The Exchanges

Q Um, couple more questions because I want to open it to, to, um, people here. Um, cloud versus on-premise, is that a relevant, is everything moving to the cloud? You mentioned transfer of data, Matt.

A I was, I was gonna raise that, that, that point earlier. It's, it's, it's a huge issue. Especially when you're talking about the enterprise, especially when you're talking about, The three letter agencies, or financial firms, or even healthcare, where the data is so incredibly valuable. There are huge privacy and security issues around it. There are a lot of places that unless you can give them an appliance, or you can load it on their servers inside their firewall, they're not going to buy it. They are not going to put their data in the cloud. I would say there, there is growing acceptance of that and comfort with security around that. But there is still a huge business for on-premise solutions, and depending on what you're, you're designing, you may need that kind of fork in your development path.

AI assessment note: “there is still a huge business for on-premise solutions”

Q do still sort of falls in this bucket, so I, I, you know, I read somewhere some of the themes that were Excited about, and, um, so in a particular order, uh, information businesses which exhibit data network effects, so including the Trade Desk, and, um, InfoSum, and Signify. Do you want to talk about maybe data network effects? What, um, that means, and, uh, why it's a desirable characteristic?

A Sure. I mean, I think we, we actually have lots of businesses that exhibit data network effects where, you know, as you, as you add clients, as you add data into the system, that the system becomes better for, for all participants, and that creates a competitive moat that is insanely valuable and hard to fight against. So, and I think we've become more sophisticated about that as time has moved on, you know, whether you're talking about something like, You know, signified and, and merchant fraud, and having an ever larger data set that enables you to make better predictions, and even, in their case, to step into the shoes of the merchant and assume the risk, right? I mean, that's, I think, the next If not the next frontier, it's certainly something where we've seen very powerful businesses be built, where they are, companies are actually stepping into the customer's shoes and saying, for a fee, we'll assume the risk that you've historically assumed, so you can focus on the thing that you're best at, like serving customers. And so I think that's one very powerful example of data network effects leading to companies being able to do things that were previously unimaginable.

AI assessment note: “as you add clients, as you add data into the system, that the system becomes better”

Q this whole concept of, of being founder friendly. Uh, and I guess that's, you know, position to what, um, happened maybe, you know, a decade or two ago where, um, I guess the power dynamic between investors and, and, and founders was different. Uh, but I think you, you have some interesting thoughts on, on this, um, When is founder friendly too founder friendly, and when do you abdicate responsibility?

A Sure. Yeah, Matt's really hitting on these, uh, these tough points. Um, I think I write too honestly. Uh, so, so, yeah, founder friendly. So I think founder friendly has Become identified or, um, associated with being permissive and being loose as it relates to documentation, as it relates to founder expectations, as it relates to founder egos. I think that is actually being very founder unfriendly because especially when, like I'll say in our case, most of our founders are first time founders, you know, like we're not Sequoia and benchmark and we're only eight and a half years old. So we don't have the stable of entrepreneurs that have exited that are on their onto their next thing. So by definition, we are largely dealing with, dealing with first time founders and Even repeat founders. They have a lot to learn, and we all have a lot to learn, together. There's some things they know really well, like the product and the vision. There's some things we know from having this enormous data set of having worked with lots and lots of companies and seen lots and lots of failure modes. And so, what we work really, really hard on, again, by spending a lot of time up front with founders and building that deep trust relationship, that enables us to be very honest and very blunt along the way. It's not personal. It's, we're trying to help founders be their best selves. That's not really p…

AI assessment note: “founder friendly has Become identified or, um, associated with being permissive and being loose”

Q and CTO of, ah, Quid, which is another very interesting Silicon Valley startup. So, Roger, maybe to start with you. So, data itself is, is nothing new, right? It's been around pretty much since computers have existed. There already have been businesses like Bloomberg or Axiom that were created on the business of aggregating and selling data. So, what is new? Why is it such a big deal these days?

A I think it's the, the combination of the sheer magnitude of the data that we now have at our fingertips, as well as the velocity of the data. So we're talking about, you know, web scale where, you know, real-time customer transactions, searches, interactions, connections on, on the web, in addition to, in the Wall Street applications of, you know, high-speed trading, where, you know, the only limit is the speed of light, and then you've got the storage element. So, Uh, on the one hand, you've got, you know, things like, you know, EC-II and AWS, which facilitate, you know, the storage of massive amounts of data without your own, um, your own big iron architecture, and at the same time, the, the CPUs to be able to process that magnitude of data has just upped the ante to such an extent that we couldn't have imagined even 10 years ago.

AI assessment note: “combination of the sheer magnitude of the data that we now have at our fingertips”

Q Have you seen, uh, customers, uh, react negatively to, to this, saying, hey, you know, I don't want you to, you know, whatever you learned from me to then go help my competitor or, like, some other company in the industry, and then what have your company's done to mitigate that risk?

A Generally not in the, in the companies that we, um, in the investments that we've done. There are certainly times, especially, I would say, Like, in financial transactions, generally not because you're abstract, you're, um, everything's anonymized, and it really is driving benefit for everybody, it's not, and the numbers are so large, it's not about specific records. But in things like, for instance, um, health, health tech companies that are looking to build panels with very, very sensitive data, you know, all, you know, PII, That's a very, very different set of problems, and needing to think about how to clean data that could potentially be merged. That is a very, very hard problem, and actually, I've actually made an investment in a company that specifically addresses that issue.

AI assessment note: “Generally not in the, in the companies that we, um, in the investments that we've done.”

Q spoken about big data a lot, and, and to that point, let's talk about small data for, for a second. Is there value in small data? If I'm, you know, a, a member of the audience here, and I run a, I don't know, a fashion e-commerce website, do, can I do interesting things with, with data without having a, you know, a PhD or, or data hacker on board?

A Yeah, I think, well, I, The question is, what's the precursor to the small data? If the big data is transformed into small data, for instance, you know, Joe's cloud score. Right, so there is a lot of value in that single metric, but behind that metric, you know, is a lot of stuff. So, if you have kind of the right, the right metrics, then yes, I think small data can be unbelievably powerful, especially in the influence realm, because there are plenty of circumstances where Wisdom of crowds and large data don't give you better answers. In fact, it's much more valuable to go to a few extremely knowledgeable people to get a particular answer. So that's where influence together with domain, so it's not just popularity, but it's with respect to a particular domain is where kind of small data becomes most valuable.

AI assessment note: “yes, I think small data can be unbelievably powerful”

Q Is that because the moment has passed for those businesses?

A We're very happy that we made those investments, but I also think that the way that we look at, so I've written about all this stuff, but just to paraphrase, like, we make relatively few investments each year. Each of us make two to three investments in a given year. Every investment that we are making, we think has the potential to be a multi-billion dollar outcome. Every single one. That, that is the bar. With that lens, there are not too many businesses in that realm that clear that bar, and those are three really interesting, successful companies. I do not think one of them is going to cross that bar, even though they'll be very good outcomes. We found in our, kind of, the power law reality of venture that we would much rather take 24 gigantic swings in a portfolio over a four-year period and work Hopefully that two will be multiple fund returners and that the next six, so call it two super performers, six good performers that cumulatively return one to one and a half X the fund and then 16 Some, some kind of outcome to shut it from shutting down to a low return of capital. That's kind of the model.

AI assessment note: “I do not think one of them is going to cross that bar”

Q Very good. Uh, you want to talk about, you wrote some very interesting stuff about the, the type of founders, especially, you know, in that context of a, you know, people that can build Multi-billion dollar companies. You had a couple of things that really caught my eye on the founder qualities. The first one was a DNA that embraces experimentation. What does that mean?

A So, yeah, we invest very, very, very early in a company's life. Again, very sparse data. Sometimes we invest, you know, again, Trade Desk was a PowerPoint. I wouldn't say that's normal, but certainly Uh, very, very, very early product. So it's really around kind of aligning on the hypotheses. That's really, that's really what it comes down to. Um, but a lot can go wrong, and a lot does go wrong all the time. So, being able to have founders who, by their nature, want to run a series of constrained experiments to test these hypotheses and then to rapidly iterate as they collect data is, in our experience, an essential element to be being a successful founder of a early seed stage company that has the chance of generating the outcomes that we're seeing.

AI assessment note: “want to run a series of constrained experiments to test these hypotheses and then to rapidly iterate”

Q set. Where do you guys, I mean, why is fundamentally a big data investment interesting from your perspective? Is that, you know, because there is this crazy demand out there, or are there, you know, intrinsic aspects to a data business, whether that's economies of scale, network effects, what would have you that, that make it, um, you know, different and sustainable in a way that other businesses are not?

A We, we believe that it, that they create unique barriers to entering, and that whether it's the proprietary data asset, the unique data that's generated by That flowing across the platform and using machine learning and AI to improve the customer experience, or if it's solving a really, really, really hard technical problem that, you know, and one other point on the, on the enterprise, you know, I'm seeing a lot of companies now that are specifically engineering solutions that don't require a rip and replace that can run alongside legacy infrastructure until that infrastructure can then be replaced, but it's, uh, It's a smooth transition, not an abrupt one, which of course makes the Bardo adoption exponentially higher. So for us, it really comes down to barriers to entry, whether it's pure technology barriers, or it's kind of the data asset that's created through the creation of a really smart platform.

AI assessment note: “for us, it really comes down to barriers to entry”

Q And then the other exciting exit recently was Flatiron Health. So we, uh, had the pleasure and honor of hosting Zach here maybe two or three years ago now. Um, so terrific story and terrific, uh, New York story as well. Any, any story from, from that?

A Uh, a little story. So I was also an original investor as an angel and invite media. So, um, I had been introduced to them by I guess it was Josh Koppelman, because first round had led that round, and we had done some investing together, so one thing that was kind of left out in the background is, so I was on the street from 87 to the end of oh four, and then from early oh five to late oh nine, um, as I was setting up IA, I had invested kind of professionally as an angel, and set up an investment vehicle called IA Capital, and put a lot of money out, um, I Seated 40 companies, led six rounds, sat on six boards, very, very active in the angel community here, and so had met Matt and Zach and was, and also knew a lot about ad tech and was extremely impressed because they had really articulated a strategy that was directionally similar to ultimately Jeff Green's and the Trade Desk, but they obviously, you know, sold out much earlier to Google, but they're now, I mean, a huge part of their core infrastructure. So, I had built that relationship, seen that successful exit, and when they started Flatiron, we were, you know, honored to have the opportunity to support them again.

AI assessment note: “when they started Flatiron, we were, you know, honored to have the opportunity to support them again.”

Q price, assume greater risk, absolutely. Okay. And, um, so for people here building data businesses, do you see, um, how do you build that, I guess, you know, building network effects is one of the hardest things to do. Um, how do you Bootstrap that and avoid the cold start problem, um, or is that something that you need to think about when you're in your five of your business?

A No, I think, I definitely think it's something that you think about at the beginning. Uh, I mean, not, not every business by its nature subjects itself to data network effects. Those that do, I think, can be especially powerful and especially valuable, but I think it is something to think very deeply about at the, at the beginning, as you are bootstrapping that initial data set, and you always have a, you know, it's very rare not to have a cold start, but there are, you know, obviously hacking that initial User base and hacking the initial data set to start to see are there ways to leverage kind of the, ah, the collective knowledge that you've accumulated in order to improve the product, whether it's the, you know, the price, the risk tolerance, what have you. So yes, I think it is something that you should be very intentional about as early as possible.

AI assessment note: “I definitely think it's something that you think about at the beginning.”

Q Yeah, that, that was exactly going to be my next question. So, um, I guess what's above this is analytics and what's on top is applications. Uh, so quick thoughts maybe from, from Roger and Matt on, um, analytics and applications. Seems to me that the analytics space is getting more crowded, um, you know, day by day. Um, at the same time, the application space looks wide open. Thoughts?

A Yeah, well, I, I think it comes down to this issue of kind of solving hard problems, and I think there's still huge, huge room to run. I mean, we backed Several companies in this, in, in this vein that sit on, sit on top and generate, you know, real-time actionable information, you know, at web scale looking at complex joints in the data that have been embedded in companies large and small now that are absolutely essential for the operate, operation of their business, which of course is the holy grail, is becoming, you know, something that you just can't, you can't rip out. Gives you the pricing power. So, yeah, by the way, the answer you guys gave- Like the Bloomberg terminal? What's that?

AI assessment note: “there's still huge, huge room to run. I mean, we backed Several companies”