually realized I don’t have so much passion for the product as I have passion for this customer and I really want to solve whatever it is that this customer is struggling with. That’s what I want to solve. And this product is the first instantiation of that. But I have three or four other products that I’m going to keep building on and releasing.
So having that passion for the problem, having the passion for the customer I think is really the thing. If, but if you’re getting signals that you’re not finding product market fit or you’re losing passion or you don’t have conviction that you have something big, then you get in the kill zone, then you say, hey, look, I’m not, I, you know, I shouldn’t pursue this. I should either pivot or pull out and do something completely different. That’s the perspective of the entrepreneur.
Jim O’Shaughnessy: And and now let’s put our VC hats on and we’re getting pitched by the person what with our own tools is going to say, hey, best of luck. But this is not fitting with us and our investment thesis.
Jeff Bussgang: Well, many entrepreneurs out there know that VCs at the early stage lean heavily on three simple criteria. Team TAM and tech. Team is the quality of the team. Do you believe that they have the characteristics and the skills to be an end-to-end successful entrepreneur? And we can unpack that in more detail. TAM being the total available market, are they pursuing a massive market opportunity that can sustain a very large company that makes it worthy of investing millions of dollars in a venture capital style, high growth style market of opportunity? And then tech, do they have a unique technology, a unique earned secret, a unique angle on what they’re building and what they’re creating that is defensible? And I’ll say on tech just at the end here, technical moats are harder and harder, as I was saying earlier.
And so we are all thinking in many ways of, I think team is being emphasized even more than ever before. And we’re all looking at the velocity of execution of those teams. As I say in the book, AI is not going to replace founders, but founders who use AI effectively are surely going to replace founders who don’t. And we’re all looking for teams that can use AI to execute at a very high velocity because in the early days your tech may have some advantage, but ultimately technology is super commoditized. In the age of AI, it’s really team, brand, distribution, execution. And as you said in the very beginning, ultimately the operating system that you create for your company, that’s going to yield the differentiated advantage. And so investors are looking at and evaluating teams as to whether they can create those things.
Jim O’Shaughnessy: I got asked by a friend, I was chatting with him about some of the startups that were looking at and he just said, we had lunch and I’ve known him for years. And he goes, yeah, Jim, but you got, you guys don’t have a Rick Rubin. Who’s your Rick Rubin? And what do you think about that? What do you think about. I have a specific point of view that, yes, I’m a data guy, but when you look at this, when you see the same patterns time and time again, you get sort of an imbued or saturated intuition. That certainly happened in my case.
Often frustrating because if the quant model, like for example, during the great financial crisis, I was still at Bear Stearns and I would wander around saying to any other SMD who would listen to me, short your house if you can. But there wasn’t a quant signal for me, so so we didn’t take advantage of that imbued intuition. What’s your strategy for that kind of mix? Is it walking a razor’s edge or is it easier than that?
Jeff Bussgang: It’s a great question and I love the framing of it because you do need the wild creative visionary thinkers out there, the disruptors, as Apple famously put in their ad campaign, those who think different. And the way we talk about that internally is does the individual that we’re in dialogue with, do they have an earned secret? Do they have a non-obvious, non-consensus view of how the future will play out? And if we also have the same hypothesis about the future, you know, do we believe they have the execution chops to pull it off? And there’s a fine line between out there creative visionary thinking and people who are just bananas and so, and so maybe the head-in-the-clouds, feet-on-the-ground, you know, methodology is still, or framing is still worth kind of exploring.
There are many founders we have backed to have just had their heads in the cloud and they couldn’t execute their way out of a paper bag. And there are some founders who have been so tactical that they missed major shifts in the market and major opportunities and may have built beautiful strategic plans and beautiful pipelines and hired, you know, reasonably well and built solid products, but they didn’t create a huge amount of equity value. So, I do think that injecting a little creativity and a little wildness is good. And in fact, we often prefer to back teams, not individuals because we think this alchemy of, you know, one really creative visionary founder and one more grounded, execution-oriented co-founder can be a really good mix.
Jim O’Shaughnessy: Do you find one of the things that I’ve played around with and we…and we…it’s not to a level where I find it satisfactory yet but as you say, since we’re also looking at preseed and seed, you know there is gold in them. There are tales, but there are also a lot of crazy people. And so, we have or attempting to build an adversarial AI that will often record the sessions that we have with the potential founders or the teams that gets transcribed and then we put it in the adversarial AI and then we bring what we think are the most interesting questions that it comes up with back to the team and are trying to develop signal there. Are you, are you doing anything similar?
Jeff Bussgang: We are. And I love that framing. We do have red team work that we do. We have a what we call an AI memo generate fees, investment memo generator. In fact, we made it public, the Flybridge investment memo generator. If you Google that you’ll find it. And we tell founders before you come in to pitch us, run the generator and see what it is that we’re going to say about your idea and your business plan. It’s like the Tom Cruise movie The Minority Report where the precogs identify crime before it occurs so that Tom Cruise can crash through the window of the apartment building and stop the murder. So, there’s this dynamic of like before we give you the.
Here’s why we’re going to pass three things like you know, hear it yourself, run through the memo generator, see those three things and modify your pitch and adjust your plan and think about how you can deal with those objections to mitigate them. So, I really like that I…one of the things we’ve also done is when we write the investment memos, if we always look for non-consensus and we look for like you were saying, the, you know, antagonistic dynamics. And so, we’ll ask the AI to find the holes in the investment memory. Like if this thing, you know, the prompts look like, you know, if this investment were to fail, what would be the reasons for the failure?
And sort of really draws out the risks and then we can look and evaluate whether we believe those risks are properly balanced with the reward potential.
Jim O’Shaughnessy: Yeah. Again, I’m smiling because I use the same movie to tell my text what I was trying to build. Oddly enough extreme technical people love that movie. But, but I, I have one who was telling me yes, but the problem with this is that we could have, AIs would be much better at this, and I’m like, well then prove that to me, please, by building it. I, I think that there’s just so much alpha in going that way and you know, learning via negativia. I always give the example of Sherlock Holmes I’m a big fan of, or I was when I was a kid of Sherlock Holmes and in the Hounds of Baskerville, the only reason that he deduces that the intruder was known to the family is the dog didn’t bark.
And it’s, I don’t think, unless I’m completely wrong, I don’t think it’s an incredibly intuitive way for people to think. Especially, you know, if they’re big fans. Right. They, they tend to fall into confirmation bias. But as I was thinking about it, you know, we humans are the original confabulators, right? We, we are really good at it, and AI is really good at it as well. And so what happens and how do you prevent the AI model from basically confabulating or fabricating realistic sounding signals in our experiments that really are not their vanity, they end up being vanity metrics that we all are trying to avoid.
Jeff Bussgang: It’s a really important concern. And founders obviously have to put humans in the loop in their major decisions. And you also have to be good about your prompts and making sure that your prompts are attenuated or attuned for these situations. In general, the AI models are too nice. And they are, they try to eager, they’re very eager to please, and they try very hard to make you feel good about yourself. And so one of the things, and I say this in the book, is one of the prompts that I often put at the end of one of the phrases at the end of all my prompts is, you know, be rigorous. I want you to be as negative as possible. My standards of excellence are very high and you won’t hurt my feelings.
You know, I really want the AI to tell me the truth. And one of my other prompts I’ll use, I’ll say, you know, don’t make anything up and only give me, you know, the facts and only give me what you think is the hard, cold truth without trying to make me feel better. So, it’s really important to use language that draws out the most critical feedback from the AI on what you’re doing, whether it’s your business plan or your pitch deck or your customer discovery interviews.
Jim O’Shaughnessy: Yeah. And it’s we just are very simpatico because it also crosses category. So, as I mentioned, we have a book publisher, we have a media arm that looks at podcasts and substacks and everything. And I was trying to demonstrate that to a colleague and who’d written a really wonderful substack, or so he thought. And I said, well, let’s see what the AI thinks about it. And we did that. And guess what? The AI, we have an on prem. We have our own AI lab here. And so, I put it through some of the standard commercial models, and it was like, this is fantastic. What a great thing. And then I put it through one of ours, which basically the prompt is, you are the most intelligent critic and then fill in the blank about what the substack is about.
But you also have a vicious streak and you have personal animosity against this author. Please tear it apart. And we put it in that. And I gotta tell you, Jeff, it was the most brutal thing I’ve ever seen in my life. I put my own stuff in there too, and have it absolutely torn apart. And I just think there’s so much value in that. And it surprises me that just isn’t something that people build right in.
Jeff Bussgang: Yeah.
Jim O’Shaughnessy: Do you have any notion for why they don’t?
Jeff Bussgang: Well, first, I’ll say, broadly speaking, a lot of people treat the AI like they treat a search engine. They give it a sentence, a handful of keywords, and they hope for the answer. And we now live in a world of the more context, the more the better. We’re approaching an infinite context window with these AI systems, meaning you can put millions of words of content to contextualize what you want to come out with. And so prompts and I have a, in my book, which you saw perhaps the at the end, in the appendix, I have an appendix on AI prompt tips and examples tuned for founders. I also have an appendix on startup valuations, which links to your other point about the public markets impacting the private markets.
But in the AI prompt tips, I talk about the formula for a great prompt, which is a lot of context and a lot of clarity with regard to what you want and the tone that you want it in. And I think it’s really important that as part of that, you are very clear about having it be rigorous and tough and critical and harsh. So, I think it’s really important. I, I, I have a, in my HBS class, I have again back on this sort of Minority Report precog theme. One of my assignments for my students is to write an investment thesis. This is in my class where I have a bunch of students who are aspiring venture capitalists. The class is called VC Journey or VCJ. And so, I have created a VCJ Investment Thesis evaluator.
It’s a custom GPT and I trained it on a bunch of excellent investment theses and I gave it a very long prompt that provides context. And the key in the prompt is, you know, using language like rigorous and you know, make sure it’s investable and it’s going to generate a massive financial returns. And, and the GPT wanted to make it societal benefit and impact-oriented. It was like, you know, and friendly and positive. And I said no, no, no, I want it to be critical. I want it to be really critical. And so that’s the, that’s the type of training that you really want to have your GPTs have and you really want to include that in your prompt every time.
Jim O’Shaughnessy: So, so what happens when everybody reads your book, goes to your appendix, puts those prompts in their AIs and like literally how much does that degrade the edge that we, that you currently enjoy as an investor? How, how, when does, when the engines get commoditized. I, my natural instinct, which is probably because of my priors, is we got to have better data, we got to have proprietary data. So first off, what do you think about that in terms of when everyone starts adopting this methodology? Where do we go from there?
Jeff Bussgang: Well, first I’ll say in my 30 years of doing this, I have seen wave after wave after wave, from Internet 1.0 to mobile to cloud to AI. And each wave people said, everybody, you know, water’s getting higher and higher. Everyone’s going to be doing this. There’s a new high water mark. How do we compete? And time and time again, people have executed on incredibly valuable business models and built incredibly valuable franchises. In fact, what’s crazy about the moment we’re in, just as a little sidebar if you’ll allow me.
Jim O’Shaughnessy: Sure.
Jeff Bussgang: It seems like when people say we’re in a win-lose dynamic, I look at the actual data and I say actually we’ve been in a win-win environment for a long time. For example, one of our most successful investments was in an open-source database software company called MongoDB, which was a New York based company. We led the Series A. It’s now a $25-$30 billion market cap company and a real leader in the world of databases. At the time of our investment, our thesis was we’re going to disrupt Oracle and IBM and Microsoft because those were the big database franchises in the SQL era. And Mongo created $30 billion of value in the last 12 to 15 years as a disruptive force in the database industry. But go look at the stock market caps of Oracle, IBM and Microsoft.
They’ve all gone up a ton during that same period. And we all know what happened to Oracle in the last few weeks. Tipping a trillion dollars here in Boston. One of the most successful startups in the last decade or two is HubSpot. And HubSpot is this incredible next gen CRM for the modern blogging era. And when HubSpot got going, everybody said this thing’s going to kill Salesforce and really disrupt the CRM industry. And HubSpot today is a $35-$40 billion market cap company and has done really well. And Salesforce, if you track how it’s done over the 10 years that HubSpot has created, that $40 billion of value, has also done incredibly well and added hundreds of billions of dollars of market cap.
So I’m not sure that we’re using the right language sometimes when we talk about, you know, competitive moats and destroying incumbents and what have you, because if the incumbents keep iterating and innovating as Salesforce has done and Oracle has done and Microsoft has done, they’re going to do great. And it’s also the case that new companies who are creating new market opportunities and powerful products and powerful franchises are also going to do great and create a lot of value. The outcomes just keep getting bigger because the whole tech industry is eating the world. As you know, the Mark Andreessen code software is eating the world. We’ve seen that play out and with the world of AI, we’re seeing it play out in every industry, globally, all at once.
So it’s a really, it’s kind of a, it’s an exciting time at a time when I, I think sometimes people get a little bit too negative and critical about hey, maybe we don’t have any more competitive moats and hey, maybe the incumbents and the Magnificent Seven are crowding out all the market and all the investments, and all the opportunity. I don’t think that’s true at all.
Jim O’Shaughnessy: Yeah, nor do I. Everything that we completely and passionately agree with you that you can still operate in win-win territory. And the old idea, which I think was really a dominant mindset for a long time, was the kind of zero-sum game where for me to win, you’ve got to lose. That is patently untrue. And you’ve just given several really good examples of how that is untrue. But I wonder about the idea of are we ever going to reach a time where we, do you remember the old Pogo cartoon ‘we’ve met the enemy and it’s us’ where we humans end up, because we maybe romanticize human judgment or we feel like, I’m not, I, like, I jokingly, kind of jokingly said to my newest chief of staff, I want you to be my last human chief of staff.
I want you to design an AI Chief of staff. And which he was like, of course I will, because that means I get to go do much more fun things. But do you think that there is a point ever where we, we’ve met the enemy and it’s us?
Jeff Bussgang: Yeah, I, I don’t know, I’m more of an optimist maybe than that. I do love your comment about I want you to be our last chief of staff. We just hired a chief of staff at Flybridge and his mandate is to be our head of AI internally as well as chief of staff, and to spin up agents that we can use more extensively internally. So, I love that you’re doing that as well. And by the way, I recommend any of your listeners, and back to your earlier comment about a lot of your colleagues and friends in your community who feel befuddled by AI and not sure how to approach it.
I recommend they really think about reverse mentoring and hiring or, you know, getting perhaps less experienced people, more junior people who are AI native and AI savvy as an AI Chief of staff or an AI native chief of staff, or even just a reverse mentor that they grab coffee with every now and then and have them show them what are the latest tools and what are they, how are they doing their work and what their workflows look like. One of the things we do in our partnership is we are always sharing the latest tools and gadgets that we’re playing with and the fun little aha moments that we’re having as we play around with Replit or Granola or Task lit or some of the other great platforms that we’ve been enjoying working with. So I, I don’t know about this.
You know, we’ve met the enemy and it is us. I, I think of it more as we’re all, as the AI newsletter superhuman, you know, articulates like we’re all becoming, we have the capacity to become superhuman. We have the capacity to extend our reach in an even more compelling fashion. If, if we walk in with 10 agents working for us into any task, we’re going to be so much more effective. Some of my HBS students and I have been talking about the fact that job interviews are shifting from resumes to portfolios. So, it’s not let me show you my resume. It’s let me show you the portfolio of work that I’ve built and created and the AI agents that will come with me if you hire me into this company.
Jim O’Shaughnessy: Again, smiling because we are doing a lot of the things in a very similar fashion. I’ve long been critical of standard interview techniques because they’re snapshots. And you know, you could have a bad day, you could be the most brilliant hair-on-fire person in the world and you, for whatever reason, you got into a fight with your partner or you got a ticket on the way over, or there’s infinite ways that you can have a bad day. There’s also infinite days ways where somebody who’s maybe not-so-great can just be on fire and be like, yeah, we love this guy. And so I, my preference has always been even pre-AI for a movie, not a snapshot. And I noted that you also use the same analogy and metaphor in your work as well.
So let’s shift for a minute and let’s look at this 10x founder. How can we distinguish him or her promote from a 1x? What, what are the things that you see in the 10Xers that you just clearly are not there in a 1X?
Jeff Bussgang: Well, in addition to mastery of the modern tools which represents table stakes, they also need to have this growth mindset. Carol Dweck from Stanford talks about fixed mindset versus growth mindset. The growth mindset is someone who’s always learning the latest tools and always trying things out and always approaching problems, no matter how hard they are, as challenges to be overcome. And so, we really look for those lifelong learning, growth-mindset-oriented individuals. And then we look for individuals who also have this capacity to marshal resources around them. And in the old days, those resources might be exclusively human capital and financial capital.
Today it’s human capital, financial capital, and yes, AI and you know, agentic capital, the ability to harness AI capabilities, whether it’s agents or, you know, excellent user of these tools to marshal resources to be more effective and more rapid in the execution. You know, again, I said this earlier, velocity is really such an important part of the competitive advantage of founders in the modern era. We also look for the earned secret that founders have in their area. I mentioned this earlier, but sort of the non-obvious insight that others don’t have that they’ve earned through the hard work of customer discovery and research and perhaps living in or having come from that industry before. So those are the types of things that we look for in the 10x founder. Another metaphor we use is clock speed.
And we have two ways that we define clock speed. And this is a term we picked up from Bill Gates from his early writing about the Internet 1.0 days. But you know, there’s one aspect of clock speed, which is their ability to make decisions rapidly and then the other aspect of clock speed which is just raw horsepower, just raw intelligence. And so, we really look for both. There are a lot of very smart people who are, you know, frozen analysis paralysis and they don’t make quick decisions. And there may be some people who make very quick decisions but exhibit very bad judgment because they don’t have strategic insight. And so, we really look for both aspects of clock speed when we evaluate founders.
Jim O’Shaughnessy: One of the things that I’ve been thinking a lot about is a lot of the factors when I was coming up in my career, so I’m 65 and when I was younger, like clock speed, that was a real advantage if you were really good at being able to answer questions quickly with solutions. If you were really good at, I don’t know, but I can find out for you quickly. Very, very strong advantages. Another strong advantage was memory. If you had a particularly good memory that proved to be incredibly advantageous when you were having to think on your feet in front of somebody that you’re trying to, in my case, get them to give you their money to manage. And then also the ability to kind of think about a problem from a variety of points of view was also quite strong.
And now I’m wondering what are going to be the advantages for we humans in the age when a lot of this can be outsourced, right, where the advantage of memory. If I’ve got a zillion grad students in my pocket, maybe not as strong as it was when I was younger. If you were like designing kind of the classic, you’re giving advice to a young smart person, a 21-year-old, they’re in your class, what are you going to tell them that they should concentrate on to be, you know, the really switched on player going forward?
Jeff Bussgang: It’s a great question and there are a lot of elements to it. In general, my belief is that we are in an era where judgment, taste, discernment, strategic thinking and insight are uniquely human. And those are the things that are going to really thrive if people have those qualities. And so, you ask yourself, you have to ask yourself, okay, well, how does one develop judgment and insight? And it used to be in any field, you would do that from the bottom up. And you would do the hard work on a first principles basis to learn the fundamentals at the bare metal level of how to evaluate a stock in your world or build a piece of software or deliver a certain quality marketing report in market research analysis.
And now if the AI is doing it for our young people, how are they actually going to know what excellent looks like? And so really being good at discernment and taste and judgment, I think is going to be really important. And for young people, how to develop that. I think it’s a moment where it’s like the Revenge of the Liberal Arts, meaning, like, go read Shakespeare and go read Homer and see the best movies in the world and, you know, watch the best TV shows and be strong at interpersonal skills and leadership skills and communication skills and really understand human motivation and understand what excellence looks like, and understand taste and study design and study art, because the technical skills are all going to just be there at our fingertips for all the reasons you said, Jim.
But what won’t be there is the discernment and the end and the taste and the strategic thinking. And that’s what we have to bring. And we bring that through this very nuanced, I think, amalgam of all of these experiences and judgments and really human instinct about what quality looks like. There’s this book, and I actually saw one of your previous guests reference this book, Zen and the Art of Motorcycle Maintenance, one of my favorite, and I know you like that book and I like that book. And I was rereading it because my son drove cross country a few years ago and we decided we’d do a father son book club. And I said to him, you should read this book as part of your journey. And I, and I reread it.
I first read it when I was myself 18, 19 years old, and then reading it later in my 50s, I, I found this passage that Robert Pirsig, the author, has where he talks about the experimentation method and the flaws in the experimentation method. And I don’t know if you know this quote, but the quote is just exactly on point here, which is to say “the experimentation method teaches you to run all these experiments, but what it doesn’t tell you is what experiments should you run? How do you choose the actual experiments to run? And how do you decide in this world of infinite experimentation what the best experiments are and what the best results are? And that’s really creativity and really human instinct.” And I think that’s really where we’re going to be focused on in the years ahead.
Jim O’Shaughnessy: Again, I wish you would say something that I disagree with because Bob Pirsig’s Zen in the Art of Motorcycle Maintenance is one of the foundational books that I try to reread at least every five years, because I just keep as it’s really interesting. It’s like Heraclitus’s famous maxim, the same man cannot step in the same river twice, right? Because the man is different and the river is different. And I find that book an excellent example of that. Every time I reread it, I’m like, how did I miss this last time I was going through it? What, what are some exper…examples of bad experiments you still see very smart, very switched on, people running. And how do you tell them about that and then say, maybe you should try it this way?
Jeff Bussgang: Well, the thing about running experiments is if you have only a certain amount of bandwidth to run experiments and evaluate the results, even though AI has expanded that bandwidth, the thing is, you have to decide running experiments that will unlock the most important parts of the business model. And so don’t run an experiment that is an obvious area or perhaps an area that investors or employees or partners view to be peripheral. You have to go after the hardest part of the matter, question the most strategic question, and then you have to sequence those experiments to kind of continue to peel away at the model. So, for example, I have a case on a company called ClassPass. And ClassPass, you may know, is a quite well-known company.
And it’s a gym membership subscription service founded in New York City by Payal Kadakia, a very compelling MIT-trained entrepreneur. She was a dancer and believed in fitness and was very passionate about exercise and going to the gym and trying out different classes. And that inspired her to create the Class Pass. But the Class Pass was her third model because she didn’t run the right experiments to test the core value proposition of Model 1 and Model 2. Model 1 was a search engine, and the search engine which she spent months building and launching. She didn’t test whether people would actually search and book online for fitness classes at studios that they didn’t know about and that they had never visited. And so, if she’d only tested that core hypothesis, she would have saved herself a ton of time.
And then her second business model was this idea of a a passport. And the passport was a bundle of passes for gym goers to use to go to the gyms. Well, that worked really well for the gyms, for the consumers, but it didn’t work for the gyms because now people were showing up who had no loyalty and would use these very cheap passes and then go to the next free Groupon like gym opportunity. And so, again, she didn’t test the hypothesis. She spent a lot of time building and didn’t and thought that she knew best because she was the customer. And so, she was overconfident in her ability. And she admits this later. This is not me being dismissive. Payal’s an incredible entrepreneur, but she admits that, you know, she almost killed the company by these failed experiments in sequence.
And so, really focusing on the business model question that matters the most, that will, unlike unlock the most value and unlock future resources from employees and investors, is really the experiment you want to run first.
Jim O’Shaughnessy: Yeah. And as you were explaining, you got a little bit into the next thing I wanted to ask you about, and that’s the question of pricing, right. People, at least a lot of people we talk to still, are kind of seeing premium as catnip. And what is your process for getting the, you know, is it framing bundles, fences, willingness to pay, stack-rank those for me and then tell me what you found to be the most efficacious.
Jeff Bussgang: Well, first, I’ll say a couple things. One is, in general, we believe in focusing on adoption first and monetization and profitable execution of pricing models later. Not that you want to do a ton of freemium forever, but you do want to get people to use the product and iterate on the product, particularly if you have some network effects and economies of scale dynamics. So, starting low when you have a 0.9 quality product and adding value over time with a 1.0 and a 2.0 and a 3.0, and having pricing rise commensurate with that is, in general, a good approach. I also think, in general, we’re in an age of experimentation on pricing that we’ve never seen because of the cost of AI, the cost of inference, and the cost of running these models. This isn’t, it’s not free.
There’s a whole riff right now in the, in the world that I live in that gross margins are compromised dramatically because people are giving away these unlimited capabilities, but they’re having these very finite and continuing costs related these variable costs from running the models and running the inference as compared to the fixed revenue on these unlimited plans. That strategy almost killed ClassPass. When they finally did get the model right, she launched an unlimited plan subscription model and it attracted the power users and the power users went to the gym 20, 30, 40 times a month and ClassPass was paying for every one of those the gyms and had this fixed revenue and this variable cost and it almost killed the company. And we’re seeing that play out time and time again in the age of AI.
So I, I do think, you know, first principles monetization maybe come second adoption-first first principles, you want to try to match revenue and pricing to costs and not get stuck in this, you know, ClassPass trap of the unlimited model. And then finally I think you want to make pricing simple for your customers in a way that matches how they buy. Many, many people lose the fact that variable pricing doesn’t work for companies as an example because then they can’t budget for it. So having tiers and you know, really understanding how they budget and how they spend and understanding where the break points are on the tiers where it can be put on a credit card versus a department approval versus a procurement approval, really understanding that model also helps a great deal.
Jim O’Shaughnessy: And what do you tell founders who are, you know, all in on product-led growth when you what what they see as a flywheel, you see as a mirage. Give me a couple examples of that.
Jeff Bussgang: I wrote an article with my former student and friend Oliver J. who runs international for OpenAI and was previously the Chief Revenue Officer at Asana. And it’s an article in Harvard Business Review that’s titled Falling into the PLG Trap and people start with PLG frequently and that works extremely effectively and it’s a wonderful way to bottoms-up get inside to an enterprise. And that was executed incredibly well at MongoDB. It was executed incredibly well at HubSpot just to pick on the two pull through the two examples that I mentioned earlier. But eventually the PLG trap is you want to sell to enterprises enterprise-wide deals once they already are using your product in three or four departments, you want to then bring that capability of building these large enterprise-wide accounts and getting six figure and seven figure value contracts in place.
The problem is you may not be enterprise-ready, your product may not be enterprise-ready, your organization may not be enterprise-ready, you may not have the on the product side, the SOC2 compliance and multi homing database, you know, constructs and various privacy and security and data, you know, integrity and accountability and transparency and explainability that you need. And your organization may not be ready. You don’t know how to go through a procurement process. You don’t have enterprise salespeople, you don’t have solutions architects, you don’t have a customer success organization. And so, people think it’s very easy to shift from PLG to SLG just by changing pricing and by saying, hey, before you used to buy this thing on a seat by seat, one off basis, now let’s just do an enterprise deal and we’re done.
What they don’t realize is that you, that transition from product led growth to sales led growth is an organizational transition and a cultural transition and a product and technology transition. So, if you want to avoid the PLG trap and break out from not only that 1 to 10 million and 10 to 100 million, but get to as MongoDB now has 2 billion of revenue, you’ve got to really go through that entire cultural, organizational and operational transition.
Jim O’Shaughnessy: Yeah. And that is something I learned firsthand at my first company where we had a very hot product. And then it dawned on me as I looked around the asset management industry that hot products are great while they’re hot and that most investment services at least, and I would extend it, I’d be interested in your view, but I would extend it to many things in the world. Most things are sold, not bought. And you know, when that little light bulb went off in my head, luckily I, I changed the organization rather dramatically. And I, and I think that the like is that, is that a demerit against one of your portfolio companies if they’re not like seeing that as plain as day or do, is it easy to miss that?
Jeff Bussgang: Well, look, commodities are bought, but we don’t operate in a world of commodities. We, we all operate in a world of very nuanced, very, you know, deeply customized solutions. Even an out-of-the-box app that you may use as one of your favorite apps, whether it’s Waze or Granola or, you know, we have a portfolio company called Bold Voice, which is this really cool English accent coaching app that’s absolutely taking off. You know, even these apps, they’re highly, precisely tuned for individual problems and they’re all very particular. And so yes, I, I believe they’re sold, not bought. And I believe the best founders, even technical founders, have to have a notion of commercialization and sales and marketing. They have to have a point of view. We invest in many technical products and we invest in many technical founders.
In fact, Flybridge is known to invest in developer tools and in software infrastructure and cyber security. But our best technical founders are amazing salespeople and they’re amazing at building commercial operations.
Jim O’Shaughnessy: Yeah. I have a friend who we go back and forth with, jokes that at the end of the day, everybody’s in sales, so you might as well get good at it.
Jeff Bussgang: One of the most popular, sorry to interrupt you, but one of the most popular books on that point is when I was early in my journey as an entrepreneur and coming out of HBS was, you know, the one thing they don’t teach you at Harvard Business School is, you know, how to sell. And and that was true in the 1990s when I was there. Today we have an entrepreneurial sales class that’s one of our most popular classes that hundreds and hundreds of our students take that class to learn how to sell.
Jim O’Shaughnessy: Yeah. Very, very important. Do you make them watch Glengarry Glen Ross?
Jeff Bussgang: That is a classic movie. Another wonderful classic movie reference.
Jim O’Shaughnessy: First prize, catalog.
Jeff Bussgang: Second prize, pair of steak knives. Third prize, you’re fired.
Jim O’Shaughnessy: You’re fired. So many great lines from that movie. As we’re, as we’re on kind of going through the scale here, I was wondering about like, the, the Net Promoter Score. Right. Isn’t that like to be devil’s advocate, isn’t that kind of noisy and easy to game?
Jeff Bussgang: Well, the general concept is customer love, and there may be multiple ways of measuring customer love. Net Promoter Score is a very smooth industry standard way, and it’s a way that you can compare against other companies. In our experience, if it’s done well, it’s so simple to execute on that the simplicity, it kind of pierces through the gamesmanship. And so it may be that it gets a little bit inflated if you know, you’re selective and who you reach out to and your sample size isn’t perfect and all that. But really what you’re looking for is a baseline and a sense, and then you use that baseline to improve over time. We have a few of our portfolio companies that…I have one e commerce company in particular, they measure Net Promoter Score and show it to the board every month.
And it’s a really key performance indicator for them. And we really love seeing it. And they’re very honest about when it goes down, they jump on it and they say, well, why’d it go down? And you know their case, this is a company called MadeiraMadeira. It’s a unicorn e-commerce platform in Brazil. It’s the Wayfarer of Brazil. They do bulky items like furniture and wardrobes and home furnishings and goods. And they are obsessed with customer love and they’re obsessed with their net promoter score because they know it tells them that shipping isn’t going well or that the quality of the product isn’t quite what they wanted or that the pricing isn’t what the customer expected. And the lagging indicator is churn. You know, churn is the lagging indicator. The leading indicator is net promoter score and usage. Because once they churn, they’re gone.
And they’ve already decided they’re not going to work with you anymore. They’re not going to buy from you anymore. They don’t want to, they don’t want to associate with you anymore. What you want to do is you want to get ahead of that early and have those leading indicators like NPS.
Jim O’Shaughnessy: Yeah. What…just as I was listening to you, I was wondering, the thought occurred to me, what’s a category that like you would invest in tomorrow because it meets a need that you have not yet found anything out there that fulfills that need, maybe even one of your own personal needs?
Jeff Bussgang: Well, that’s a great question, Jim. I will say I’m really interested in email as a category. We made one investment in that space. It’s a product called Tasklet. There’s another company in that space, Superhuman, that was recently merged with Grammarly. That’s got some momentum. But as we think about the proliferation of email and all the messaging apps and you know, having some sort of unified communication that’s AI native, that allows us to be more efficient in the processing of information and inbound. I think that’s really interesting. Just personally, I would love to have more sort of control over my inbox and having an agent that helped me with that. You mentioned the AI chief of staff. There’s a lot of energy around that. I think it’s very cool.
But, but at Flybridge, we’re also spending a lot of time thinking about what the workforce of the future is going to look like. What are these very service intensive, fragmented businesses going to look like in the age of AI? We’ve invested in some services-oriented companies that are AI native and AI enabled and we’re looking at another one right now. So, I, I think there’s a lot of interesting dynamics at this intersection of technology and services. There’s a joke I heard somebody say where they said, you know, I don’t know who’s going to win at this inflection point. Is it going to be the VCs who are bringing AI into the services industry?
Or is it going to be the PE firms that are, you know, kind of bring, you know, used to bringing capital and are now bringing, you know, capital and management efficiency and view AI as just another one of those things, you know. So, the AI, the VC firms know AI, but they don’t know services and capital. The PE firms know services and capital, but they don’t know AI. You know who’s going to win that battle?
Jim O’Shaughnessy: Yeah, who do you think?
Jeff Bussgang: I, I have no idea. Like I said before with regard to Oracle and MongoDB, maybe everybody will win. Who knows?
Jim O’Shaughnessy: Right. Right. Well, and also as I was listening, we’ve talked about the 10X founder. What are we looking for when we’re looking for a 10X joiner?
Jeff Bussgang: First, I’ll say so the idea of a joiner, which is a term I, I think I’ve invented, but others maybe used it as well in the last 10 years in my book. Entering Startup Land is a book that’s focused on joiners. Those are the folks that join startup as employees number three through 300. I was a joiner in my first job out of business school. The public company, the company that went public that I mentioned, the rocket ship. And so, what we’re all looking for are 10X joiners who can, like the 10X founder, leverage the modern AI tools to be an incredibly effective head of marketing, product manager, head of engineering, head of customer success. And instead of looking to bring their team with them from their previous company, they bring their AI agents with them.
In today’s era, we are seeing organizations of 20 or 30 or 40 execute as if they were organizations of 200. The company Bold Voice, I mentioned the Accent coaching app, and this is public information. They, they have announced this. They’re part of a very special club, the Tiny Teams club. Over a million of ARR per employee. They’ve crossed over 10 million of ARR with seven employees. And you can bet those individual joiners who have joined the company, the head of growth, some of the various engineers, they’ve come in with this incredible skill of leveraging AI to extend their reach.
Jim O’Shaughnessy: Yeah, and that is something that I have specific experience with because that’s kind of the way we run OSV. And what I wanted to ask you was what does the ideal human team look like when the company is treating AI as a true and real partner? That’s what we do. We do it in, like across all of our verticals. I’ll give you an example I’m writing for the first time, a fictional thriller. I’ve written four books, all nonfiction. Much like Dorothy Parker, I hate writing and love having written. And my daughter, who’s a Newbery Awardee for middle grade fiction, was here and I looked at her and I said, I had no idea how hard it is what you do.
But we’re taking a somewhat different approach in that we have a writer’s room helping me write this book, and in that writer’s rooms are human writers and editors, but also AI writers and editors. And there’s been some kerfuffles with my human team saying, like, did the AI suggest this? Is this your turn? So I, I’ve, I’m in the process of learning that I’ve got to develop a new skill set in not only dealing with team member queries like that, but also the question I have for you is, does that lead us to a different type of human to put on that team?
Jeff Bussgang: You’re answering your own question, and I think your example is an excellent one. It’s a cultural shift where the humans need to be very comfortable directing the AI agents, in some cases overruling the AI agents, putting the guardrails up for the AI agents. We’re all going to operate in organizations where there are more AI agents working for us than humans. That’s just a fact. It’s going to be the case across business, across government, and across nonprofits. And so, the humans in those organizations need to be more and more comfortable orchestrating, managing, controlling, overruling, or being overruled by the AI agents. And they need to sign up for that.
Jim O’Shaughnessy: Exactly my thought on it, and what I’m still struggling with is like, how do you have any heuristics or little tricks that you could give me when you’re talking about bringing a new human on? That have worked really really well for you in determining, yeah, this person is going to get, is already incredibly comfortable with the idea of having this AI as a true partner.
Jeff Bussgang: I mentioned to you earlier that at HBS, we’re talking about resumes, moving to portfolios, in the job interview process, asking questions like, what have you built with AI? What are your favorite prompt techniques? What’s a fun use case in your personal life? What’s an app that you wish you could create, and why haven’t you created it yet? Those types of questions in the interview process I think will really help bring that to life.
Jim O’Shaughnessy: Yeah, that.
Jeff Bussgang: One of the questions I ask frequently when I speak to business audiences is on a scale of 1 to 10, how would you rate yourself where 1 is an AI newbie and 10 is an AI native? And then on a scale of 1 to 10, how would you rate your organization? And then you. When people identify how they rate themselves in the organization, you then begin to say, well, why aren’t you a 10? And why isn’t the organization a 10? What’s holding it back? And when you find the eights and the nines and the tens in the audience, you say, well, what are you doing? Give me some examples. And people’s eyes light up when they hear what their friends and colleagues are doing. So, it’s a really, I think, important thing is to draw out the AI native in all of us.
Jim O’Shaughnessy: Yeah, that’s good advice. So, Jeff, this has been absolutely wonderful. Generally speaking, I very rarely come across somebody who’s like, viewing the world very similarly to the way I view the world. And generally speaking, you don’t want to just have a me too. But you’ve, you’ve given me excellent answers to things I…that I hadn’t thought about. So I’ve learned a lot, and I want to thank you for that. On Infinite Loops, we…we have a, a final question for all of our guests, which is this.
We’re going to wave a wand and we’re going to make you the emperor of the world. Rules are you can’t kill anyone. You can’t put anyone in a re-education camp. But what you can do is we’re going to hand you a magical microphone and you can say two things into it, and the entire population of Earth is going to wake up whenever their next morning is and they’re going to say, you know what? I’ve just had two of the greatest ideas, and unlike all the other times, this time I’m actually going to start acting on them right now. What two ideas are you going to incept in the world’s population?
Jeff Bussgang: I would say the golden rule. Do unto others as you would have them do unto you. And the second idea I would have is be kind. Just be kind to each other. I think we’re entering into an era of tremendous disruption, of tremendous uncertainty. Yes, they’re going to be haves and have-nots and arguably is the wealth grows more rapidly in these because of these turbocharged technologies. We’re going to have greater income disparity. And I just think people need to have grace and be kind.
Jim O’Shaughnessy: I love both of those. Those are simple but not easy and like a lot of things in life. Jeff, where can our viewers and listeners find they can get your books at anywhere. Your website we will have in the show notes, but do you have any personal website that they might go to or any other sites that you would want to mention?
Jeff Bussgang: They can go to Flybridge and see what we’re investing in flybridge.com they can follow me on LinkedIn and certainly, yeah, go buy the books and sign up for the blog that I write, which is off of the LinkedIn newsletter.
Jim O’Shaughnessy: Perfect. Jeff, this has been delightful. Thanks so much for your time.
Jeff Bussgang: Thank you, Jim. Great to be here.
