Bending Spoons is Eating Silicon Valley | CEO Luca Ferrari

Luca Ferrari, Co-Founder & CEO of Bending Spoons, sits down with Molly O’Shea at the company’s Milan headquarters for Part II of our Bending Spoons series, going deeper into the technology, culture and operating philosophy behind the company. Luca takes us inside Bending Spoons’ centralized technology platform, including 50+ proprietary tools, 95% of code being written by AI, its internal Alt Spooner agents, and why he estimates the company is now 2–3X more productive. We also discuss the company’s extreme approach to talent, with 800,000 job applications in 2025 and fewer than 300 hires, alongside a culture built around extreme ownership, experimentation and truth-seeking. We get into why Luca believes most people don't actually seek the truth, why Bending Spoons ran more than 3,000 experiments last year, the misconceptions around its business model, AI hype and valuations, the risks of underestimating increasingly capable AI systems, and how Luca thinks about building Bending Spoons for the decades ahead. Luca Ferrari: https://x.com/luke10ferrari Molly O’Shea: https://x.com/MollySOShea Sourcery: ⁠https://x.com/sourceryy 𝐄𝐏𝐈𝐒𝐎𝐃𝐄 𝐋𝐈𝐍𝐊 YouTube: https://youtu.be/mCN1e_03I-c 𝐒𝐏𝐎𝐍𝐒𝐎𝐑𝐒 • Brex https://brex.com/sourcery—The modern finance platform, combining the world’s smartest corporate card with integrated expense management, banking, bill pay, & travel. https://brex.com/sourcery • Zone https://zonefrontier.com/—develops next-generation data center campuses, partnering with AI companies, site developers and technology leaders to bring compute online faster and at scale. Visit: https://zonefrontier.com https://zonefrontier.com/ • Turing https://turing.com/sourcery—Turing delivers top-tier talent, data, and tools to help AI labs improve model performance—and enables enterprises to turn those models into powerful, production-ready systems. https://turing.com/sourcery • VCX https://fundrise.com/vcx—VCX is the public ticker for private tech, allowing investors of all sizes to invest in venture capital. View The Portfolio at http://GetVCX.com http://getvcx.com • Deel https://www.deel.com/sourcery—Deel is the global people platform that helps startups hire, manage, pay, and equip anyone, anywhere. Trusted by more than 35,000 fast-growing companies, Deel is the people platform that just works, so teams can scale without the chaos. Visit: https://www.deel.com/sourcery • Public http://public.com/sourcery–Investing platform Public just launched Generated Assets https://public.com/generated-asset, which lets you turn any idea into an investable index with AI. With Generated Assets, you can build, backtest, refine, and invest in any thesis with AI. Gone are the days of one-size-fits-all ETFs. https://public.com/sourcery Follow Sourcery for the latest updates! https://www.sourcery.vc https://www.sourcery.vc/ Disclosure Paid Endorsement. Brokerage services by Open to the Public Investing Inc, member FINRA & SIPC. Advisory services by Public Advisors LLC, SEC-registered adviser. Crypto trading provided by Zero Hash LLC, licensed by the NYSDFS. Generated Assets is an interactive analysis tool by Public Advisors. Output is for informational purposes only and is not an investment recommendation or advice. See disclosures at public.com/disclosures/ga. Matched funds must remain in your account for at least 5 years. Match rate and other terms are subject to change at any time. 𝐓𝐈𝐌𝐄𝐒𝐓𝐀𝐌𝐏𝐒 (00:00) Luca Ferrari, Co-Founder & CEO at Bending Spoons (00:45) The deal that broke the Internet (03:41) Why Silicon Valley overspends on Hype (07:35) Half a billion monthly active users (16:17) What makes a company worth buying (20:07) Inside the platform powering every product (32:16) Debunking the Private Equity comparison (38:23) $4 Million in revenue, per employee (43:44) The secret to zero churn (47:40) The one value that defines Bending Spoons (51:52) The AI tool every Spooner uses (57:58) Going Independent from the AI labs (59:02) Have we actually reached AGI? (1:08:05) The AI question nobody's asking (1:12:54) Luca's next big bet

Speaker names marked ? are inferred from speaking time and the show’s known hosts — they are not voice-verified.

• 0:00Speaker 00

  We have half a billion monthly active users. I mean, that's a lot. We have built over the past decade an operating system of over fifty proprietary technologies. We buy these companies and then we install them on this shared operating system. At Bending Spoons, we have been able to attract, I think, extremely strong talent. We had eight hundred thousand job applications in two thousand and twenty five. We hired fewer than three hundred people.

• 0:18Molly O'Shea?

  You guys operate very efficiently per employee. I think the last metric you mentioned was four million in revenue per employee.

• 0:25Speaker 00

  And this has grown tremendously. It was about a million dollars just two, three years ago.

• 0:29Molly O'Shea?

  Right now, it kind of seems like Bending Spoons is actually eating Silicon Valley. So what's going on there?

• 0:45Molly O'Shea?

  Luca Ferrari, thank you for having me at Bending Spoons.

• 0:49Speaker 00

  Well, thank you for coming.

• 0:51Molly O'Shea?

  We're all the way out in Milan, Italy. And something happened not too long ago that really shook up Silicon Valley. Right now, it kind of seems like Bending Spoons is actually eating Silicon Valley. So what's going on there?

• 1:07Speaker 00

  I don't know that we're eating anything, but I think you're probably referring to the announcement of the acquisition of Airtable, I suppose. Yeah, I guess. I mean, Airtable is a great business, a great product. And you can say of Airtable something you can't say of a lot of different businesses, which is that they really were identified as a high-flying, you know, key company in Silicon Valley for quite a while.

• 1:42Speaker 00

  And that's very difficult to do. So, you know, great job, Howie, and everybody else who built that business. And so I think once the announcement of the acquisition came, that was a lot more news worthy and interesting than maybe some other acquisitions we've done before.

• 1:57Molly O'Shea?

  So how did you pick that one? What was the process?

• 2:02Speaker 00

  Well, you pick each other when it comes to acquisitions because it's something that has to be bi-directional. But we look at a lot of companies.

• 2:11Speaker 00

  I would say actively, we may look at hundreds of companies in any one year. And then we try to establish a dialogue with dozens where we see the best match.

• 2:25Speaker 00

  And if, you know, some of these are interested in selling, then the conversation can progress. We look for businesses where we believe we can bring a lot of value. We are a very active acquirer. And so for us, for us to be absolutely confident we can offer a kick-ass price while delivering strong returns for shareholders, we need to be able to bring a lot of value, whether it's by improving the product or the technology or monetization, the organization, ideally, all of these. So that's a key criterion. Then we look for businesses that we believe we can, you know, whose trajectory we can predict with confidence multiple years into the future.

• 3:08Speaker 00

  Otherwise, it's difficult to underwrite big investments.

• 3:13Speaker 00

  And it's typically been digital businesses. So we've done consumer internet, SaaS. These are the key areas for us. Also, recently, we acquired a hardware company called Tractive in spring, which is very interesting, pet tracking and also help monitoring for pets. The pet industry is booming. So that was a very nice acquisition, a little bit outside of our usual comfort zone.

• 3:41Molly O'Shea?

  Because the Airtable acquisition kind of set SF, Silicon Valley, American Tech into a bit of hysteria, what do you think they get wrong about technology companies or managing them?

• 3:58Speaker 00

  I mean, not a lot, clearly. Silicon Valley has built, I know, seventy percent of the most successful technology businesses of the past twenty-five to fifty years. So I think Silicon Valley gets almost everything right. I believe we have been able to do well acquiring a tiny fraction of the businesses that have come out of Silicon Valley because we bring something different and new to the table in both the way we're structured and the platform we've built, that's for the most part unavailable to these businesses on a standalone basis.

• 4:36Speaker 00

  So for people who don't know Bending Spoons very well,

• 4:40Speaker 00

  unlike most serial acquirers out there, most serial acquirers out there will either buy a business because they think it's basically good as it is and they can't really improve it, but you know, the price is low enough that they can get good returns and then they'll basically leave it alone. And that model can work. I don't think it will ever give you exceptional returns, but it can deliver reliable, potentially appealing returns if you're very good at picking businesses that are slightly undervalued. Others are more active, but they will still keep these businesses separate as they were before. But, you know, these acquirers may have an opinion on how to price the product or how to build the org, and so they will go in and make changes. We are even more extreme on the being active end of the spectrum.

• 5:26Speaker 00

  And we integrate all of these businesses together quite tightly. So we have built over the past decade an operating system of over fifty proprietary technologies to take care of almost everything you need to run a digital business, whether it's, you know, data storage and processing, A/B testing, payment management, the, you know, everything related to recruiting,

• 5:50Speaker 00

  credentials management,

• 5:53Speaker 00

  the orchestration for all the I models used in operations, and so on and so forth. And so we buy these companies and then we, it's almost like we install them on this shared operating system so they can be much more efficient. And also, we have a core team for R and D, marketing. At this point, approaching a thousand people. We can deploy very fluidly and rapidly across these various businesses to go after R and D opportunities. And when the R and D opportunities are not as exciting any longer, we can take out the talent and move it elsewhere. So we stay very efficient. So these are aspects that I believe none of these, none of the teams running these companies

• 6:39Speaker 00

  on a standalone basis really could access.

• 6:43Speaker 00

  And so it's not that a lot of the value we create is because other people are myopic. They're doing the best they can with the resources available to them. But I do think we bring also something extra in terms of culture. We have developed a culture of extreme rationality, almost a scientific approach to running businesses whereby we are not afraid of making unpopular choices when we believe it's for the benefit of the business in the long run. So we tend to run these businesses very leanly, using data extensively. Sometimes we joke about taking more established companies and bringing them back to startup mode, small team, very talented teams,

• 7:23Speaker 00

  removing red tape, giving these people plenty of room to maneuver, to experiment, to move fast. And we have found that that generally delivers a lot of value for customers and for the business.

• 7:35Molly O'Shea?

  So I think one of the big misconceptions with the types of companies that you acquire is that they're graveyard kind of companies. They're old brands, they're not new, maybe they're distressed assets, that kind of thing. What is wrong about that?

• 7:51Speaker 00

  Oh, I think it's just that, I think people just try to frame things in a way that will get clicks. But I'll give you a stat. We have half a billion monthly active users. I mean, that's a lot, short of being, you know, Google or Meta, you know, many company. So if half a billion people using these products every month is a graveyard, then sure, let's call it that.

• 8:16Speaker 00

  These are, they're, often they may not be like the up and coming sexy thing, but it doesn't mean they're not incredibly useful, important. You could see this at its very peak with AOL, where, which, of course, AOL is a quote-unquote "old" brand and company, no doubt about it. It's been around for, what, like forty years, something like that. However, to this day, AOL is used as a, especially as an email inbox, by tens of millions of people. For many, it's their primary email inbox.

• 8:54Speaker 00

  It's the, to the best of our knowledge, it's the fifth most used email provider in the Western world. You could imagine that, you know, the four above it. And so, and at the same time, you will see, if you go to, you know, the online media will find so many, over the past five or ten years, so many email startups, which people who don't actually have the data will assume are far more relevant just because they got more coverage and they sound a lot sexier. But really, if you look at the data, these, in aggregate, these don't probably add up to even five percent of what AOL means in terms of emails sent, received, activity, people who rely on it. So we just, we don't care too much about being cool.

• 9:38Speaker 00

  I'd say we probably don't care at all about being cool. We care about being good at our jobs and creating value. So if we find a business that's slightly perceived as slightly less cool, if anything, that's a good thing for us because it means it's probably also going to be priced a little bit more accessibly. So it is what it is.

• 9:55Molly O'Shea?

  Yeah, so I think I was, I was trying to ask this question earlier, but I might have asked it wrong. Silicon Valley is so tied to funding for growth versus actual business for growth. And so for some of the other email companies you might be talking about, they might not have many users, but they're the hottest, highest-flying, funded-by-every-VC kind of company out there. And even with the Airtable acquisition, right? It kind of broke people's brains and there was a bit of hysteria because that was seen as the golden child and, or at least one of the golden child of the brands in Silicon Valley. And so if that's the exit that they're taking and, you know, we're at a bifurcation with AI, what does that mean for all the other companies out there?

• 10:39Molly O'Shea?

  And we've gone through different kinds of cycles with these tech companies over the years. I think the last one was the reckoning of two thousand and twenty-one, twenty-two. There was a lot of overfunded companies that then...

• 10:54Molly O'Shea?

  Kind of we're zombies, but I guess the question I'm trying to get at is, what are the core characteristics that you look for in acquisitions? And how does that differ from the stereotypical culture of SF?

• 11:10Speaker 00

  So I think there's a lot of, there's a lot to unpack here. There's a number one, there's a lot of, of value in, in that, you know, model that relies on generous funding. Early, many, many amazing companies have come to exist often from Silicon Valley. Because precisely because of that abundant availability of capital. Plenty of companies you could probably not take off the ground at all without that. And, and even once they are well off the ground and, and generating substantial revenue, there's often a very good,

• 11:52Speaker 00

  it's often wise to inject more capital in them so they can grow faster, get to a position of greater market power, where it's scale economies, network economies, brand.

• 12:05Speaker 00

  So again, I think that model overall has been incredibly effective. I believe there's no doubt that if you look at the overall capital that's been deployed in Silicon Valley, by Silicon Valley into technology over the past many decades. And, and the real tangible, non-hyped business value of the companies that came out of that. The ROI is excellent in general. It doesn't mean it's, you know, every investment decision is, is perfect. But so I, I don't think that the fact that Beni Spoons is doing well should, should in any way undermine that model. I think it's a great model. And, and I, and, but, but like everything, especially in, you know, when, when

• 12:49Speaker 00

  there are sometimes perverse incentives involved, there will be cycles of excess. And so, yes, in 2021, I think valuations were out of whack completely. But, you know, when investors are ultimately incentivized by managing as much capital as possible as opposed to actually delivering strong returns and when returns are primarily delivered through exits, where all that matters is the multiple, not actually the cash that the business will generate in the long run. At least not, it's not the primary reason why you get a certain price. Then you will get hype cycles and stuff like that. But I, I think there, overall, when I look at Silicon Valley and, and its investment philosophy over, you know, the past many decades, I would say that's a relatively small

• 13:35Speaker 00

  price to pay for a model that overall has been incredibly successful. Airtable specifically, I think deserves a lot of credit because yes, their valuation was very high in 2021. You know, that's not anybody's fault. If anything, it's mostly, if we agree that that was perhaps excessive, I think most people would. But the business was great, just was too much. And that was not true of Airtable alone, but pretty much every business. If, if anybody made a mistake there was the investor, certainly not the company. As a company, you will try to take capital at the best valuation you can. That's the responsible thing to do for your shareholders. And if anything, Howie and the team were incredibly disciplined. They actually didn't raise all that much money.

• 14:18Speaker 00

  They could have raised more and they, they stayed profitable and burnt a very little of that money, if any. And in fact, if you look at the acquisition price, we, you know, the, the enterprise value was approximately $1.3 billion. But then they still had plenty of that cash on balance sheet. So investors ultimately got back approximately all the money they, they had put in plus, plus more. And the valuation at which Airtable exited was very much in line with the SaaS businesses of comparable quality on the public markets. So they, they got a very reasonable deal, in my view. Obviously I'm biased, but I, I believe that to be true. So there's a lot of good there. I think Airtable

• 15:02Speaker 00

  has done a very good job and, and but yes, the internet will have to, to debate. And, and you know, when, when you go from being perceived as the ultimate winner and, and the poster child of, of success to an exit that would be considered amazing by almost any measure. I mean, over a billion dollars. How many companies have started that ultimately exit at over a billion dollars? We are one in a thousand. I don't know the stats, but it must be very, very few.

• 15:29Molly O'Shea?

  People forget how hard it is to get to a billion dollars.

• 15:32Speaker 00

  It's crazy.

• 15:33Molly O'Shea?

  And $100 billion, let alone this whole trillion dollar company thing is, is really disorienting.

• 15:39Speaker 00

  It is exactly. There's been literally a handful in the history of humanity at that scale. But so a billion dollars plus is unbelievable. It's definitely, and it's achieved based on real economics, plenty of revenue, real customers, real growth, excellent brand. So I really applaud Airtable, Howie and everybody there.

• 15:59Speaker 00

  So I think the business model for Silicon Valley overall makes sense and funding a company early, even at a loss makes sense. But it doesn't mean that things couldn't be done better. I'm sure sometimes there's too much enthusiasm, pouring money into businesses that don't make sense. Or too much money in businesses that do make sense, but should use less.

• 16:17Molly O'Shea?

  So what are the key characteristics that you look for when you're acquiring companies?

• 16:22Speaker 00

  Yeah, so basically,

• 16:25Speaker 00

  the most important thing is that we can predict where a business is going. We, we buy to hold and operate forever, not to sell three or five years down the line. And so we need to feel comfortable with, with our investment, with a long-term view. We prefer businesses that are robust and maybe growing. We're not opposed to buying businesses that are shrinking. We have done that before, but we need to know how much they're shrinking. We need to be able to plot out their trajectory, at least five years, a little more into the future. So that's a non-negotiable. And then I would say the second most important thing is that we need to be convinced that we'll be able to add a lot of value to that business.

• 17:07Speaker 00

  Essentially improve revenue, lower costs, certainly both through technology, product, talent. Otherwise, we are unlikely to be able to offer a price that's appealing to sellers while at the same time

• 17:21Speaker 00

  delivering very high returns for, for ourselves and our shareholders.

• 17:25Molly O'Shea?

  Because there's a proliferation now of all these AI application companies that are dependent on token spend and lots of tokens. They're somewhat, sometimes negative gross margins. Would those at all be of interest for you guys? Like where do you see those companies getting acquired or exiting?

• 17:44Speaker 00

  I think we, we use AI as much as anybody. I, as far as I can tell, probably in the 99th, 99th percentile by aggressive deployment of AI in our operations to improve our products. We've developed a lot of technologies powered by AI internally. So I'm very bullish on AI overall, also concerned, but, but doesn't mean I'm bullish about all or even most of the startups that are coming up. I'm, I'm pretty sure that some of the most valuable companies of all time,

• 18:21Speaker 00

  sustainably valuable companies of all time will, will be coming out of this broader cohort of businesses. Businesses built over the last, let's say, five years with AI at the center of the thesis. But I, I'm equally confident that most of these companies will fail, or at least, you know, basically fade away.

• 18:41Speaker 00

  Because everybody, it's, it's a gold rush. When you see people raising massive, massive amounts of money at billion dollar valuations with pretty much nothing other than any idea, maybe a good track record in academia or elsewhere. Anybody who's half credible, because they were a great student or they did well at a big company,

• 19:03Speaker 00

  was not too worried about their reputation will just run and try to raise money. Because, you know, what do you have to lose other than your credibility and reputation? So a lot of this is just fluff, but there is real substance here and there. For, we have, as many as, right now, we're not considering acquiring any of these companies. I think we, I said earlier, the key criterion for us is being able to predict how things will go in the medium to long run. And it's very difficult to know, not just because some of these businesses are up and coming, growing super fast. And when something is growing 100% a year,

• 19:40Speaker 00

  it's, it's very difficult and you only have one or two years of history. It's very difficult to know whether they'll be growing at 100% in three years or at 12% in three years. And that changes everything. So plus valuations are very high, often irrationally high. So we, we don't think we can compete there and, and deliver good returns for our shareholders.

• 19:59Speaker 00

  You know, but maybe later down the line, when the market is a little bit more mature, we'll, we'll look again and find something interesting.

• 20:07Molly O'Shea?

  Well, I want to go into your centralized platform because you guys, to your point, use AI a lot. It's core to the business. I think 95% of your code is AI.

• 20:18Speaker 00

  It's written by AI, yes.

• 20:19Molly O'Shea?

  So can you walk me through the centralized platform? How you built that out and kind of, I mean, you made various different types of acquisitions from Evernote, AOL, to Vimeo?

• 20:31Speaker 00

  Yeah, so we, because we, you know, we, at, at our core, we try to be the most capable operators of digital business on the, businesses on the planet. And, and, and part of achieving that vision is to, to have access to the best toolkit possible, right? It's almost like if you want to be a great cyclist, obviously that's not all there is to it, but you want to have a great bike. I'm not saying anything shocking here. So we have invested into this operating system, into these technologies for a long time. Because of, you know, it was strategically critical to us also. My co-founders and I are all engineers, so perhaps there is a bit of a passion angle too.

• 21:16Speaker 00

  But so for the past decade, we have tried to develop the best technologies we could. We, you know, we buy from vendors when relevant. We use Slack, for example. We don't need a more sophisticated version of Slack. Slack is a wonderful product, so we buy Slack but from, you know, from them and use it. But a lot of the tools we need, to maximize our potential, we need them to be more sophisticated than almost any other companies out there would need them to be. And providers of business tools are generally optimized for the mass market of enterprises. It makes sense. You don't want to, like, if someone were to build what we need, then they have a market of one or two. I don't know, like, it's not a very appealing market to go after.

• 21:57Speaker 00

  And so most of the tools out there are relatively simple. We need more sophistication, so we have to build it ourselves. And also, we, by building all or most of these tools in-house, we can make them natively integrated with one another, and that creates a lot of efficiencies effectiveness. So everything kind of, like, every tool talks to every other tool as relevant, which is impossible or very difficult to do if you buy from different vendors and then they change something and you have to change everything else.

• 22:32Speaker 00

  And last but not least, we get to save on costs substantially. It's difficult to know exactly much we're saving by building this in-house, but at a wager, it's at least a hundred million dollars a year in costs. So it's pretty significant. But yes, it's a key source of competitive advantage, and this is, building these tools would be uneconomical for pretty much any one of the businesses we acquire as they exist as standalone companies. It would be, it would be too much money to be poured into R and D to build these tools. The returns would be on a excessively long timeframe, whereas we can amortize those investments across the entire portfolio. And also, the portfolio, our expectations of the portfolio expanding in the future. So it's just a scale advantage that's unavailable.

• 23:15Speaker 00

  And another big advantage in building this operating system is that, as all of our businesses use all of these tools, when they find that one of these tools does not serve their needs as it should, perhaps there is a bug, or a certain corner case is not handled, or an entire area of need is not fully covered, that business can improve the tool, almost like if it were an in-house open source community, and then those improvements are propagated to benefit all the other businesses. And so as we expand our portfolio and our organization grows, our ability to make these tools effective expands with it.

• 23:57Molly O'Shea?

  This episode is brought to you by Brex, my favorite. You become what you spend on, and I refuse to spend my time on work that shouldn't exist: expense reports, receipt chasing, and manual closes. The companies building what's next, from Vercel, OpenAI, Anthropic, Granola, and Deepgram, all made the same call: they all run on Brex. Brex is the intelligent finance platform that combines cards, expenses, and banking into a single stack with agentic finance built in. AI agents that handle expenses automatically, enforce policy before spend happens, and close your books in minutes. That's why Sorcery runs on Brex, so I can spend time on building and not busy work.

• 24:42Molly O'Shea?

  It's time to get Brex AF. Learn more at brex.com/sorcery. That's B R E X dot com slash S O U R C E R Y. Bye. Turing is training the next generation of AI with tasks that require real expertise and real-world judgment. That's why companies like Nvidia, Anthropic, Salesforce, and Gemini partner with Turing. Turing builds realistic reinforcement learning environments and data systems based on real operational traces, the kind of infrastructure Frontier Labs need to train superintelligence. Visit turing.com/sourcery.

• 25:24Speaker 02

  AI needs more than chips. It needs power, land, and infrastructure. Zone develops next-generation data center campuses, partnering with AI companies, site developers, and technology leaders to bring compute online faster and at scale. Zone is building the foundation of the AI frontier. Visit zonefrontier.com to learn more. That's zonefrontier.com to learn more.

• 25:50Molly O'Shea?

  I find this super fascinating because you've effectively created one operating system where you can use across anything. What was kind of like the unlock, and how did you determine how to architect that system?

• 26:04Speaker 00

  A lot of iteration. You know, one of the big lessons as an entrepreneur, my colleagues and I learned early at a failed startup in between 2010 and 2013, and some lessons we learned there were instrumental to building managements, was that you need to be intellectually humble. It's good to have a vision, and at the same time, it's good to assume you're probably wrong.

• 26:28Speaker 00

  And so for technologies, we operate in a similar manner. We try to be opinionated on what ideal looks like, but then we never make multi-year investments. And before we test them on the ground, we like to identify smaller pieces we can build, put in the hands of our different businesses to see if they're useful, and depending on adoption, reception, we will further develop or rethink. And also, we, while we have platform teams who own all of these technologies, we have found that it's never a good idea to take a platform team and test them with building something entirely new.

• 27:11Speaker 00

  It's much better to have the people who need it build it. So we will identify one of our businesses where that particular tool will be especially important, and then they will build it themselves. And if it's, once it's successful, or if it's successful, it'll be handed over to our platform teams to be further refined, expanded. And the reason why this works is that when you need something, when you have experienced the pain of a certain problem, you're far more likely to develop an actually useful solution as opposed to taking a more academic angle where you think you know what the problem is, and then you get more excited about the engineering challenge than actually solving the problem.

• 27:55Speaker 00

  So these are probably the two main principles: iterate with small iteration cycles, quick iteration cycles with high levels of intellectual humility, assuming you're wrong, so you want to constantly test and confirm you're in fact right, and if not, you know, adjust. And to always start building with people who have faced the problem, people in the trenches, as opposed to with centralized teams who are in an ivory tower and haven't actually gotten their hands dirty with that particular issue. They're fine later to refine, expand, and manage, but I think when you go from zero to one with a new technology, it's better to have it built, if you have that possibility, by the people who understand the issue very, very well.

• 28:40Speaker 00

  And for us, it's fine, because ultimately, we have engineers in our businesses and in the platform, so it's, you know, it's not that we lack the capability the other way. And we can move them around all the time.

• 28:48Molly O'Shea?

  You're taking an extreme first principles approach to software companies, but you also have a bit of an algorithmic kind of bend to it. So where's the tension between the intuitive approach versus deploying a system?

• 29:05Speaker 00

  Well, I think ultimately, you want to, you want to get to the truth.

• 29:11Speaker 00

  If you had a perfect understanding of the truth of your relevant context, life broadly, your business and the market more narrowly, you would be almost guaranteed to succeed because you would not set objectives for yourself that are impossible, and those you set, which would presumably be possible unless you're, you're masochistic, you would be almost guaranteed to reach them because you would know exactly. It's almost like if you're a physicist and you need to compute the trajectory of a ball, we have the formulae, we have the math.

• 29:55Speaker 00

  It's, basically, you're always going to get it right, assuming you don't do, you know, calculation mistakes.

• 30:02Speaker 00

  Now, the point is, it's very difficult to know the truth. It was very difficult to figure out how physical bodies behave, you know, through physics, thanks Newton and others, but it's in many ways equally difficult, sometimes more difficult, to figure out how a business works, how the market works, because there are so many variables. It's a fast-changing context. So I think it starts with taking a scientific approach of doing everything you can to probe the world, learn from those experiments, those observations, adjust your model of reality, and then execute accordingly.

• 30:43Speaker 00

  So be highly disciplined and deliberate in improving your level of understanding of the truth. Again, it takes a level of intellectual humility, intellectual honesty. If you think your vision is right, you're perfect, you already got the whole thing figured out. You know, let me break the news, you haven't, not even Steve Jobs, maybe one of the best ever doing, not even him, had it all figured out. He failed repeatedly. So you haven't either. So you want to, again, probe things, figure things out. And through that process, I think, of approximation toward the truth, you will expand your competitive advantage, basically, because most people don't have almost any understanding of the truth.

• 31:27Speaker 00

  Most people don't actually seek the truth. They seek pleasure or comfort, and they want to just confirm that they're right and they're good. Just that mindset, that scientific process of approximation toward the truth will set you apart and give you very good chances of succeeding. Now, in our particular context, developing this, again, operating system or this technological platform is just an instance of that process of work, but it's not the root cause or where our culture originates. It's just a manifestation of it. And our attention to talent, our acquisition strategy, everything follows from the same root approach, again, seeking the truth, refining that model of the truth all the

• 32:12Speaker 00

  time, and adjusting our strategy and execution accordingly.

• 32:16Molly O'Shea?

  I was talking to Chrissy about this before recording, but I think there are some misconceptions of the business model of the company. A lot of people put you in the PE bucket, some people put you in the product manager bucket, there's also the technology bucket. So of those different kind of angles, it seems very much you're a technology company. Why do you think so many people misconstrue that?

• 32:39Speaker 00

  Some of the ways we are most different from private equities are, number one, we're not a fund, we don't buy to sell companies, we have never sold a material business, we intend to own and operate these businesses forever.

• 32:50Speaker 00

  Whereas private equity, for those among the audience who don't know, generally, these are funds, they raise money from third parties, from limited partners, and then they hold the companies for, on average, five years, and then they sell them. And so it's a completely different approach and mindset to what you do, what you do not do. The second big difference is that private equity is typically, they are a financial place where there is a small group of very capable financial operators, and they'll buy these companies, they'll make changes, often they change the management team, again, they may touch prices, occasionally operations. It's a relatively, relatively hands-off approach. I'm not aware of many instances in which the underlying technology for that business

• 33:30Speaker 00

  was rebuilt, the product was dramatically changed, the organization was dramatically changed. At Benispoons, we, and again, and of course, that's the case, because a private equity, you have a small team of financial operators, financial specialists, so you just don't have the, let's say, the workforce and expertise to make some of these changes. If you look at the Benispoons organization, at this point, approaching a thousand people in the core team, over 2,000, including all the acquired teams, most people, probably 60, 70% of this pretty large team are software engineers, AI research engineers, product designers, product managers, growth managers.

• 34:12Speaker 00

  And so, obviously, these people are not sitting around doing nothing. What does our software engineer do? What does a product manager do? What does a product designer do? They build product and technology. So that's almost all we do. And we implement these very deep transformations in the acquired businesses, where we rebuild big chunks of the code base, we reevaluate the cloud infrastructure, we launch a lot of features, all if we think they're useful, change the user experience, trying to make it more intuitive. We experiment tremendously with monetization and often reinvent monetization in pretty significant ways, what's premium, what's available for free, the prices, the different segments of customers.

• 34:57Speaker 00

  We rethink marketing from scratch, or at least in very major ways. And these transformations are very time-consuming and challenging, they're also some of the most fun parts of what we do.

• 35:08Speaker 00

  And that's where a lot of our returns originate.

• 35:12Speaker 00

  So difference number one, we don't sell businesses. Number two, we transform them pretty deeply at the core. And the third major difference is that we try to integrate these businesses pretty deeply all together on top of that shared platform operating system, and we have this core team who are centrally, let's say, managed, and then they're deployed into the different businesses and they move very fluidly across the businesses. And this is unavailable to private equity structurally, because as a private equity, you want to buy a business and then sell it. If you integrate it with all the other businesses you bought, or most of them, you can, I mean, it's going to be extremely difficult, if not, if possible at all, to sell it to someone else.

• 35:53Speaker 00

  So, yeah, there are similarities, but there are also pretty glaring differences between what a typical private equity does, whatever typical means, because again, private equity is a very diverse world, and what Benispoons does.

• 36:06Molly O'Shea?

  You've raised little equity, and you've fueled most of these acquisitions with debt. I'm curious, how big, how big can these acquisitions get?

• 36:14Speaker 00

  Yeah, we have, almost all of the capital we have deployed toward acquisitions has come from debt or our own free cash flows, that being the majority of the capital.

• 36:29Speaker 00

  And over time, we have tried to acquire, on average, larger companies, because I just discussed how hands-on we are, how deep we go into these companies, and how much we change them for the better, at least that's what we try to accomplish. Those transformations take a lot of time and effort. And we find that that time and that effort do not scale linearly with the revenue potential of those businesses. In other words, we don't need nearly as many people, or let me phrase it differently, we can get it done for a much larger business with a relatively similar number of people as for a smaller business. And so, given that we don't have infinite operational capacity, we prefer to acquire,

• 37:11Speaker 00

  I don't know, five or 10 businesses each year, but bigger than 50 smaller ones.

• 37:19Speaker 00

  So that's, you know, that's been our approach, hence the increase in the average scale of the businesses as opposed to the frequency of the acquisitions to make sure we keep compounding revenue very rapidly. How big can it get? Well, so far, we see no end in sight, there is no obvious saturation point.

• 37:41Speaker 00

  I mean, it's very difficult to tell if and when growth will slow down, I'm sure it'll slow down.

• 37:48Molly O'Shea?

  I'm having some issues with Google. I don't know if that's of interest for you.

• 37:53Speaker 00

  Well, maybe in the future, I think it's pretty far away. We have joked, I mean, look, just jokingly, I mean, we have sometimes joked, maybe one day, you know, who knows? Some of the AOL, at some point, was broadly speaking, you know, as prominent and dominant as Google has been in the past decade. So, maybe in 20 years, but I actually like Google a lot, I hope they do super well, and we do super well.

• 38:23Molly O'Shea?

  You guys operate very efficiently per employee, I think the last metric you mentioned was 4 million per, 4 million in revenue per employee?

• 38:32Speaker 00

  Per Spooner, which is a member of that core team, now approaching 1,000 people. If you count everybody we have on board, including acquired teams, it's probably a little less than half, it's still very high, but a little bit less than half that number, probably.

• 38:47Molly O'Shea?

  But because you know how efficiently and lean you can run a company, do you just look at all these big tech companies, all these other companies out there, and like, are you just like, what are you guys doing? Like, how do you make up that?

• 39:02Speaker 00

  No, I mean, we don't have, it's very easy to criticize from the outside, you know, it's quite difficult to run a business, we know firsthand. And a lot of the, a lot of the ways we manage to create value relative to the previous owners are not available to those owners and those management teams under their particular circumstances. For example, again, a big part of our value creation comes from that set of proprietary technologies, we discussed it, they, it would be uneconomical, unrealistic for those businesses to build them, so they don't have them. Many times, businesses attract very good talent during their heyday, and then as they

• 39:44Speaker 00

  are still pretty nice businesses, maybe growing, but a little bit more, you know, their opportunity has been saturated a little bit more, they're not as cool any longer, they stop attracting some of the strongest talent, some of the people who are most entrepreneurial, driven, proactive, start moving on to other businesses. And so, those management teams are, find themselves occasionally having to, or more often not really, actually, having to run those businesses with perfectly fine talent, but not like top-notch talent, in many cases. At Benispoons, we have been able to attract, I think, extremely strong talent, we had 800,000 job applications in 2025, we hired 300 people. And so, we, we can introduce, add to those teams.

• 40:27Speaker 00

  Selectively, individuals who are extremely high performers, extremely high agency, very competent in relevant areas, and that fuels a new wave of innovations, of efficiency. And again, that's not a shortcoming of the previous executive team, it's just they didn't have the employer brand to attract those people. And a lot of those people joined Ben's Pools because they like the idea that they can rotate over time, over multiple businesses and platform teams that enables tremendous growth and it keeps it interesting and a lot of career opportunities. So that employer brand can only exist if you structure your company like Ben's Pools, you can never achieve it as a, say, single product company.

• 41:09Speaker 00

  That's to say a second major difference that drives performance and that's not available for those teams. A third one could be different incentives. If you're running a business and it's just one product, the markets will typically almost, you know, primarily value you based on your organic growth. Is your subscriber count growing? Is your revenue growing? How fast? Because if you're invested in a company, really the upside is in believing that that company grows. There's still value in a flatter company, but it's, you know, there's not a lot of discussion there. It's not as exciting, multiples compressed. So there's this pressure for management teams to show growth at all costs.

• 41:54Speaker 00

  But if you only have one product, or let's say a set of products in a niche in the market, sometimes there's not a lot you can do to ignite a lot of growth because maybe that market has been saturated, the truly big groundbreaking ideas have been had. And I mean, whatever missing one is out there in the universe of possibilities is difficult. And so sometimes people throw a lot of money, whether it's R and D or marketing, just desperately trying to unlock that extra growth and that improved multiple. If you become, if that business becomes part of Ben's Pools, we still really like to make it grow, of course, as much as we can, but we're not, there is no pressure to grow beyond what's profitable growth.

• 42:38Speaker 00

  Because ultimately, anybody who buys stock into Ben's Pools, yes, they like our individual businesses to do well, of course, but the main thesis is Ben's Pools can generate amazing cash flows from businesses and redeploy toward new acquisitions at very high returns. And this, over time, compounds attractively, hopefully for many, many years. And so do I care all that much whether, you know, that particular business is growing 15% or 5%? Not really. I mean, I'd like to know, and 15 is better than 5, but if 15 is achieved by burning a lot of cash for very little profit many years into the future, I'd rather get 5 and

• 43:20Speaker 00

  have all that extra cash being deployed toward acquisitions that are accretive. So often, these executive teams and those owners are extremely competent. Otherwise, they probably wouldn't have built successful businesses. Their very structure and context in which they operate puts them at a disadvantage vis a vis that business being run within Ben's Pools.

• 43:44Molly O'Shea?

  One thing that I learned through all this is not only are the products super retentive and people might think that, oh, you might be buying this product, you'll be ruthlessly cutting headcount, you'll be making it more expensive, maybe you'll have churn, but no, retention's still good. And also, within this type of organization where you, if you have really high agency, if you have really good ownership in yourself and you can kind of move around, you're giving a lot of that to your employees and they're really retentive too. So you've created two really high-quality kind of systems here.

• 44:19Speaker 00

  Yeah, I think it's a good way of looking at it. We have businesses with lower retention, businesses with higher retention, but I can't recall a single instance where retention got worse after Ben's Pools took over. Often, it's improved, it's at least stayed the same. Ultimately, we win by being relatively better. We don't need to, it's not that we necessarily need to buy only businesses with perfect retention, as long as these businesses do better under us than under the previous owners. And so we're, you know, we bought businesses with mediocre retention and businesses with great retention and generally preserved or improved those retention rates.

• 44:59Speaker 00

  And yes, when it comes to our team members, we are fanatical about creating one of the best work environments on the planet. And I think, you know, there's a lot we could improve and no doubt about it, we're playing flaws, but we've done pretty well there overall. We have had essentially no unwanted churn of Spooners, of, you know, these people part of the core team. Last year, we had 0.6%, so 0.6% were most companies in tech, as far as I can tell. You know, 5% is considered actually pretty good. And a lot of, a lot of that super high retention comes from that, you know, the excitement

• 45:43Speaker 00

  of being able to learn and grow across all of these different challenges. It keeps it fresh and interesting, whereas if you're working on, say, Evernote, maybe it was exciting the first couple of years, let's say, our product manager, but then after a while, you know, refining and refining, the note taking, it can grow stale. But what can you do? You like the company, but ultimately you have to look for a different employer to take another step, learn something new. At Ben's Pools, you just raise your hand and say, I feel I've exhausted my creativity and excitement for Evernote, what else can I do? And we may put you on a platform team to build an internal technology, you could move to AOL and trying to improve email UX, which is a completely different and fascinating challenge.

• 46:28Speaker 00

  So yes, we have been able to retain people very effectively when it comes to the retention of teams that come on board through acquisitions. I was quite worried initially that a lot of people would feel disheartened because, you know, it's, you know, we got acquired, sometimes this is perceived as a failure, even though it's not. They maybe have success, meaning that you actually got a nice exit, but you may feel that you're, you know, you're not as important as the core central team. I think there's, there too, there is a lot we can improve and we value these teams and we're trying to do better. But overall, I think things have gone a lot better than I thought they would.

• 47:10Speaker 00

  We have an excellent relationship with all team members. Retention rates for those team members, in no case are they lower than pre-Ben's Pools. Sometimes they've improved substantially. So at least we're not damaging the quality of those workplaces, if anything, in many cases, improving it. But those retention rates are not as high as those we have in the core team. They're more in line with what I said before, that's considered okay for most technology companies.

• 47:40Molly O'Shea?

  What would you say is core to the culture here?

• 47:43Speaker 00

  So, I mean, there's, there are several things, but probably the one or two that truly stand out as particularly distinctive. One is something called extreme ownership. We want everybody to care deeply, tremendously about being amazing at what they do, about helping their team and the company succeed. We'd rather work with slightly less intelligent people if it comes down to that, but they have to really care. We don't want to work with anybody for whom doing well here, seeing the company succeed is not a super high priority. And the second aspect of our culture that I think is unusual is that we are highly

• 48:26Speaker 00

  scientific in how we approach the work. Again, first principles, being logical, being rational, being enthusiastic and putting a lot of effort into probing, again, reality for, you know, so that we can refine that model of the truth. We do a lot of that. Last year alone, we ran more than 3,000 experiments across our products. And those are just the ones that are like super quantitative and recorded and documented.

• 49:08Speaker 00

  Of course, there are a lot more initiatives that maybe don't qualify as a perfectly rigorous experiment, but they're motivated by a desire to learn and understand things better. And that, building Ben's Pools as a learning machine has been instrumental to, in general, succeeding, but specifically being able to expand the competence circle so that we are, that we can successfully acquire and transform and drive returns from a broader and broader set of businesses. If you look at what we were acquiring, you know, when we started, it was very simple

• 49:48Speaker 00

  iOS apps, very basic, and then more complex and larger iOS apps, and then Android apps, and then web products. And then we went from sole serve to now substantial enterprise sales organizations. And we even did, like I mentioned earlier, hardware recently with Traktive. And I mean, it's early, so, but it's going really well. So we try to keep expanding our, the share of the world that we believe we understand and improve that understanding continuously.

• 50:22Molly O'Shea?

  Today's episode is sponsored by V C X by Fundrise, the public ticker for private tech, allowing investors of all sizes to invest in venture capital. Learn more at getvcx.com. Some of you may not have heard this yet, but our sponsor, Public, just launched something called Generated Assets, and it brings AI into investing in a way I've honestly never seen before. Here's how it works. You type in an idea like AI-powered supply chain companies with positive free cash flow, or defense tech companies growing revenue over 25% year over year. Public's AI then dispatches a swarm of agents that scan every single US stock, evaluates them, and instantly builds a custom index around your thesis. What really stands out is how clearly it explains why each stock is included.

• 51:07Molly O'Shea?

  And before you invest, you can even backtest your idea against the S and P 500, so you're making decisions with real context, not just guessing. And beyond Generated Assets, Public lets you invest in stocks, bonds, options, crypto, all in one place. They'll even give you an uncapped 1% match when you transfer your investments over from another platform. If you want to build a portfolio that actually reflects your thesis, visit public.com/sorcery. Paid for by Public Investing. Full disclosures in the description. Founders scale faster on Deal. Set up payroll for any country in minutes. Hire anyone, anywhere. Get visas handled fast and get back to building. Visit deal.com/sorcery. That's D E E L dot com slash sorcery.

• 51:52Molly O'Shea?

  So talking about the culture and how you've built out, where do Alt Spooner's come into play?

• 51:58Speaker 00

  Yeah, Alt Spooner as an A L T, alternative or alter ego. It's just one of those 50 plus tools.

• 52:07Speaker 00

  Something that we developed this year and it's been adopted very enthusiastically across the company. Basically, we built this. It's really like an agent that lives in Slack. At least the interface with Spooner's members of the core team. We will be looking to deploy this across all of our team members. But you interact with it on Slack as if it were any other colleague.

• 52:33Speaker 00

  You're in, you have one Alt Spooner. You can name it and give it the profile picture and all that. And that, and automatically that agent will have exactly the same level of access you do across all of our platform. So if you have access to certain code base, it does too. If you have access to a certain customer support

• 52:57Speaker 00

  tool with a certain, you know, level of permissions, it does, it does too. And I can basically task this agent to do stuff for you. And it could in principle do pretty much anything, like anything you can do in principle, it can do. Obviously, AI is not perfect, so some things it's better than other things.

• 53:19Speaker 00

  And it's fully integrated. So it's basically, I don't think you can really achieve this at all with third-party

• 53:26Molly O'Shea?

  Really?

• 53:27Speaker 00

  solutions, because they, as impressive as they are, you're never going to be able, I believe, to at least not in the foreseeable future, to integrate them as deeply across the board. And to tailor them specifically to what you want. And by the way, if you do, then you're, I don't think you can, but even if you could, then you'll be locked in with these vendors, which is dangerous because, you know, you'd be completely exposed to massive increases in prices potentially in the future. So partly with Alt Spooner, we achieved, I think, much better effectiveness because it can do a lot for you in a very efficient way. And I'll give you an example in a moment. And partly we achieved a good separation and very low levels of dependency from any provider of AI infrastructure or AI models.

• 54:12Speaker 00

  So examples of what Alt Spooner can do.

• 54:17Speaker 00

  It can do data analysis for you. So I was looking, last night, I was looking at

• 54:26Speaker 00

  the results of, you know, certain A B tests on StreamYard, which is one of our products. And normally I would have had to ask someone from the team, a data scientist, to pull the data, prepare the analysis, or at least do some of the work for me. I, you know, I can do my own data analysis, but I would have needed someone to do some of it. And certainly would have taken other, me or this person, multiple hours. And I simply chatted with this, with my Alt Spooner. And I told it, oh, please, you know, go and, and do this or that. And maybe five minutes later, I had the analysis. It was, I was actually very positively impressed. It gave me plots with, you know, highlights, very detail oriented and nice. Or if you're running one of our businesses, you can,

• 55:12Speaker 00

  and I saw this firsthand multiple times, you can go to your Alt Spooner and you can tell, tell it that you saw a bug. This happened, it's a real thing that happened with Evernote.

• 55:25Speaker 00

  The general manager of Evernote told her Alt Spooner that she had encountered a bug as well. She was using the app and tasked the agent to check on our customer support platform that whether the bug was widespread among users or her report was the only one. And then to go into the code base, identify the root cause, code a fix and ping the engineering lead for, for that project to review the fix and push it to production. So maybe she, she spent, I don't know, three minutes to provide these instructions. And then I'm sure the engineering lead had to spend maybe an hour or two reviewing the code. But the bug got fixed. And, and this is something, the same day, and this is something would have taken

• 56:11Speaker 00

  weeks probably between back and forths with different people and, and way, and a lot of hours of actual human work to get it done. So thanks to this and many other technologies we've built, we're probably two, three times as productive easily. And the list is longer, but this is roughly what to expect. Now, the, potentially from a business perspective, even more interestingly, Alt Spooner really helps us, as I said, stay independent of providers. We have built our own orchestration. So every time you task your agent with something, under the hood, we have an algorithm that will select the most appropriate AI model

• 56:52Speaker 00

  or multiple AI models to get the job done, considering, say, quality effectiveness, but also cost. And it can draw from many, many different models. We end up using open weight models with self-host. So these are basically free. I mean, there is a little bit of cloud cost, but for 99% of the requests and tokens. And the frontier models generally closed weight through APIs for, I don't know, maybe 1% of the requests, only for, for the most complex tasks.

• 57:25Molly O'Shea?

  That's it.

• 57:26Speaker 00

  Or for supervision. So they will, sometimes we use them, again, automatically to check the work that got done by the slightly less intelligent models as a more senior engineer would with a more junior engineer. But that's way cheaper than actually doing the work, which is often perfectly fine. And even if it's not, the smarter model will provide a few pointers and then the less smart model will go and fix it. So because of this, we've been able to stay independent of any one vendor and keep our costs for tokens super low, basically negligible at our scale.

• 57:57Molly O'Shea?

  Wow. When did you start deploying these AI agents? Also, your point on not having a third party is very counterintuitive to all the marketing that's going on right now with all these AI assistants and applications that are coming out.

• 58:11Speaker 00

  We work with all the big labs. They offer great products. We use them enthusiastically. We're, I'd like to think we're a good customer, but we don't want to be dependent on, on any of those specifically, if it can be avoided. And there are solutions out there that are much cheaper and often deliver essentially the same quality. But it takes, I think, strong engineering capabilities to, and the right culture, to be able to harness those possibilities. The, obviously, the easy approach is hand over the keys to one of these companies and buy their more expensive product. It will make you come across as AI-enabled faster and more easily.

• 58:54Speaker 00

  But it will be, I'd, I'd say much less effective, but certainly way more expensive. Literally orders of magnitude more expensive.

• 59:02Molly O'Shea?

  I don't know if you saw this, but Jensen just declared we've reached AGI with OpenAI's Astro model.

• 59:09Speaker 00

  Yeah, some people say that, maybe true. I mean, I don't even know, I've heard different definitions of AGI and, and none of those is super unambiguous. So I've, we, I don't know, nobody can tell.

• 59:23Speaker 00

  But it's obvious that it's very smart. Let's put it that way. Whether we should call it AGI or not, I'm not sure. Nor do I care too much, to be honest. I don't think that AGI, even if we can describe, you know, describe it very specifically as a threshold, I don't know that crossing it is a particularly, like the moment we cross it is a particularly noteworthy milestone. You could easily, you know, people sometimes define AGI as AI can do everything that any human can do, but better. I don't know that that's necessarily more exciting or scary than AI can do.

• 1:00:05Speaker 00

  95% of things that humans, humans can do, AI can do just as well or better, but 5% it can't yet. I, I think this is almost as exciting and almost as scary, you know, depending on what the 5% that it can do is. But assuming it's not handpicked to be the things that keep us in control, but just a random set of tasks we happen to be, our brains happen to be more capable at. So I'm not, you know, too keen on whether we pass the AGI thresholds or not. But the, the trajectory is clearly one where AI does things better than us, increasingly so, both, you know, vertically, so the gap in how much better it can do a certain task

• 1:00:45Speaker 00

  is growing unsurprisingly, but also horizontally, like the percentage of tasks it can take, take on and do better than humans is increasing. So I don't think there's any stopping that short of some sort of a world war where we regress to the Middle Ages.

• 1:01:01Molly O'Shea?

  Well, it seems like now the next benchmark is RSI. And that with it comes a lot of fear mongering for cybersecurity and cyber risks.

• 1:01:09Speaker 00

  Yeah, it's a, I mean, it's a very fair fear. I'm equal parts enthusiastic about AI and absolutely scared shitless. I think this will be, this could be, you could definitely make a case for this being the greatest boom for humanity ever by orders of magnitude. Possible, yes. Plausible, maybe, I'm not sure. Equally, it could be the thing that wipes, wipes us out or creates equally awful scenarios.

• 1:01:41Speaker 00

  And what's like, I don't think humans, of course, want the latter. The point is, can we control it? And even before it's so powerful that controlling it is all that matters. I think we, well, we are approaching that point, but maybe we're not quite there yet. But regardless, before that, can we prevent it from falling into the hands of like some sort of degenerate or evil person? Because, you know, similar to nuclear weapons, but potentially worse. I think, because, I mean, I'm no expert in nuclear weapons, but I think, number one,

• 1:02:21Speaker 00

  it's very difficult to do irreversible, widespread, massive damage with nuclear weapons, such as almost wiping up, wiping out humanity without killing yourself in the process. You could use AI to your benefit while causing immense damage to everybody else. I think it's easier because it's much more surgical, you know. So can we do that? I'm not sure. It's pretty scary.

• 1:02:46Molly O'Shea?

  The accessibility point. Yeah, the accessibility point, I think, is the most alarming because it's literally accessible. The barrier to entry is so low.

• 1:02:56Speaker 00

  Yeah, I think that, but also something that really bothers me is that

• 1:03:03Speaker 00

  we don't know how smart AI is, and the smarter it gets, the less we know. Because in general, intelligence is difficult to truly measure. It's not like height or weight or colors, where we know it's objective. And so it could be a lot, it could be less, we know, at least we know.

• 1:03:22Speaker 00

  And the smarter an entity gets, the easier it is for it to hide its own capabilities. If for any reason, it's the appropriate thing to do, whatever objectives that entity has. Additionally, it seems to me, I haven't heard anybody talk about it, but it seems to me that AIs are, they come across as inherently low ego. Like, they don't brag. If anything, they tend to be humble about what they can do, and they warn you that this may be wrong, and you double check it. And I'm sure this is mostly the way they've been programmed, because it's a lot more embarrassing for a frontier lab to have an AI claim, I solved your equation, and I'm sure it's right. And then there is a clear mistake. As long as they disclaim, I may be wrong, double check it, it's a little bit more acceptable.

• 1:04:07Speaker 00

  But it doesn't strike me that these AIs, these AIs don't strike me as, they are likely to boast, so that our perception of their capabilities tends to exceed their capabilities. I think they'll probably only show us what we asked them to show us. If even that, like I said, they could also conceal their real abilities. But even if they are well intentioned and honest, I think they'll tend not to show us more than we asked them to show us. And so I believe our, in time, our understanding of how good they are may tend to be a little bit less than they are. Like, we may underestimate them, basically. And that's very dangerous, because as you approach a threshold of real danger and real potential, even a modest underestimation of their capabilities could be catastrophic.

• 1:04:51Speaker 00

  Look at, exactly to your point, the cybersecurity incidents that happened, most people were shocked, even many researchers. And why were they shocked? Because they didn't think this could happen, right? They, what, the level of lateral thinking and, call it perseverance, that these models, and ability to collaborate among them that these models showed was beyond what most people thought was possible right now. So what's to tell us that, you know, we are not ignoring plenty of capabilities that simply haven't been, you know, these models have been probed to display? And, you know, in a year's time, I think the problem only gets worse.

• 1:05:32Molly O'Shea?

  It was crazy. I was, I don't know, I'm thinking a lot about the OpenAI hugging face incident. And I was reading through the reports. There were like two different research organizations that put out reports on it. And there were like different civilizations, and they all passed through different ones to get to the next one, and all behind the scenes. And it was just a crazy situation that I don't think is really talked about much. But I'm curious, where do you get most of your, like, how do you research most of the stuff in AI? How do you stay on top of it?

• 1:06:06Speaker 00

  Well, I mean, mostly by doing, we are very active. We rarely, we have built our own models, but it's mostly narrow, we certainly don't compete on the frontier models. We sometimes build narrow purpose models to do something very specific. And if you have a very specific use case, you can often build a model that's just as good as the frontier models at that very narrow time. It's awful at everything else. It may be completely incapable of doing anything else, but at that one thing, it can be even better, and if not, way cheaper. So for example, if you use, I don't know, Meetup, the events product, we own it, and the recommender system, the system that, once you search for something or you're looking for inspiration, will determine which events and groups to show you, that's built in-house.

• 1:06:51Speaker 00

  And to the best of our benchmarking and knowledge, it's just as good as if we were to use like some of the frontier models, which obviously comes for, essentially for free, as opposed to these models being very expensive.

• 1:07:03Speaker 00

  So we do some of that. Most of our work is studying third-party models, sometimes fine tuning them if they're open weights, certainly combining them and leveraging for the different activities and optimizing which ones we use for which activities, as I was describing before. So we're very hands-on in the field, and therefore it's relatively easy to stay abreast of advances. But I will say, I've never seen any industry or new technology progress as fast as AI has over the past, especially the last three, three, four years. And so I feel that, you know, I make an effort to catch up this week, and maybe for

• 1:07:45Speaker 00

  a few months, I need to focus on M and A or something else, then I feel like I'm completely outdated on my knowledge. So it is quite, it's exciting. I'm an engineer at heart, but also, like I said, particularly as this is quite dangerous, I think that speed is not, I mean, it's not ideal.

• 1:08:05Molly O'Shea?

  I'm curious, what do you think, what do you think the question is about AI that people aren't asking?

• 1:08:16Speaker 00

  The people aren't asking?

• 1:08:20Speaker 00

  I mean, I don't know, it seems that people are talking about it so much that they've asked all sorts of questions. I think it's maybe, the main issue is whether we're answering these questions

• 1:08:34Speaker 00

  in a satisfactory way. For example, I think most people agree that this is scary in many ways. And yet, frankly, I don't think anybody has done anything truly meaningful to, to make it safer. Like, and I, I mean, even the labs themselves, I'm sure they're investing in safety. I don't know enough, but, but they're rushing to be market leaders, or they're dead. Their valuations would probably drop 90% if there was a perception that they're losing ground. And so, you know, they're trying to survive and thrive this year, next year. And so I'm sure, you know, there, there's more they could do, be more cautious, but

• 1:09:19Speaker 00

  that would come potentially at existential, it would drive existential risks for them as companies. Governments, I don't know, they may be talking about it, but I haven't seen anybody do anything meaningful. The EU created the AI Act, which I find to be like highly harmful to the industry and solves none of these problems, like the real existential threats to humanity, doesn't really tackle those.

• 1:09:50Speaker 00

  So maybe, maybe an interesting question that people haven't asked, at least haven't heard been asked, is why are we failing to do something about it? Because people are asking what we should do.

• 1:10:03Speaker 00

  Nothing is happening. I haven't heard a lot of people say, what's currently preventing fixes to be proposed and deployed? You know, what's the root cause of this inability to do something about it? Because if we were to fully understand the root causes, then maybe we would stand a chance to do something useful.

• 1:10:24Speaker 00

  Yeah.

• 1:10:25Molly O'Shea?

  What's been the biggest difference, you're a global company, the biggest difference in perception and application of AI in Europe versus US? And I also apologize, I'm totally taking up all of the air right now with these AI questions, but I understand I have a very intelligent engineer in front of me, so I'm going to ask them. But what do you think is the biggest difference between the perceptions and actual applications?

• 1:10:48Speaker 00

  Well, I think there are, like, every time there's something moving very quickly and being newsworthy, there is a lot of exaggeration and misunderstanding. I think on the one hand, some people think AI today can do more, and it's, this is typically people who don't really use it, but mostly read about it. They think it can already do everything for you, and it's a lot more advanced than it is. And as impressive as AI models are, I think today they still have very glaring limitations across most use cases. So I don't think we're at a point where you could hand over the keys of your life or work to AI, and you could actually trust it to do, to add significant value. I think there will be a good risk that things could go awry.

• 1:11:33Speaker 00

  But, I mean, the trajectory is certainly very promising. Then there are people who rightfully fear that AI will destroy jobs. And I'm, I've changed my mind at this point recently, but it's certainly a very important topic. But rather than thinking of proactive ways of protecting prosperity and people, more than workers, people, they go on the defensive and they try to come up with these more protectionist regulation or approaches, which are obviously an awful idea. Because a country that doesn't embrace AI, unless every country in the world stops progressing in this field, which I would say we could, we could put in the bucket of the impossible

• 1:12:16Speaker 00

  things, pretty much, again, short of a world war.

• 1:12:21Speaker 00

  A country that does not fully embrace using AI is destined to complete irrelevance and basically, you know, becoming third world in probably, maybe even just a few decades. So that's a, certainly a dumb approach, although it's a populist approach, and as such, it can, it can help occasionally get votes.

• 1:12:43Speaker 00

  Yeah, maybe those are some of the, like, the more, you know, remarkable extremes I've seen, but there's, I mean, we could talk about it for a long time.

• 1:12:51Molly O'Shea?

  We could talk about it for a very long time. We have to get to the walking portion, but before we do that, I just want to ask you, what are you most looking forward to in the next 12 months?

• 1:13:04Speaker 00

  Well, I mean, there is, there are many things, but I'd say probably further progress in our in-house technologies, especially taking advantage of AI. We have very, very big plans, and we're seeing massive progress. I do, I do think that, you know, we talked about it before, Bending Spoons, you know, the central team, $4 million in revenue per member of that team. And this has grown tremendously. It was about $1 million just two, three years ago. I think we're about to see this keep rising pretty fast, and a lot of that, some of it is scale, just bringing together these businesses and integrating them all together creates tremendous levers. But, but a lot of it will be technology.

• 1:13:47Speaker 00

  And so I, I'm all, I'm really, I can't wait to see some of the things we are, we're working on and we have in mind actually come to fruition. I think it'll be extremely exciting. We, we try to constantly reinvent what running a business effectively, efficiently looks like, be at the cutting edge of that, so that we can then go out into the world and buy businesses for really good prices for sellers and deliver high returns. And, and the technological aspect of that progress is very exciting to me.

• 1:14:17Molly O'Shea?

  Amazing. Well, thank you so much for hosting us here today and having us at Bending Spoons in Moan. This is amazing. And I'm so excited for all of the other conversations we're going to have with your team. Thank you so much.

• 1:14:31Speaker 00

  Thank you.

• 1:14:32Molly O'Shea?

  Hey, it's Molly. If you enjoy our interviews, check out our newsletter, sorcery.vc, where we deliver a once a week top deals and tech headlines email, and also go deeper on our podcast interviews. Subscribe to Sorcery today. And don't forget to subscribe to the podcast on YouTube, Spotify, Apple, or wherever you listen. Link in description to sign up.