Dmitriy Berenzon: Crypto Infra, ZK, and Working with Founders
Ash sits down with Archetype’s newest team member, Dmitriy Berenzon, for a wide ranging conversation. The two discuss the crypto ecosystems of NYC and SF, hardware becoming cool again, weird stuff at the intersection of AI and blockchain, and being both thesis and founder driven.
Dmitriy is the newest Partner here at Archetype, and will be driving research, investments, portfolio support, and fund strategy. He was previously a Research Partner at 1k(x), where he focused on research, investments, and working with founders on fundraising, business development, and token model design.
📬 To keep up with the latest from Archebyte and receive articles and research from the rest of the Archetype team, subscribe to our newsletter: http://eepurl.com/iCApL2
- - - - - - - -
1:05 Dmitriy’s background
3:18 NYC vs. SF crypto ecosystems
8:30 What Dmitriy is on the lookout for in 2024
10:49 Today’s crypto infrastructure
12:46 UX and abstraction
15:17 Building for consumers and crypto-natives
19:13 DePIN, DeVIN, and hardware becoming cool again
27:16 Modular vs. monolithic
30:34 Will Solana become modular?
32:46 ZK potential
39:16 AI meets crypto
44:31 Being thesis and founder driven
👋 FOLLOW US
Hello, everyone, and welcome to Archebyte. I'm Ash Egan, founder here at Archetype. Today I'm joined by my good friend and Archetype’s newest partner, Dmitriy Berenzon. Dmitriy, welcome.
Great to be here. First podcast I'm doing in person, actually.
They’re always better in person. Today we're going to talk about a lot of different things. But before we jump in, it would be awesome to hear a little bit about your background and what you were doing before joining the team here at Archetype.
Yeah, for sure. So I started my career in traditional finance. I worked for a bank for six years, first four doing a variety of strategy roles, last two building out their innovation labs. Heard about Bitcoin in 2013 when I was doing research for the bank's head of payments. Did a proof of concept with the labs in 2016.
Ultimately took me until the end of 2017 when I was in Berkeley for grad school where I finally had my “aha” moment, met the Blockchain Berkeley folks who are these 18 year old galaxy brains. Individually very inspiring, but they also had this good course on Bitcoin more from a distributed systems and game theoretic perspective. I looked at it from the lens of defi—kind of saying that the last 30 years of fintech innovation was lipstick on a pig, you know, pretty front end on top of 1970s rails. And at least with Ethereum you have 2015 rails. So I was really all in since then.
I worked on a startup for about a year, a portfolio management tool and robo advisor, ended up sunsetting that and working with a fund called CoinFund out of New York doing research and it was super fun. Ultimately wanted some more responsibility, so joined a crypto family office to lead research and investments there for two years called Bollinger Investment Group. Did around 25 investments with the firm. Smaller checks, you know, $50K-$500K.
What I particularly learned there was I enjoy working with founders, had the opportunity to join a larger fund who did more lead and co-lead. So I joined 1kx just about three years ago and had a great time there, you know, was doing research, investment, due diligence, portfolio support, probably bending more on the infrastructure side of things. And yeah, really got to see what an S-curve looks like for a fund and kind of what works and doesn't work and you know, now have the opportunity to do it again with you.
And I'm psyched about it. You're a New York native. You grew up in Brooklyn. You've lived the last few years in the Greater San Francisco Bay Area. You're in New York for some amount of time here with the team in person. Maybe just to start, like the ecosystem in SF, you know, we met back in 2017 and it felt like SF was the epicenter, you know, coming here to New York here in 2024. Give us a rundown on the New York vs. SF ecosystem. Are these the most talented places to build and invest in crypto companies and protocols? Why don't we start off there.
I think largely, yes. I mean, there's macro issues, I suppose, with being in the US, but I think both ecosystems are exciting. I think the benefit of the Bay Area is you have the effectively constant annual admission of galaxy brains coming out of Berkeley and Stanford and that is tough to compete with. I think they are great institutions.
You know, I've been kind of shocked at how developed the New York ecosystem has been. I used to run fintech meetups here back in 2015, and, you know, it was hobbyists, a lot of people wearing suits and ties. There were the Consensys folks, I suppose, in Bushwick, but there wasn't a gravity of founders, and I think there were a couple of path dependent events that happened. I think COVID was definitely one one issue.
I think when you look at innovation ecosystems historically, usually they need a certain success story to look at. This happened in Berlin with things like Rocket Internet where the German culture is more conservative and I think, you know, looking at Rocket Internet made it less taboo for people to get into e-commerce.
And I think historically New York has been number two. I suppose not as competitive when you look at which startups are successful, you know, maybe MongoDB on the deep tech side, Etsy on the e-commerce side. But the difference with crypto is I think we have Uniswap and OpenSea here and I think that provides a very powerful flywheel and I think it worked.
You know, when I came here, even just visiting in 2021, I was shocked. Even relative to me spending some time here in 2018, the gravity of founders and investors here, it has staying power.
I always like to call Spring Street Crypto Alley. Just given the concentration of crypto startups, venture capital firms and ecosystem folks. Maybe just comparing New York City 2024 versus San Francisco 2017 when we met. Do you think New York City is in a similar spot to where San Francisco was in 2017 in the crypto lens?
It's hard to say. I think the beauty of what was happening in San Francisco around 2017 was the Crypto Economic Security Conference, CESC. It was organizing a fund that's probably no longer operational, Decrypt Capital. It was a lot of the Blockchain Berkeley folks. That was just around the last bull cycle and that was really powerful.
After CESC, I think now it's called the Science of Blockchain Conference. That was around probably since 2017, 2018. It still goes on now. I think those are really powerful events because it's kind of a no shill zone or it really tries to be a no shill zone. I think CESC happened a year or two ago and they have submissions. It's whitepapers, there's a committee that reviews them. I think that is powerful.
I was surprised, pleasantly surprised, that SBC is going to be in New York this year. So it seems to be that there is more academic rigor, which I think is great. You know, I think what New York probably needs is more horsepower on the distributed system side of things. And I think once you start to have the more hardcore academic conferences—and there's Cornell IC3, I suppose that happens here too,might be upstate New York—but I think having more of that gravity happen in New York City proper, it'll attract more of the technical founders. And I think that's probably been more of a weaker spot historically. But it looks like that's changing.
Yeah, I think those are great insights. Zooming out to the maybe feature, maybe bug cyclical nature of crypto. We're coming off, I don't want to say max pain, but a two year span of banking events, FTX, Terra Luna, just a lot of craziness. And it feels like we're pre-bull, we're not full bull, but you just sense the amount of optimism in the air in early 2024. What are you most excited about for 2024? Like what do you think are things that are going to be happening that really come to fruition, whether it's the same stuff we've seen historically, rise of new L1S, more depth around the consumer application layer, middleware, types of entrepreneurs. Give us a rundown on what we should be on the lookout for 2024.
Yeah, I think cycles aren't bad. I think cycles and speculation and bubbles are a side effect, I suppose, of innovation. I think in crypto, because everything is highly financialized, you do get quite painful deleveraging events at every cycle. You see these wars I suppose, and it happened with L1s. And then I think we're probably seeing this with L2s and other execution layers. And I think we're probably going to see this with data availability layers kind of splitting out, more like the transition to modular infrastructure.
But I think, you know, since probably 2017, 2018, there has been a lot of infrastructure built. And I think last year there was also more of a focus for infrastructure for UX as well as the infrastructure for scalability finally coming online.
And I think we're at a place today where I think we have the puzzle pieces in place to actually support more large scale applications on the consumer side and by large scale, it's hundreds of thousands to millions of users at reasonable fees. And I think that's something that was not available in the last cycle. And you put that together with better on ramps, account abstraction, better wallet infrastructure, and I think now we have also more consumer awareness. And I'm excited to see now that these pieces are finally live, what we can actually bring to market on the application side.
Yeah, feels like in the COVID era, with everyone glued to their computer screen, you had just more eyeballs, more experimentation, and it felt like we crossed the chasm. In hindsight, you know, generalized L2s were just getting off the ground if we look back to ‘21 and ‘22. And now you have a lot of this infrastructure, high throughput, low cost in the form of L2s, a lot of this infrastructure has gotten built out. You're a highly technical entrepreneur who wants to build crypto infra. What do you build today? Are we actually in the late innings of what the crypto infra stack looks like or early innings?
I think we’re late innings on scalability and early innings on UX in the sense that I think we have enough scalability to usher in another flavor of innovation and that cycle will continue to pendulum. I think what is still missing is better on ramps, off ramps, and abstraction.
I think we're seeing this with embedded wallets and what Privy is doing, I think is great. I think that experience needs to be collapsed. I think for more of the entrants today, I still think it is difficult. And it's interesting because this is not a new topic. It seems like we've been talking about, you know, private key management for years. But I think it's still not there yet.
You know, I think 4337 you know, in the Ethereum world, I think is an interesting evolution. When you look at the usage metrics, I think it's still quite paltry. It's definitely an improvement. But you probably need to ask yourself why is 4337 not being adopted?
You know, when you look at the volume that bundlers and paymasters are doing, why is that still in the low double digits, three digits? So I think that is still not there yet, we need to ask ourselves how can we get that better?
Yeah, I think that's a good way of framing it, like scalability—well, you know, who knows? Like, are we at the point where you can build decentralized Facebook, decentralized Uber? Feels like we're getting close. It's hard to say exactly, but I think framing as infrastructure around UX, UI, making these things more usable, I think that makes a ton of sense.
Yeah, like there's a debate on how much crypto do you abstract? And a lot of the crypto OGs are saying, you know, you need to feel the pain. And I think there's another flavor of people that say, to an end user, crypto should be indistinguishable from a web2 application. And I'm more of the latter. I think you can embed tokens in applications. I think the most difficult part is I suppose the custodying of those tokens, which is why I like projects that are trying to use email, socials as a way to spin up—even if it's a custodial wallet—somewhere to access the tokens. And then you have a funnel by which you have people who come for the tech, stay for the tokens. I think that is for better or for worse, the way you hit more mainstream.
And when you look at a lot of these crypto UIs, I mean you look at the Curve UI, it’s so crazy to think like, how is that ever going to be mainstream? Looks like something that was done in 1994. And I know they did it for a certain aesthetic and for a certain user group, but I do not think the next cohort of millions of people are going to onboard onto something that just feels so unsafe, I suppose, and so non-intuitive to use.
I think the crypto developers that are coming in have more of a focus on on UX. But yeah, that's still an evolution.
I have this internal battle, always. It's like, do we build the crypto native audience or do we just abstract all this stuff? And there's some categories, like NFT and tokens come in many shapes and sizes, but NFTs, you know, like the floor price and just the price of the NFT is front and center. It's going to be tough to abstract those kinds of things.
Yeah. But I think you can go after a crypto native audience with good UX. The Blur UX is really good and they spend a lot of time A/B testing what works and what doesn't work. And it just goes back to, you know, product management fundamentals and design fundamentals that I think the early cohort of crypto builders didn't pay attention to, which is fine, but I think there's more experts in the room nowadays, at least the people from the FAANG companies that are more excited to get into crypto today. And I think they can level up a lot of what the UX is.
So at 1kx, one of the investments that you led was in Pudgy Penguins and it feels like Pudgy has done both building a great crypto native audience but also crossing the chasm with toys and various outlets. Do you think that model of addressing the crypto native crowd but also being able to cross the chasm will be the new strategy for entrepreneurs building these NFT collections or consumer more broadly?
Yeah, it was interesting when we made that decision, along with Peter and Chris and Michael and the team, it was contentious, I suppose, when we also spoke to other investors. I mean, first and foremost, what was exciting to us was Luca as really the force of nature. And I think he was and is very inspiring and he really has the face of a movement, I suppose. And I think that is very important in a more consumer project with a different demographic, I suppose, than your usual defi traders.
I think the approach is different. It depends on whether you're looking at PFPs vs gen art in the NFT sector, you know, because it becomes more complex over time. For gen art. I don't think you need to do the same strategy that Pudgy did. I think for a lot of gen art projects, their goal is to become a store of value. So what they want is to, you know, be in Sotheby's auctions and I don't think they need any real world presence. They just need to be in the mindshare of the collector base and to make sure that is a growing cohort, which I think it is, of more traditional art collectors looking at gen art as a store of value.
With PFPs I think a lot of the benefits of PFPs is around giving people a feeling of belonging and there's really interesting cohorts that are dependent on the PFP and you know, you could talk to the founders, get a sense of who the collectors are and they are very different personas. And the persona I think that Pudgy was targeting, part of why it was exciting is because it was relatively untapped, I suppose. Like the folks that they were targeting were maybe more female skew, maybe younger, and a lot of their channel and distribution strategies were supporting that.
You know, they had a big focus on memes and gifts and being present on TikTok and the physical assets I think work for them because you might want to gift a plushie to your cousin's daughter for Christmas.
And I think it need not be the strategy of every PFP. I think it probably depends for these projects, they need to think about what is their target persona on the demand side, and be very specific. A lot of crypto projects don't do that exercise. But you know, is it male, female, between what ages, what demographics, what do they do in their free time, where do they like to do on the weekends? How can you craft a holistic strategy to actually target that group and give them a way to kind of feel a sense of belonging?
And I think the real world assets are interesting because it gives an additional source of revenue, I suppose.
And just brand recognition.
Yeah, and with PFPs, I think that's the name of the game, you know, how can you get as many eyeballs on this thing as possible?
Yeah, it's hard to make blanket statements on what the perfect type of approach is, right? Whether it's like skewing more crypto native, like trying to cross the chasm. But I think for both, you have this sense of, you mentioned building a movement, you're almost like manufacturing memes to an extent. And for things that are more tangible, maybe jumping into DePIN and DeVIN, you spent a lot of time these last few weeks really surveying the landscape there. And we've talked quite extensively about it. What are the founders that are going to really break out in DePIN and DeVIN? Like how should they be thinking around manufacturing these memes or creating that kind of movement that we talked about?
That's a really interesting question. For the listeners, DePIN—decentralized physical infrastructure networks. DeVIN, it's a flavor, it's a categorization—decentralized virtual infrastructure networks. The difference is that with DeVINs you're thinking about things around storage, compute, bandwidth. These are typically fungible resources and they're not geographical in nature. For the DePINs, this is more around wireless, geospatial, mobility. These tend to be more nonfungible, and they do have a geographic dependance.
So I think a lot of the crypto folks are more familiar with the Filecoins of the world and maybe, more recently, things like Helium on the DePIN wireless side has gotten a lot of eyeballs.
I haven't thought about how to marry those two. It's super interesting when I think about why you know what the thesis is behind, DePINs, DeVINs, let’s just call it DePINs, it is that you can use tokens to bootstrap the supply side. And I think Helium was almost like the Compound moment for DePIN where the success of that supply side was inspiring for a lot of what I think, you know, electrical engineering PhDs, nerds, to say, hey, maybe I can do this for my piece of infrastructure.
When I think about what is the venn diagram between the individuals who are buying these pieces of hardware. You know, I can be like dashcams in the case of Hivemapper, radio stations, in the case of WeatherXM. Why are they doing that? How crypto native are they? Are there weird ways to bootstrap or to further grow the supply side there? Does it make sense to release some kind of, you know, PFP collection for these individuals who are bootstrapping the supply side? On the demand side, I think that's typically been the issue.
That would be a really cool area to experiment. You know, like can you have a way to help penetrate the demand side with a feeling of belonging where, where maybe they are not only using the service because it's cheaper, because I think that's a difficult vector to compete on over the long term, but are they using it because they get a sense of belonging from being in a community with others who use it? Yeah, I think that's been an interesting angle to experiment with.
Yeah, I think back to lower school and like the flex. Well, first off, how I walked into a classroom and I saw one of my classmates with a Gameboy, the little Gameboys, Gameboy Advance, and I was like, I want one of those. And eventually convinced my parents to buy me one.
And I felt like I was part of a collective there, you know? And then with Pokemon and all these games, like it felt like a movement.
Just recently you had the Rabbit hardware device where it feels like there's a similar kind of craze and almost flex for hardware. And so, yeah, maybe there is something at the intersection of creating a movement or creating this sense of belonging associated with these DePIN networks. And I don't know what it is exactly, but that's why we're doing this. We're just riffing on stuff. We haven't talked about this before.
Yeah, yeah, that's really cool. It's almost as if hardware has become cool again. And I forget who said this, but someone said, if you are serious about building good software, then you build your own hardware, and I think we're seeing a trend there. To have hardware as a flex. I think it’s a very interesting phenomenon. Maybe we're actually seeing this with the Solana Saga phone, you know, I just preordered the new one. So maybe there's a part of that where I can see someone at a conference and say, like oh you have the new Saga, so do I. You know, I have a sense of belonging with the other saga phone holders.
You know, that's not, I suppose, traditionally how we might think of DePIN. But I do consider that as within that category.
Not in the DePIN category, but one area that feels ripe for tangible plus a movement is the emerging category of digiphysicals. And I hate that name, but you know, companies like End State where you can buy these sneakers and you have an NFC chip where you know, there's an NFT associated with that physical item.You have sneakers, we're starting to see more around different kinds of apparel. Maybe this is an area where you can combine the physical nature with this sense of belonging, a movement of sorts. I'm curious if you have any thoughts there.
Yeah, totally. I prefer a digiphysical much more to phygital.
That's way worse. Yeah, I do think it is almost like a Trojan horse into crypto adoption. And you know, it's kind of related to the conversation around real world assets. I think part of the issue historically with RWA adoption is that, you know, like these are more financialized. It was things like receivables financing and the persona of individuals who want to invest in that and the institutions who want to put that more onchain, they're just slower, you know, and they're more risk averse.
I also see a chip on a sneaker as a real world asset, but the buyer of that is typically a much younger and much more open to experimentation. And it's interesting that there is more willingness to pay for these physical assets, particularly in the fashion space, where if they have, you know, one of these chips, there's a greater willingness to pay. And if you take that to a logical conclusion, then you'll have hundreds of thousands, millions of these different articles of clothing that have some connectivity to being onchain. That's actually more real RWA adoption than what we initially thought as what are RWAs.
I'm also excited about what comes from the world of digiphysicals because you got to think Foursquare 2.0 is like the next logical move. I mean, sure, you can use your phone to check into places, but what if you didn't have to do anything, and you had these quests?
And I think of Pokemon Go, like the craze that was around finding these Pokemon. I was living in Boston at the time and people were like in these parks running over children to try to capture Pokemon. In a world where these NFC chips, you know, are communicating geo and where people are and you can have these like almost mini games and real world experiences powered by digiphysicals. Any thoughts on that or is this a world that's like ten years away, not a few years away?
I think, you know, objects in the mirror are closer than they might appear. When you combine that also with token incentives—
Yeah we saw that with Stepn to an extent.
Yeah, it's a very powerful mechanism for doing large scale human coordination. And so if you put those things together, I think there's some really interesting entropy that can happen.
Yeah, who knows? But could be cool. Hard pivot here. And I'd say the most hotly contested topic on crypto Twitter, let's call it over the last few years, the modular versus monolithic infra debate. How should people think around how this stuff plays out? And that's a broad question, but I'm framing it that way purposefully for you to take a stab at it.
Yeah, I mean, I think in that context the usual conversation is when you split out, you know, Ethereum was actually monolithic itself. And then you split that out and you have your execution layer, your settlement layer, and your data availability layer. And the form factor for a lot of the separate execution layers are your roll ups and all of the different flavors of your validiums and things like that.
And then you have on the other side the argument, you know, Solana is your fully integrated kind of monolithic chain and a lot of folks say, you know, the benefit there is better UX, cheaper. I've heard, you know, memetically people refer to monolithic as being more of the Apple and modular as being more of the Android. You know, so I'm more on the modular side of things as almost an inevitability in open source.
I think developers often gravitate to customization and specialization for their own needs. And if you do believe in ideas around application specific rollups and giving applications the choice to say what components do I want onchain, offchain, do I want to share block space? I think it kind of pushes you to having more diversity.
And I think when you look at Solana, it's also becoming modular. It almost doesn't have a choice. When you look at something like Eclipse that uses the Solana SVM but its flavor is an Ethereum rollup, so it uses Ethereum for settlement. I think that's one flavor of being modular.
You know, I've heard other anecdotes, people saying I think Solana is the best execution layer and Ethereum is the best settlement layer. I'm not saying I agree with it, but it does point to saying, assuming a fully open source environment, developers will kind of gravitate often to the thing that does the job better. And I think there are also a lot attempts at improving the EVM and there might be some success with different flavors there.
And I think whether we like it or not, I think the entropy around developer adoption kind of pushes us to, ultimately, choice. And I think it's hard to avoid a market structure in an open source environment where we just have three or four or five different flavors, you know, and if there are enough developers across enough use cases, I think that natural fragmentation just happens.
Yeah, it's interesting where in 2018 with the ETH2.0 roadmap that Vitalik put out, rollups weren’t really part of the discussion at the time. It was around sharding, and Ethereum was still a monolithic blockchain, this is pre proof of stake, 2018. And there was a mindset shift by Vitalik and the EF, and I don’t know what it was exactly. I've heard some of it was influenced by Polkadot, maybe by Cosmos, but evolving to more of this modular type approach. I do wonder, is that an evolution that the leading monolithic chain, Solana, is going to go through in itself?
It's a good question. I truly don't know the answer to that. I haven't spoken to Vitalik about it. I think part of the issue is, to my understanding, sharding is difficult. And I think you can get a lot of the benefits of that architecture by just focusing on execution. And there were other ideas around plasma and a lot of the unknown unknowns—
Yeah, we'll see. I think Solana is really trying to kind of stay as integrated as possible. And I think their focus is on how do you send value at a 100th of a penny and what are the trade offs that you need to get there? And maybe there iare trade offs on the decentralization spectrum and maybe socially that's fine. But I think to really see what breaks at scale, I think to see whether an integrated approach will work or not, you probably need to add a couple of orders of magnitude of users and transactions. And I think that's the real difference here between vertical and horizontal scaling, because that's the difference. I think the Solana approach is largely to increase the node requirements and the bandwidth requirements. And the Ethereum approach is more horizontal where you just add more machines to the network that can do the execution itself.
Yeah, the fun part about investing and being a part of this industry is like, there's no right answer. And you have very different trains of thoughts focused on their particular ecosystem. I think that's one thing I absolutely love about this space is just the diversity in thinking. Maybe just to jump into a little bit of a different topic. A technology that's been around for some time but feels like it's getting to production grade and enabling different types of applications, things like that—ZK, zero knowledge proofs. We were investors in Lagrange. You can speak more to what Ismael and the team are doing, but yeah, maybe ZKPs and what this infrastructure, assuming we're getting closer to production grade or maybe there already, what kind of stuff does it unlock and where does this fit within the monolithic modular type frame?
I think the technology's magical. I also think it's largely, I won't say misunderstood, but people focus on a different component there, where a lot of people focus on the ZK side, the zero knowledge side, the ability to say—to prove that you have some information without revealing what that information is—that has some privacy properties, though it doesn't give you actually very strong privacy guarantees.
But I think the more interesting part is actually the snark. It's the compression part of ZK. And I like to elevate the topic and look at it under the umbrella of verifiable compute. And this topic is not new. It's been around since probably the early cloud computing era where there were companies, when they wanted some proof that data that they were receiving from a server was not tampered with. And the research in zero knowledge proofs as it relates to verifiable compute has been going on for over a decade. And I think the difference is today there are much better algorithms and much better hardware to actually how the economics work at scale.
What I like to frame ZK as, ZKPs—this might get too deep bug I’ll attempt it—when you think about, you know, what makes up a computer, you have transistors and logic gates and you have these and gates and or gates, and they effectively either let electrons pass or don't pass. And then you get these zeros and ones at the end. Super basic explanation.
It turns out that you can represent these physical logic gates as mathematical equations, and that's a super mind blowing concept that I think it took me two or three years to really understand. So what you can do is express a program as a set of mathematical equations, a lot of them, and effectively have a proof that says, I can prove that this output was run on this program by effectively providing a mathematical proof.
So you are effectively turning code into math. And I think I like that framing more because it talks about more of the verifiability and computational integrity rather than the privacy aspect. And I think it's really cool because when you think about what blockchains are, they're also verifiable computers, but they do verifiability in a different way. They do it with re-execution, with consensus. Everyone needs to execute the same stuff. And then you say what the majority, either by some mechanism, you know by stake weight, do we agree that we all achieve the same output? And I think you might not even need to do all of that if you're just able to do the proof, and then have others verify the proof.
So then only one chonky machine needs to do the actual calculation and then everyone else will verify it. And I think it's quite a different design than what we've historically seen with blockchains. But I think it's a super powerful technology and frankly I think it has implications that potentially span beyond blockchains themselves. But the initial use case for ZKPs and verifiable compute has been more on the privacy side with Zcash.
And what then happened was more around computational compression, how I frame it, which is really L2s, ZK rollups. And now I think we're trying to figure out actually like after compression, what then?
And I think verifiability is another interesting component and maybe that's where like Lagrange plays and a couple of other projects in the space Modulus Labs on like the ZKML side, Axiom around like blockchain introspection. It's like how can I reduce the compute that I do onchain by doing some compute offchain and then providing a proof that says that that data or that computation was done correctly.
So I frame it as you're effectively giving smart contracts superpowers. And I think there are some really cool things that you can enable. Potentially in defi, you can have more complex collateral management. In gaming, you might be able to do more complex economic rebalancing. Yeah, so I think there's still like a lot of the projects are going through the idea maze of figuring out where they can use ZK.
But I think verifiability is probably the next step that's particularly exciting to me for ZKPs after the partial privacy and compression use cases.
We've also seen, and you sort of hinted at it, the intersection of ZKPs in the lens of AI, right, for model integrity and things like that. We're large investors in Ritual, and it doesn't have to be in the ZKP framing, but the intersection of crypto AI. If we were a traditional venture fund, all we would be talking about would be AI. We're lucky that we get to look at a bunch of different sectors and categories and things like that. But crypto AI, what can we expect ‘24 or ‘25 over these next few years?
I think there's always a venn diagram, you know? And when you look at the breakout companies like Uber, that has a lot of components under the hood. There is the mobile phone, there's internet, there’s the GPS. These are all technologies that are weaved together to actually deliver a new experience.
And I think there's going to be quite a large venn diagram between AI and crypto. I think the technologies actually serve as checks and balances to each other. I actually think AI is philosophically more of a controlling technology and blockchains are more of a liberating technology. So I think from a societal perspective, I think they'll serve both important roles.
You know, my prediction is, for example, in five years, 5% of the content online is going to be human generated, like we just won't know what's real. And in a place where you do know what is real anymore—and it's already there if you look at a lot of the gen AI content that's coming out—there is going to be a societal premium on humanness. And I think that's a bit different than with privacy because, unfortunately, I don't think there is great societal importance placed on privacy or else no one would be using TikTok.
But I think with verifiability, that and humanness that is such a fundamental thing and blockchains and ZK are very well suited to this. So I think it almost rides the trend of generative AI.
You also hear like just tops down the calls for “we can't rely on centralized AI” and the paranoia around a select few number of companies basically dictating how AI's ushered in and deployed in the US, but globally. And it feels like the crypto meets AI is prime from a tops down standpoint. I think the open question is from a bottoms up standpoint, is it things like Unibot 2.0? That seems like the low hanging fruit from just a bottoms up. What's going to come of this? I'd love to have a personal LLM, but yeah, like curious on how this comes to fruition from like a bottoms up standpoint. Is it going to be a big infra type play like Ritual that really ushers in these net new applications? Is it going to be OpenAI shuts down their API and devs building on it need an alternative? Yeah. Walk me through how you think it's going to play out.
I like the weird use cases. I think it always starts out weird and fringe. So for example, I think people are becoming more and more lonely. You know, I think the pandemic was a big factor there, but I think as more of everyone's lives are hyper socialized, I think human loneliness is going to continue to increase.
I think AI girlfriends are a very real future that will happen—
Yeah. Not for us, but yeah, but yeah—
Yeah, I would never pay for one. But if I did, maybe I want proof that this is my AI girlfriend, and that's my model because I'm paying $30 a month for her. So maybe I want some verifiability that this ML model is actually the one that I have trained, maybe on the texts of my ex. And I think there's probably a willingness to pay for that.
I think another super weird one is religions. You know, what if someone train an ML on the Bible, and then you have your AI Jesus. And then you start to have a cult following, do they want any verifiability guarantees that this deity that they're following is actually the one that they trained?
And it happens, I think in these fringe examples and I think it becomes more mainstream quicker than we thought. Because I think at some point people thought it was weird to get into the cars of strangers and to rent your house out to a stranger. But Uber and Airbnb are things. So social behaviors sometimes change quicker than you expect.
We've talked a lot about different categories and ideas and some really wild ideas. As an early stage crypto VC, I think it's great to be thesis driven and almost have a sense of where the future may be going. But I think you also have to factor in that the most compelling entrepreneurs are going to be the ones that create new markets or figure it out. And so how do you think around just balancing looking for compelling teams in these particular categories versus just tracking very talented folks in the crypto ecosystem?
Totally. I think I have evolved my thinking. You know, I've been investing for six year in crypto, and I think I've evolved my thinking to be more founder driven and then thesis driven over time. I think no investor really has a crystal ball, and a lot of these trends happen out of the blue. And I think if you lean in too much on thesis, you become quite rigid. And I think there's also an ego involved for a lot of investors because I think they want to feel like they're right. And so if a founder says something that is not aligned with their thesis, I think they get a little hurt there. I am totally fine with and with the founder coming to me and saying, Here's why you're wrong and here's why I think the time is now. And I want to enable those founders and give them the resources and support to have their world view come to fruition.
And I think thesis building is a good exercise and I think it provides good guardrails and it's one that we do as well. But I think I've seen a lot of funds make the mistake of becoming so rigid that they will not look at an opportunity because it doesn't fit in that category. And I think it makes sense, you know, for an investor like why they would do it. You see, you know, hundreds of opportunities a month and it takes a lot of brainpower and everyone tries to look for shortcuts, it's very evolutionary, you just want to burn less calories. And so if you like, God forbid, need to rethink a thesis that they just spent three months developing, maybe you just say no, it doesn't align. But maybe something was different there. And I think oftentimes it's actually the founder that's different.
Yeah, I think that's spot on. Something we talk a lot about here and how you define founders, like these archetypes—our name Archetype, it means something broader—I think something that we’ll continue to build data around is are there certain features that we can look out for that sort of foreshadow an entrepreneur's ability to build something remarkable, sort of cross the chasm or whatever it may be? I think it's something that we're going to continue to revisit over the coming months and years. But it's helpful framing to not lock yourself too much into particular categories while also balancing entrepreneurs and timing and things like that.
Anything else that we haven't discussed that you want to hit on, Dmitriy? We've hit on a lot today.
I think we hit on a lot. The last thing you know, it's been good to be in New York and obviously community is a big focus for us here. So if any listeners are based in New York or are visiting here, hit us up. We'd love to grab a coffee, host you here in our offices here in SoHo.
If you're lonely, you know where to reach us. But yeah, we're all in person, we're very intentional around building community. Get people in front of each other. There's so much to learn from each other as entrepreneurs and as investors. And I find that I learn every single day, and I'm just able to do that faster in person. So yeah, I feel lucky to have someone like you, Dmitriy, joining the team and, you know, it’s going to be a blast of a few years ahead of us.
Yeah, this is going to be fun.
All right. With that, signing off, thank you to the listeners for listening in and we'll see you on our next podcast.
Cool, thanks for having me.
DISCLAIMER: The information in this video is the opinion of the speaker(s) only and is for informational purposes only. You should not construe it as investment advice, tax advice, or legal advice, and it does not represent any entity's opinion but those of the speaker(s). For investment or legal advice, please seek a duly licensed professional.