a16z’s David Haber: Recording Every Meeting, Firm Over Fund & Why Fund Size Equals Conviction

0:00Robinhood, Chime, Revolut, New Bank, the first wave of fintech over the last 15 years, built some of the most valuable companies on earth. My guest today thinks that wave is over.

0:16David Haber: And so I think it's just very, very difficult to build the next Robinhood today. I hope that's not the case, but I think the game is largely won. That's David Haver, general partner at Andreessen Horowitz, a firm with about$100 billion in assets under management. David has been investing and building in fintech since 2011. And lately, he's been obsessed with an edge most companies don't even realize they're sitting on, their own internal conversations. I think every meeting in a company basically should be recorded. Most of them already are, whether you know it or not. Conversations are where culture resides.

1:00David Haber: It's where expectations live. It's where edge case handling gets done. At A16Z, that's part of a bigger idea David calls firm over fund. Running an investment business like a startup company. Most people are running funds. And the objective function of a fund is how do I generate the most carry with the fewest people in the short amount of time possible? How do I build sources of compounding competitive advantage? What are my moat? And for David, recording everything is part of that moat. So much signal lives in every founder conversation, in every past note, in every investment memo. What we've lacked historically is a system of collective intelligence, that brain to stitch all that unstructured context together.

1:43David Haber: But if everyone knows they're being recorded, can they still be honest? But I do think that the Overton window has shifted pretty aggressively. You've been told for a long time not to put something in writing or in an email that you wouldn't want to have in a lawsuit. I think the same is happening in meetings. I'm bullish on where I was building skills. Yes. You may have like a little mini skiffs inside of your office. I think this room might be a perfect candidate. David is one of the most thoughtful investors I know. And at heart, he's a founder who never stopped building.

2:20The nature of how you think about building a business has definitely changed.

2:27David Haber: Certainly since we last spoke about three years ago. Since we last spoke on the record, just we've spoken in between. so one thing that you've written about that i actually really connected with and liked is how everything's being recorded now yep not just podcasts yep right um it started with note takers yep probably or i guess you can even go back to email yeah right it started with emails but now like note takers are almost everywhere you always have to assume that someone is recording. How do we get here? Yeah, let's talk about that. Why did you write about this? Yep. And then I have some thoughts about the matter so we can talk a bit about this.

3:11David Haber: Yeah, I mean, well, first, you know, I was a very early user of Granola. Yes. You know, probably one of the first couple dozen. I'd known Chris since his days at building Socratic when I was at Spark. And just credit to him for the team for building an amazing product. And I think, you know, one of their insights was that it should live kind of ambitently in the background, not join a Zoom meeting. And so I think part of the adoption of Grinnell in particular is just reducing the friction, you know, even if it's like the, you know, the uncomfortable sort of social friction of having one running.

3:49David Haber: The bigger insight that I had kind of came twofold. One was they did this end of year recap. I think it was last year. Maybe it was the year before. I forget where I had done like 750 meetings in Granola. Yeah. I think it was last year in the prior year. Yeah. And the, you know, it's like the equivalent of Spotify wrapped, you know, it does sort of this like end of year recap of your work life was like shockingly thoughtful. Yeah. And like identified, you know, mannerisms, things that I said in, you know, common phrases firm over fund, which were probably, you know, like there's just some people that things that I find myself repeating in conversations quite often.

4:26David Haber: And I was like, wait a second, like, Renault probably has more context on me and also is on A16Z as a firm than basically any other product, right? Because there's so much value in context that lives kind of unstructured in these transcripts that doesn't live in any other sort of system of record inside of a business, right? Your CRM doesn't have this. It is starting to, but historically, at least, hasn't had a lot of this sort of conversational information. And the other catalyst for me was I had this amazing, you know, associate sort of VP on my team. You probably met Mark in Drusco, who's awesome.

5:10David Haber: I had worked with Ed Goldman and, you know, go to Plaid and he was here for four or five years. He recently left to go to Oak to help build out their early stage practice, which, you know, I was very excited for him to go do. But I also realized to some degree I was spoiled. You know, I had this amazing guy kind of riding shotgun with me on most of the companies and deals that I was working on. And I used it as an excuse to just go really deep into a lot of these, you know, agentic coding tools like CloudCode and CodeX. in an attempt to basically automate a lot of the work that we would have done together.

5:45David Haber: And the first insight was, I'm just going to begin ingesting all my real notes and use it to categorize all my meetings so that it will know the difference between a board meeting, a monthly update, a one-on-one, a candidate to interview a podcast. And depending on the type of meeting that I'm having, it should kick off a bunch of work in the background. you know um so now this kind of product that i've been building you know uh before a board meeting goes into my email inbox pulls out a pdf of the board deck or a google slides or the docs end parses that document gives me a pre you know pre-meeting brief um you know diffed by the prior board meeting so it's like here's what changed you know since the last time you had a board meeting here are some of the things that i've identified you know um structurally that you may want to talk about.

6:35David Haber: You walk into the board meeting, you granola the board meeting, it layers in the granola notes. And then I've structured to the output is, you know, the board notes in the exact format that I want in my voice and it auto drafts an email that I can then forward to my team so that they have visibility into, you know, what's changed. And the product I've been building has expanded far beyond just that, but it's been an amazing window into, again, the value of a lot of this unstructured data. And I think it's, I've become so bullish on that, which is why I wrote the piece that I think every meeting in a company basically should be recorded.

7:13David Haber: Most of them already are to your point earlier, whether you know it or not. But it's because conversations are where culture resides, right? It's where expectations live. It's where edge case handling, gets done. And all that context is ultimately what you need to feed agents to actually do real and valuable work inside of a company. And so mapping that context, I think, happens largely through recording conversations. So ideally, this would strengthen the culture of a firm. 100%. Right. One thing that I remember, both you and I worked for very large banks that have been around for a century or two and everyone there talked about how that's a meaningful thing but i would always kind of find it amusing that the people that i was working with had only 10 20 maybe 30 years of experience they didn't know of the hairy deal that closed in 1910 correct that they could be learning from right totally that can actually be the case if you know the next MD at Pick Your Bank in 2100, right, can actually go back to all the notes and the actual conversations from 80 years ago.

8:36100%. That's where the institutional memory actually becomes something that you can brag about. Totally. Part of the way that I thought about even this tool

8:51David Haber: that I've been building is like, how do you deliver value at the edge? Meaning, how do you give the individual a valuable tool to like, you know, automate a lot of the work that they might not have wanted to do, make them smarter, free up time to actually spend time by the entrepreneur versus kind of the rote work they might have done, you know, previously. But then how do you sort of use, you know, that to build up, bottoms up the collective intelligence of the firm, right, to your point? And, you know, as an organization, even if you just think about injuries and horroids, there's just so much signal that lives in every founder conversation, in every past note, in every investment memo, and ultimately how those companies perform over time.

9:34David Haber: What we've lacked historically is a system of collective intelligence or a sort of a brain or central nervous system to stitch all that unstructured context together. And I think if you can do that, which I think is now possible with AI, you can really build a compounding advantage and reinforce the things that you want to reinforce, right to your point. Where, you know, the newest sort of expression in this tool that we've been playing with is it's actually a Slack agent that lives ambiently in the team. And so we recently had an offsite last week on the West Coast and Mark and I think ChatGPT wrote this sort of investment charter manifesto that like pushed you to think about, you know, where sources of defensibility may reside, where value may through, etc.

10:23David Haber: etc. And so I basically fed that into the product. So that anytime we're debating a deal on Slack, this product that I call Vantage internally, if it has something thoughtful to say, we'll actually jump into the conversation proactively and be like, but wait, have you thought about like point number three in this investment charter? And it's reinforcing, again culture but also just debate and also an investment thesis you know potentially or just ensuring that we're thinking deeply about kind of the edge cases yeah um again like i've just been so um like i have so much energy you know today from what's possible in uh because Because, you know, coding was sort of precious historically.

11:17David Haber: I didn't grow up as a developer. I took a couple of computer science classes in college. But I was always the product guy. I was always sort of a creative person. And I think back to my time building Bond Street. And, you know, I almost mourn that period because the feedback loops of actually like coming up with a product idea, working with my co-founder of the engineering team. They go off and build something for three weeks. they come back and be like, what do you think? Yeah, it's not quite right. And it was just every, it was, you know, a month in between each cycle. Now it's just, you know, context is the rate limit.

11:51David Haber: Like knowing what good looks like means that the speed of iteration in building product is like incredibly fast. And so I feel almost unleashed, you know, in what's possible. And it's why I've just gotten super deep down this rabbit hole. You've kind of given the answer, but how would you build Bond differently today? Oh, man. I mean, there are other flaws. Well, we're at Blondstreet's business model beyond software. Maybe most people are probably not aware of the business, but we were a small business lender providing loans from$10 ,000 all the way up to a million bucks. the biggest challenge that we had was acquiring that customer at the transactional point of intent.

12:37David Haber: Nobody's going to take out a loan unless you're ready to do something thoughtful. You're not taking out a mortgage unless you're ready to buy a house. In their case, it was to open up a new location or hire employees or buy a piece of equipment or maybe refinance a more expensive credit card or something. One of the ways that we tried solving that, although it didn't fully actualized was we had built this product called beacon um which effectively was like you can think of it as like mint.com or a pfm for um for the small business owner where um you know you could off access to all your different accounts your quickbook your you know your bank account um etc and then use our tools to sort of set budgets manage cash flow and strategically the idea or the hope was that like we would widen the top of the funnel.

13:24David Haber: We would begin a relationship with the entrepreneur with a way to help as opposed to a transaction. And then we would nurture that user eventually over time into becoming a transactional customer. And by having visibility into their financial data, we could push a product and say, oh, it looks like you're going to have a working capital gap in two weeks. Like push this button, draw down on this line of capital. and i think the core thesis there actually is much better positioned for today than it was then like the idea of sort of abstracting away the financial complexity of being an entrepreneur like most people start a company because they're passionate about the product or their service or their craft not because they want to be the cfo and so that was kind of the bigger mission of bond street was like we're not we're going to lend people money but in a perfect world we would actually just allow them to focus on the thing that they're passionate about and solve the financial part of their life.

14:16David Haber: And I think AI is like incredibly well positioned today. Oh, I haven't seen much. I'd love to. People are working on it. Yeah, that can kind of help coach an entrepreneur in anticipating the big financial challenges. And now maybe with computer use actually executing on some of them. So we're going to talk about the differences of building today versus years ago. But before we get away from what Jack Dorsey calls every company will be a mini AGI, right? Like the central computer brain. There are some drawbacks or there can be some drawbacks, right? Like consent needs to be very clear. And I feel right now it's very chaotic, right?

15:01I personally try to tell people even just a message. Hey, by the way, we're recording. Let me know if you want to turn it up. But it goes beyond that, right? If everyone knows that the meeting is being recorded, you think that they might be less genuine, less honest? How do we get past that while still kind of maintaining that institutional memory?

15:22David Haber: I mean, that is the downside for sure. You know, people still need, I think, a space for privacy, you know, to have the kind of authentic one-on-one conversation and be able to like share, I don't know, a personal challenge they're facing And like, you know, you probably shouldn't be recording that dialogue. But I do think that the Overton window has shifted pretty aggressively where, you know, you've been told for a long time not to put something in writing or in an email that you wouldn't want to have in a lawsuit. Like, I think the same is happening in meetings. Because, again, I think the default, whether you and now it's like whether you realize or not, is that most things are going to be recorded.

16:06David Haber: um and we haven't we haven't looked at them seriously or invested in any but i think it probably creates the opportunity for like the anti-recording devices oh absolutely yeah i was actually thinking about this i have a few friends who come out of the cia i'm sure you had to as well and the cia the nsa they have this i think it's called skiff yes yeah i'm i'm bullish on where is building skills totally where are those entrepreneurs no you may have like a little mini skiffs inside of your office yes yeah i think so i think this room might be a perfect candidate exactly uh good so let's talk a bit about a concept that you absolutely love uh and that's firm over fun yep uh because you know i think every time i talk to you that that is a relevant topic because at A16Z, you guys keep building.

16:57Yeah. Right. So, you know, first of all, tell us, I think I've never heard you talk about the KPIs that you track at A16Z, other than obviously, let's call it VPI, you know, the actual returns to investors. But other than that, what are some important KPIs for you, for the firm, given that you're building a firm, not just a fund?

17:28David Haber: Yeah. Well, look, I first started saying this as a way to sort of articulate where I feel best, which is sort of somewhere in between being an investor and an entrepreneur. I get a lot of energy running around and hopefully making good investments. But I also find that I need a place to channel my builder energy. hence why I think I'm so bullish on this AI you know, Gentic coding, you know, product right now but it's informed kind of the aesthetic I would say of the style of firm that I most resonated with and ultimately why I'm here which is to say most people are running funds and I think the objective function of a fund is how do I generate the most carry with the fewest people in the short amount of time possible by the way, you can be unbelievably successful running a fund.

18:16David Haber: The second, a firm, you know, in contrast, like certainly has to deliver exceptional returns. That is sort of a prerequisite, you know, for doing either well. So ultimately like DPI is, you know, what is required, you know, at the limit. But the second variable I think is the other really intellectually interesting one, which is how do I build sources of compounding competitive advantage? Like what are my moats? If I think about Andreessen Horowitz or an investment business as a company, not just a fund, what would I do differently? Right. And I think that question is a very entrepreneurial one.

18:54David Haber: Every entrepreneur wakes up every day asking themselves that same question. What are my source of comparing advantage? What are my moats? How do I how am I building a defensible business? But I don't think every fund manager does. I think they think about the next incremental or marginal investment. Um, and, um, you know, and I think the firm, like, Andreessen was always sort of like one of the archetypes and sort of thinking that way, in my opinion, because from the very beginning, Mark and Ben, instead of paying themselves, basically took all the management fees and reinvested into the operating platform.

19:24David Haber: So one expression that was, you know, the operating capabilities of the firm across go-to-market, marketing, HR, talent, you know, now media, um, uh, you know, and how do we sort of put that war machine to bear for our founders right and the firm is now about 600 people the vast majority of our headcount is in the operating team you know and so it's not a one talent person over here it's like hundreds of millions of dollars every year that we're spending again to tread tilt the board in our founders favor um but if done well and it can change the outcome of some of your most successful companies like it pays for itself obviously meaningfully um the other one that i think the firm was early to was was content was you know exactly what we're doing right now absolutely um where you know how do you sort of um build brand awareness in some way is it sort of like a is it relatively new firm i mean people forget but andreessen isn't that old it's an organization that started you know 2007 2008 uh time frame I'm not like the decades old firms of most of Silicon Valley.

20:30David Haber: And I can't take any credit for their success, but I think they've done an amazing job sort of like jumping into that echelon of conversation quite quickly. But to me, the question goes beyond just venture capital. I love teasing out this question even in other asset classes because I think there's lots of amazing examples of amazing entrepreneurs who build sources of commanding advantage in other disciplines. And it's not to say that we are trying to become them, but I think with humility, we want to learn from other people and pull the pieces that make most sense for our culture and our firm.

21:10David Haber: And it's one of the reasons I've started my own mini podcast series, you know, chatting with people like Tony James, who built, you know, DLJ and then built Blackstone and Mark Rohn, who built Apollo. So, you know, in their case, like they built compounding and managed through permanent capital, right, through Athene and their kind of balance sheet. And, you know, effectively having a lower cost of liability and originating, you know, proprietarily sort of higher spreads is sort of this like wedge that that business creates. And they're just compounding over time at scale. Anyway, so I just like that intellectual question.

21:48David Haber: it doesn't easily in my opinion boil down to kpis in the short term it's more of a culture it's more a relentlessness and a very entrepreneurial nature of the firm where we ask ourselves constantly like what is broken what is working if it's broken we're not afraid of tearing things down and rebuilding we're not afraid of building an organization we're also not afraid of like taking risk and putting a bunch of chips on the table to go hard at something you know like building a big new media team or going deep on AI right now. Because I think there's a belief that that will have compelling advantage.

22:25Yeah, and you've mentioned a couple of very large asset managers in different categories, right? It's gotten to the point where at least the mega VC firms, where obviously guys are in that category, there's almost like the lines are blurring between some of the asset managers maybe in private equity credits yeah you know so on so for even publics right um you guys are approaching 100 billion wait i'm just over that yeah just over um and and so how does that evolve uh you know the first evolution i guess going from cottage industry to more professionalized And now there seems to be almost two worlds or even multiple different products within venture capital.

23:21How do you envision the next five to 10 years?

23:24David Haber: It's a great question. And look, I think different firms will explore this differently. And I have a lot of respect for people kind of pushing the edges of what's possible and sort of exploring the frontier of what venture capital actually means today. I think to Mark and Ben's credit, though, they've done a very good job. And I pushed them on this and they've done a very good job pushing back in some ways of making the main thing the main thing. Which is to say, you know, I think three or four years ago, I was like, shouldn't we go build like an AI private equity arm? You know, it seems like pretty reasonable to assume that you could like acquire and transform a legacy business with technology and change the margin structure.

24:07David Haber: and those could be interesting canvases for other early stage portfolio companies to sell into. And, you know, Mark's retort was basically like, if AI is going to be the biggest technology of my life, of all of our lives, like, why are we going to screw around with credit equity? Why don't we just do 10x more of what we're doing, you know, and just like put, you know, billions of dollars more into our best companies, basically, versus like hiring and like firing and dealing with the kind of organizational complexity of like owning an asset.

24:43David Haber: And I think they've been right. It doesn't mean that that opportunity doesn't exist. And I actually think what, you know, to some degree like Joan Callis has done with their opportunities fund and build it doing kind of like this AI roll-up strategy, what, you know, my friend Josh Kushner has done with Thrive Holdings and now maybe less so with Eternal, but like, you know, in a permanent capital capacity, you know what my college friend Alex Topman is doing with Long Lake I think are just like amazing expressions of those ideas I'm not sure that that's what we will do never say never but that's not certainly the immediate focus we want to just work with the best kind of early stage and growth founders in kind of pure play tech investing but again like I think what GC and Thrive who I would argue are kind of pushing the frontier or most aggressively right now are illustrating is that venture capital has already and will continue to sort of further industrialize in some ways in the way that private equity did over the last 25 years, I think in a much faster timeframe in the next like five to 10.

25:50David Haber: And to credit Mark, like software's eating the world. I think AI is eating software. And so the impact of technology in every facet of our economy is only going to continue to grow. and I think there will be more convergence and overlap. Like even in the conversation with Mark Rowan, you know, he wants to provide the debt to the big data center build outs and like the capital requirements of some of these companies are way beyond venture capital, right? They're way beyond just equity actually. And so I think they come at the world from a very different place than we do, but there will be more and more overlap over time.

26:27David Haber: And I think that's just going to be a very interesting dynamic. Does that convergence manifest itself also in risk for LPs, right? Is it all converging mainly about AI? Or how should LPs think about it? Um, it could. I mean, he made this argument on that podcast that there's actually more concentration risk in the public markets today than in the private markets. You know, basically the biggest public tech companies, like the debt issuance that they are making to fund most of their compute needs is concentrating like investment grade debt into like four or five names. and actually to build diversification, you need to go into the private markets.

27:14David Haber: This is his pitch, obviously. But I think it's true. And it's why they've spent so much time focused on other strategies. But I don't know. It's hard in some ways to invest in technology and not be investing in AI because it's just becoming this sort of like intelligence primitive that I think is going to be embedded everywhere. So are there going to be bottlenecks? Could there be air gaps in the evolution of this technology or the proliferation? Maybe. But I think in the longer arc, we're still unbelievably early. And it seems like the demand for intelligence is like infinite TAM. And so I still think we're so early in that curve.

27:59Who doesn't want more intelligence, right? Yeah. So let's talk about building a company these days. is it and you've talked a little bit about this is it easier or harder than ever to be a non-technical founder I think it's

28:21David Haber: well it's interesting I think I think it is easier in some ways to build a company as a non-technical founder if you are intellectually curious and you become fluent in the tool sets like you know I think you need to understand how LLMs work and how to apply that technology to jobs to be done.

28:44David Haber: I will still say that the scarce resource is that, though, which is to say people's understanding of how to apply all items to jobs to be done is not like evenly distributed. And so one reflection that I've had is that five or six years ago, if you're investing in a vertical software business, you probably wanted the person deep in the industry, right? Coupled with a technical co-founder. That was like the ideal archetype of a founding team. It's like the person that came out of construction knows all the players, plus like their engineering friend from school or whatever. You know, now I think what we're seeing more often than not are younger and more technical founders who are less deep in the industry, but they are fluent in the tool, in the tool set.

29:36David Haber: and that there it doesn't mean that the context in the industry is relevant actually i think it matters a lot and more maybe than ever but they're hiring for that context early um just because again today the understanding of how to apply the l1 into you know a health care context or construction or you know finance is that is the the rate limiter um i think the pendulum will shift back as like people's understanding of how to use ai um becomes more ubiquitous um but at least for now that's sort of been an unusual thing like some of my you know fastest growing companies um were founders that were not deep in their domain of expert or they ended up focusing their vertical application um but they've done unbelievably on they become fluent on that industry as well it's almost like the question used to be do you have a cto among your founding team now it's like you have an industry expert right amongst your founding team yeah and and um thinking about that how how is is software really in in from when you're seeing your portfolio companies um kind of like how the Carlson brothers have said it's more like pizza you know okay it's only good if it's fresh you know yeah are you seeing that level of iteration and portfolio companies, I think at the earliest levels, it makes sense.

31:02It becomes much harder at the more mature levels.

31:06David Haber: Yeah. I mean, I forget who said this, but somebody said it was like building sandcastles. You know, that you just recognize that you may need to rebuild, you know, mostly because the model capabilities are changing so quickly. And so what was possible a month ago or six months ago, certainly, like changes really dramatically. and I think ultimately what you want to be for many of your customers is the sort of like translation layer of like the context of their work and their industry and the job to be done and like the capabilities of the model and like the distance between model capabilities and that is the application layer.

31:43David Haber: Right, this is how I often talk about like why the application layer needs to exist. It's like one-shotting a prompt in a model in the context of like even my job, like it'll generate something. but like it's the harness it's the guardrails it's the e-bails it's it's all the product in between that actually it makes the output relevant for a gp at injury center whatever right um and and so you need to kind of manage both of those lenses like you need to continuously deeply understand your customer their context their job their work but then also stay current in my opinion on model capabilities and that largely in a perfect world you could just swap and new model capabilities and you have sort of a robust eval system internally in the company to know, okay, this new model just was released.

32:30David Haber: How does it benchmark against the prior versions and what is now possible? But in some cases, it may require you to kind of rebuild and re-architect that end-to-end system. Hopefully not from scratch, but large chunks of it. We also see it amongst the technical founders. No one is keeping up to date with the latest releases like they are. Yep. Right. What do you think of this concept that companies are just going to in-house most of their software tools? There's a spectrum, but how do you react to that? I think this conversation comes up every like 20 years or something. Like Goldman, you know, famously had like built their own email client called Orbit.

33:20David Haber: It's like, why is Goldman Sachs in the business of building their own. Like they didn't use Outlook or Gmail for a long time because they need their own proprietary email client. That doesn't make sense. It doesn't mean that companies shouldn't insource and build the things that are core to their compounding advantage or their kind of intellectual property. Again, because software is becoming cheaper to build, you'll probably see that more and more. Like more things will get built than ever. At the same time, I think like focus matters, right? Somebody who's 100 % focused on building, you know, voice models, like you should probably just rent the voice model from the best voice model company in the world and not try to like build it yourself because, you know, whatever you build is going to get commoditized in six months as they kind of push out the frontier of what's possible.

34:10David Haber: And so, again, I think like figuring out like what are the things that are your core moats or your source of defensibility and like pressing hard there and probably borrowing or renting from other people in service of that moat is probably the right answer.

34:33David Haber: The flip side is kind of like where we started this conversation is like you can really build personal software and you can like, there is no market for the David Haber, like VC operating system or a limited one, but you can build something uniquely tailored to like my personal preferences and my unique workflow if you want and that is also equally exciting but things work a shit ton of work yeah things break this is not like one shot all of a sudden it's it's functioning you know properly yeah you uh and you learn that it's actually made me more bullish i would argue on the application layer having actually gone very deep in building you know software um because i think a lot of the conversations like oh my god you know entrofic is going to eat everything you know the the labs are going to eat every vertical and build our own applications and um i just don't think that's true yeah yeah specialization still matters yes david what happens to our beloved fintech industry in all this context because you know five years ago it was very clear if a company was a fintech or not yep that's not the case anymore yep look i've always taken uh an expansive view of what i consider fintech um i mean angela you know run every company's a fintech company which was uh uh true for a long time um but i think you know i've been doing fintech since 2011 right so i think people's definition of fintech historically were financial product-led companies right like the first wave of fintech was basically unbundling the bank starting with a better version of trading or a checking account or a loan putting it online or into a mobile app building meaningful distribution and then bundling other financial products to try to own the end customer relationship and i think many of those companies have done unbelievably well robin hood chai and revolute you know you can go on and on and on um they benefited from platform shifts they benefited from you know being early on mobile when like jp morgan didn't have a mobile app they benefited from the cost of acquisition being low or lower on sem and social um and so i think is just very, very difficult to build the next Robin Hood today.

36:58David Haber: I hope that's not the case. I hope AI somehow unlocks a new consumer preference that changes the dynamic and sort of explodes. And maybe Kalsh is just on this in prediction markets, who knows? But I think the game is largely won. I think there are lots of interesting investment opportunities, but they're mostly our growth colleagues who are investing in Revolut and other other companies at a later stage. I think where I've spent a lot more time, even pre-ChatchyPT, and certainly more now, is in software-led fintech companies as opposed to financial product-led fintech companies. And part of this, candidly, was born out of my own PTSD of running a lending business, where despite what I described about Beacon, the core of the business was a loan.

37:49David Haber: And at the end of the day, you're selling a commodity. And I didn't have a cost of capital advantage certainly versus a big bank. And I didn't have a distribution to manage. And so that becomes a very difficult business, in my opinion. I'd much rather compete with banks on software. And there's infinite numbers of problems, I think, increasingly large problems to be solved with AI in just the way these firms run. And you've seen that early with a lot of success in areas like customer support, you know, in loan servicing, in collections, you know, many of these large financial institutions have thousands and thousands of calls, you know, folks in a call center answering questions about their credit card payment or doing welcome calls or payment reminders for, you know, auto loans or other things.

38:38David Haber: AI is incredibly good at that, you know, because you can speak in 50 languages, right? You can have the agent who speaks Swahili hiring that person for like the 50th, you know, whatever, 1 500th a percentage of your customer didn't really make sense. Now it's free, right? The marginal cost of serving every language basically is free, effectively. And they're infinitely patient, available 24-7. And so I think that's been an early wedge, you know, for a lot of these AI companies. But I think we're quickly moving into a new frontier, which is longer horizon agents that are solving much more complex problems.

39:19David Haber: So, you know, customer support was sort of a bounded one. But if you think about like the process of originating a mortgage or a student loan, that might take months. And there's all different sort of off ramps and leaky parts of the funnel where it's, you know, follow up or missing documents or credit checks, you know. And you may, humans historically have done this, like, you know, nurture that lead from a first email, you know, through text, voice, outbound calls, you know, and so that funnel is not perfectly linear. It's actually like, I think it'll zigzag across channels. And I think the models have gone to a place with the right harness to actually have shared context and nurture that lead through months of a funnel and ideally get paid on an outcome right because that whole supply chain that i sort of just described is very expensive to support with people you know both acquiring customer but also just like having marketing touch points direct mail like you know nurture that lead all the way down to actually like you know refining their credit card or taking out a mortgage or what have you so the willingness to pay for that agentically i think is like really high and I like that style of FinTech company.

40:40David Haber: I could go on different examples of FinTech, but I think just the idea of solving work inside of large financial institutions is where we are spending a long time. Do you think financial services is actually one of the most attractive spaces to disrupt? I mean, certainly the profit pools are there. 100%, and I'm biased because we're FinTech guys, but I'm also biased. But no, I think it's true, right? I mean, financial source is probably one of the most human capital intensive industries on the planet. You have thousands and thousands of highly paid people living in Excel. I certainly saw this at a place like Goldman.

41:20David Haber: Not using Excel, by the way, as a modeling tool, but using it to track work across legal compliance, risk, banner onboarding treasury. like so much of how these the biggest financial institutions in the world and the long tail by the way like operate is very manually um and i think like where what ai is quite good at and kind of the meta theme that we've been spending time on is um it's beginning to capture labor budgets not just it spend and so the tam the addressable tam for many of these software companies is massive and if you think about you know the big banks have pretty big tech budgets they still spend way more on all the other headcount combined.

42:00David Haber: And so how do you sort of both help on a cost reduction, but even to what I just described, how do you help drive revenue? And so I think there will be massive companies built across all different layers of financial services. I think we're very early in this. So in that context, How is Andresa Horowitz structuring your funds? Because before it was very clear, like gaming, software, fintech. It's very different now, right? I mean, I know a little bit, but I'd love to hear it from you. It's not perfect. And actually, like, one of my favorite things, I probably said this to you before, is like that opportunities live between fields of expertise.

42:45David Haber: And in some ways, I like the things that aren't easily bucketable. Yeah. It's like a good thing for me. that if it's somehow falling in a crack between our funds. The way we've organized it so far is we've split the venture fund into three. So we raised an AI infrastructure fund, which my partner, Martin Casado, leads with a handful of other folks. So investing in things like data infrastructure, cybersecurity foundation models, the kind of overarching way to organize that is like anything sold into the developer community. So something with like a computer science primitive or the end buyer is somebody very technical.

43:21We have a new business we've been building

43:23David Haber: called American Dynamism, which is supply chain, energy, manufacturing, public safety, basically things with atoms, not just bits, or atoms and bits, right? Not exclusive, but in large part, they sell into government, right? Into the Department of War or state or local, federal government. And so that's sort of its own practice. where I know sit in, you know, my colleagues, Alex and Anish help lead is, is our AI apps fund. And so you can kind of think about that as, you know, if this is sort of B2D, like business to developer, this is B2G business to government, we are B2B and B2C, you know. So anything consumer still lives within the apps fund.

44:10David Haber: And then B2B covers lots of industry verticals, legal home services, you know, et cetera. And then horizontal software. So things that are sold into the kind of functional buyer that could be true across companies. So the CFO, the CMO, the CRO, the COO. And so that's kind of how we think about that. We also, it's somewhat complicated, but we have a bio and healthcare fund that does everything from like hardcore biotech and, you know, therapeutics to care delivery to health tech. The health tech is probably the closest thing that sometimes bleeds into apps. Crypto, which is the floor that we're actually sitting on right now if you're here in Indreuson, is becoming, I think, I would argue increasingly more fintech because things are centralizing more around stable coins and tokenization.

45:00David Haber: And then the kind of other horizontal beyond speedrun is our growth fund that sits across the entire kind of early stage platform. And so when things are working, you know, my joke is that we are the customer acquisition and they are the LTV. or they're the monetization. So they will write much larger checks in a much later stage. So that's, again, at a high level kind of how we've organized. I think it's not a perfect science, but that seems to be working at all so far. And you're always experimenting. Correct. Yeah, yeah. We're in your New York office. It's one of your biggest, not your biggest.

45:37Did you read a report that now in New York has more tech jobs than so far in the Bay Area? I haven't read it closely, but I saw the headline. Yeah, yeah, yeah. I mean, can we also... I read one or two paragraphs, Max. Yeah. But, I mean, it's a reality that tech is becoming one of the most important... And also New York is becoming one of the most important tech hubs in the world, without a doubt. I mean, international founders, they now want to build either Silicon Valley or New York. Come on. Right. They don't want to go to London. They don't want to go anywhere else. You are New York through and through.

46:18Right. Do you think this keeps growing or, you know, how do you have that conversation between most of your firm that says Silicon Valley versus New York?

46:31David Haber: I think it's hard to bet against New York, you know. and that may be more controversial today than ever because of politics and fears of AI and knowledge workers or whatever but I still think that if you're a ambitious young person New York is one of the most attractive places to live and I think part of that comes by the way from the heterogeneity of industries and people that live here you know back to my kind of like opportunities live between fields of expertise the thing that I love most about the New York tech community in particular is that and the most new york companies in my opinion are those that live at the intersections of technology in one of the big incumbent industries whether that's finance or advertising media or health care and you know increasingly um i think what the the knock historically on new york was we lack engineering talent i think that's actually been mostly resolved and i think actually with ai it's it's less of the issue i think the biggest challenge the New York has is actually the, I would argue like the executive layer, which is like the VP of marketing or the COO who's taken a company from like 100 to 1 ,000.

47:37David Haber: It's the hyperscale. You know, you just have a much deeper bench of talent for that in Silicon Valley because there's so many more companies. The other thing that I think Silicon Valley does better than anywhere else in the world is deeply technical problems, right? You know, hardware robotics, like a lot of the frontier labs, I think there's a reason many of them are based there. I think that'll get more diffused, but with AI, we've definitely seen momentum shift back to San Francisco and the Bay Area broadly, where that sort of like human network effect was beginning to unbundle in COVID, and a lot of people came here.

48:16David Haber: We were big beneficiaries of that. But in the broader arc, I've been in New York since 2009. It was a much smaller community in the wake of the financial crisis. And being in tech was sort of non-consensus. I think that's flipped pretty dramatically. I think we'll continue to flip because again, like software's eating the world and you have the biggest companies in the world have a meaningful presence in this backyard. Oh yeah. So before I let you go, let's just do some rapid fire. And you can't pick this one, best block in New York City. i mean bond street i still own bond street.com oh you wish i'm in a phoenix from the ashes at some point uh but i i chose it because it was sort of an allusion to finance but it's also my favorite street new york city i think the cobblestones and the architectural design is amazing and it keeps getting more and more fun that's amazing right yeah so uh one thing that you believed about the industry tech industry when you joined a 16z that you don't anymore

49:33David Haber: um i think um i think the thing that i changed my mind on in some ways is um the scale of ambition You know, I think there's a common principle in venture and a lot of discussion historically on like fund sizes. And I think, look, the laws of physics of returning big funds is really hard, no doubt.

50:03David Haber: And if you have a smaller outcome, it can have a real impact in multiple and a much smaller fund. And I think the thing that Mark and Ben had changed my mind on is like your fund size is sort of your conviction on the future. And that actually like the TAM for software companies are going to be way bigger than I even anticipated coming into it recently, even five years ago. Right. The fact that we have trillion dollar plus companies going public right now is like unfathomable. Multiple, right? Yeah, multiple. I mean, it's crazy. you know the fact that you can have big M &A in very short time is unfathomable and so it doesn't mean anything's easy we don't take anything for granted but I think like if you're really convicted on AI and really convicted on the future putting a lot of chips on the table against that is has been something I really admire and I think they'd push me to think much bigger in that capacity.

51:07A founder or investor that is underrated these days?

51:14David Haber: Man, so many. One of the most rewarding things of being an investor at the early stage is sort of having deep conviction in the individual far before the market realizes, far before it's manifested in any sort of metrics or anything legible. And, you know, for most of the companies, I was like the first investor or the first board member, certainly. And I think many of them are still early in their journey. You know, because this is a fintech podcast, I mentioned some of the fintech ones. I think Moment is a deeply underrated company still. You know, we sell capital markets software into the biggest financial institutions in the world.

52:00David Haber: I've been very involved in helping them build that business. and we've announced that we now power LPL, Edward Jones, Hightower, or many, many others kind of in the wings, some of the biggest in the world. That moment is now powering portfolio construction compliance risk far beyond fixed income into equities and other things. And I think it's a good example of a company that actually didn't start, I would argue, with an AI wedge. They started with the harder problem, which is how do we become the core kind of data model and the execution system for all of your clients and all trading? And I think if you, as they've done, earn that right to sort of aggregate that information and be the kind of core system that every client's portfolio gets built upon and all compliance runs and risk runs through.

52:59David Haber: layering in other agenda capabilities on top to power things deeper in the middle and back office for these financial institutions, maybe up the funnel for clients or the advisor, let's say, is very ripe. But I think that part of the story is still early. what's a recent article or insight anything that you've read that really stuck with you a problem that you think about an insight anything like that um it's mostly been through conversations recently um but i've been again kind of formulating this thesis of like uh effectively the case for an AI native Red Ventures. So there's this amazing company based in Charlotte called Red Ventures that was started in I think the late 90s or early 2000s by this guy named Rick Elias.

54:04David Haber: And their thesis was basically let's own the sales and marketing funnel on behalf of the highest LTD, kind of most considered products, right? So things like, I think DirecTV was their first customer, but Telco, financial services, insurance, these products are often sold, not bought, or again, they have this sort of leaky funnel problem, and they basically got paid on a CPA basis for managing that entire funnel. So they would take a lead, and instead of it being run through American Express's sales org, they would do it on their behalf, and they would do it much more effectively and they would get paid on an outcome.

54:47David Haber: I'm still researching this, but over the past 15 years, they've started actually moving further up the funnel and actually buying the demand gen itself. So they own bankrate.com, they own the points guy, they bought CNET, I think they sold it. But there's a whole media property angle that exists to kind of own the SEO effectively for these high LTV products. I don't have any visibility into how the business is run today candidly, but I think they're very at risk of AI because SEO is changing, discovery for all things is going now through some of these big model providers, so things like GEO or whatever you want to call it are shifting that dynamic and then kind of the earlier conversation, agents, like AI agents, I think are much more capable in some ways of owning that end-to-end funnel with shared context across channel and so instead of having thousands of people sitting in charlotte managing a very linear funnel from like you know a direct mailer through to like an e-commerce checkout um actually i think an ai agent doing it across voice and text and direct mail and uh is actually going to be way more effective um in every language like for every product um and it's a huge company the most people want to know about it's a private business And so I think like people will build vertical versions of Red Ventures in lots of categories, having these sort of long carotid agents that solve this lithophanal problem.

56:13David Haber: And so I've just been trying to chat with people to go deeper on that. And I've been listening. I think Patrick O'Shaughnessy had a podcast with Rick from several years ago. And so I've just been going down the rabbit hole, kind of pressure testing this idea. It's a good topic to end on. David, thank you. This is very fun. a lot of fun. Thanks for doing this again. My pleasure. I'm glad we did it in person. Yeah, me too. It's time. And yeah, man, let's keep betting on the future. Likewise. Thanks for having me on. Thanks. Thanks for tuning in and I hope you enjoyed this great conversation with David Haber from Andres and Orwitz.

56:51If you want more interviews, please make sure to subscribe, follow, and leave a review on Spotify, Apple, YouTube or wherever you get your shows. It helps and means a lot. And if you have any suggestions or thoughts about the show, just drop me a line on LinkedIn. See you next time.