Fireside Chat with Kartik Talwar & Evgeny Yurtaev at Pragma Prague 2025

[Music] We're going to do one next fireside chat before we take a quick break. So without further ado, I'd like to bring on Ekini to tell you all things Zerion. Please give him a big round of applause. Thank you. Pleasure to be here.

Thanks for being here. Um, we've known each other for a while and um, it's always uh, awesome to kind of talk to you. Uh, usually we don't do this on stage. So, uh, it's going to be one of those conversations where we're trying we're going to try to do the same thing that we usually talk about, but, uh, with other people watching. Um the part that is interesting to me is um anytime I ask somebody if you know about Zerion um I get a different answer for the product that they use from Zerion.

So my first question to you is when somebody asks you what Zerion does, what do you usually answer? That that's a great one. Uh and that's true because like Xerion was changing so so many times. we've evolved uh so we started in like the earliest days of Ethereum like since 2016 and evolved and went through so many iterations. So I guess the answer really depends on um who I'm talking to.

So to the users Zerion is uh the most intuitive self-custodial wallet. Um I think the philosophy when we were approaching building Zerion was it should scale. The more power user you become the experience should scale and I think Zerion is following that trajectory. So, it's as easy to start as any other wallet and probably even better. But then if you are a DJ trying to like flip meme coins or do all crazy DeFi things across every chain, uh Zerion also works for you.

Um and then for developers, Zerion is primarily the Zerion API these days and that's a natural I guess evolution. So Zerian API happened to be the one of the most successful products that we've built recently. Uh and uh this API allows you to replicate and build your own wallet without building full wallet infrastructure as we've done. So we're just saving time for developers to go to market faster. That's awesome.

So I want to talk a lot more about the API and then there's also a a few things I'm curious about on how you think about the feature of wallets al together. But um was the goal of the API to just build the API so you can use it for the Xerion wallet or was it to satisfy an actually external demand for that data to be used by third parties? Uh yes. So Zerian API was initially not even like productized. So we had our internal API.

We've worked for years to make the we always focused with the Zerian wallet when we launched on the user experience. So we wanted to have an API that works across uh like all of the clients. So we don't need to replicate a lot of the work and don't go to the lowest primitives like RPCs. Uh and uh it happened that over time more and more people started reaching out to us and saying look like we want this infrastructure. Uh and this is how Zerian API happened in the first place.

We've existed with just one client for maybe two years. It was not like we're not charging for this. Uh so Rainbow Wallet was actually our first client who reached out and uh and this means what was Rainbow Wallet using the API for and and what is the yeah what are the functionalities here? Yes. So so pretty much everything data related.

So when you create a wallet obviously the keys inside your app but then all the information about tokens on your wallet your transactions that happen across all chains this was fetching fetched from from our API. Um and then later more and more customers started just reaching out and we realized that we need to make this a real product and there's like a lot of demand um on the market and it honestly with like 2025 starting it it just uh continued exploding I would say. That's awesome. Um what what do you think people get wrong when um either designing this or using this um kind of data API? Because you would think, oh, I can just uh save all the RPC data on in a database and just give it to you as an API.

Is it just is that basically it? Like what goes into this and and why was it uh why had nobody else done it before? Yeah, that's a that's a good question. I think um first we didn't have so many chains. Uh and I think the complexity of the ecosystem was a little bit uh less than right now.

And that's why people were thinking, okay, I just need this thing and that thing. I'll stitch them together and it's going to start working. Uh but as we now, you know, in 2025, we have uh hundreds of chains, we have uh thousands of D5 protocols. If you're trying to build a product that really works, you need to invest a lot in infrastructure. So uh in terms of RPCs, even managing RPCs for all the different uh blockchains is is a mess.

Then aggregating this data, it's just a very pragmatic thing to do. And I think companies started to realize that um it's just not their business. Easier to outsource than to do it. So that's a straightforward thought and I think it took some time for people to to get to that point. But not even right now like not every company is like a lot of companies have replicated a very similar infrastructure.

Um and I think what we're seeing right now the there's a big emergence of uh B2B businesses uh and APIs because people don't think that like why do we need to maintain 15 different businesses inside uh and I think it's a great sign of uh maturity of the industry. Absolutely. And and all in a way just proves that uh this is something that people are willing to actually pay for um in addition to uh just it being necessary. Um, one of the comments that we uh heard from uh one of the earlier speakers uh this morning was uh Bogdan from Tenderly saying uh running RPCs is the easiest thing ever. Try building EVM debugging tools.

So I guess my question to you is uh is that true? And uh and uh what are the complications you run into uh maintaining I guess what 50 plus chains that uh you do for the API? The funny part is that we don't run RPCs because this is not our business. Yeah. In terms of like managing Yes.

Yeah. So for us like we we never wanted to cuz we we knew that like it's a different piece of stack that uh you know there there other companies you know quick alchemy they they do it really well. Uh so it doesn't make sense uh for us to to do everything inhouse. So for us the biggest part is really the infra that we've built on top. Uh and that's that's hard.

Um we're spending uh like over 100k a month just on AWS for example. U sometimes even more. uh and this is uh on like a fairly uh you know crypto is I would say still very niche and uh the number of users on a daily basis is fairly low especially now uh and uh with that kind of usage we are already at like a huge infrastructure spend so um yeah it's expensive to maintain um so I guess is that in a way saying that it's going to be a world where people are going to use RPCs less and just rely on APIs who are ingesting the information giving it out there or do you kind of feel like there's a more interesting kind of outcomes here like how do you think about future of RPCs? I think RPCs are going to be specialized. So um RPC becomes expensive if you're just trying to call like query everything from a single provider and I think uh what is happening uh right now is that there are even RPCs themselves are specializing.

So uh you would have RPCs that are better for sending transactions and some RPCs are u better for like fetching historical data. some RPCs are just faster to respond etc etc and I think uh use case specific uh APIs are starting to emerge so you'd see like I mean Zion is one example obviously for uh user data I mean the graph in a way uh the graph API was uh one of the first I guess specialized um approaches to the the RPC and there are many more examples and I think as we as we grow this more specialized ones are just better optimized for the use case and uh RPCs are going to be you know very low-level thing that no will really need to know about uh when you're building a I guess modern composable uh application consumer application. What's the value of those specialized endpoints being RPC still if they can just not necessarily Okay. Yeah. I think the biggest value of RPC is the composability.

So they're standardized. Um but they don't have to be RPCs in a way. Um yeah. No. Okay.

I was making sure I didn't misunderstand something. Um, you've obviously been building a wallet that's very successful for for many number of years. Um, what do you think about kind of the surge of a lot of wallet applications or wallets being embedded into almost everything that we're seeing? Um, is that a is that a good thing? Do you feel like uh we're going to learn the limits of how much uh different things we have to maintain as a user, complexities, all these things?

Like what does kind of the wallet world look like to you? Yeah, that's a great question. Um, I I think there's a a bit of a fork that happened um a few years ago, I would say, when we there are two types of wallets that I think are going to exist. Uh, so I think application specific wallets are definitely going to be there. Uh, and this is the biggest trend I guess obviously so embedded wallets, but it doesn't necessarily need to be uh like social vidget like wallet.

it can be embedded on a much deeper level can be even custodied inside the application that is building it and we see that that happening in fintech as well. uh so like fintech apps just integrating straight into their applications the the crypto like keys and uh non-custodial stuff too and then separately uh there is a a trend of like new neo banks emerging from the wallet um from the non-custodial wallets uh and I think it's it's really about the brand and positioning of in a way so there are only a few wallets in the ecosystem that are uh large enough to really become the new neo banks of like the the future future world obviously the the biggest ones and we see the a big race from centralized exchanges like OKX Binance they're all launching new wallets investing their existing wallets because this for them this is just distribution and uh preventing people from leaving to to new wallets um so yeah I think it's going to be a duality of that and there's probably nothing in between and it's very hard to build for for both um so so you think it's still hard to see kind of the future we end up in where are we just kind of expecting wallets to exist in multiple places for a given user or we kind of expect like one one or two super apps that just uh hey this is my wallet and this is the wallet app I go to just like the way you have only one bank and maybe sometimes two but 90% of people will not have multiple bank accounts um is that more likely or or it's going to become so easy that it it you may have a 100 different wallet accounts somewhere I I think it's going to be closer to you having your own application which you trust where you store the money uh kind of long-term you use that for payments like similar kind of use case to a neo bank but built on crypto rails and separately if you are using you know uh I think the easiest example would be something like hyperlquid or like anywhere you're doing you whatever application specific wallet yeah farer for instance yeah there you would have a separate wallet kind of like with uh say YouTube creator rewards, you accumulate it there, you withdraw as needed, but you're not going to put your life savings into YouTube uh wallet. So, I think that's the difference between different types of bullets and I think that's going to be pretty much mimicked from what we see right now. And I there's nothing wrong with that. How has that changed any of your thinking around evolving Zerion wallet then?

Another great question. Our thinking about Zerion is always evolving as Yeah, fair enough. But I should say that we've started putting much more effort into um simplifying the experience. So we looked at u I guess if we want to compete with neo banks getting into crypto, we have to be at least the same in terms of user experience and that has not been the case. And I think we're currently losing this war uh to the broader ecosystem of banks and like um existing web two players because they are using the what we've built as a community just putting their uh stuff on top their distribution just on boarding a bunch of users um and kind of leaving behind a lot of great companies that are in the the Ethereum ecosystem and beyond.

Uh so I think we need to fight back. So, um I think we as an industry we need to step up and uh just start finally start using the stuff that we built uh and put so much like you know sweat and tears into uh and and use it. We don't need more. It's that's enough. I think that's something all of us can get behind.

Um, what are the Let me see if I can rephrase this because I was going to put this more negatively than I think I really mean to, but I assume just like everybody else in the last 12 months, you've had to support so many different networks. Um, how does that complicate things or what benefits kind of have you noticed or do you have any advice for people starting new chains when they eventually have to integrate with a wallet that uh don't do this because it's going to anywhere from take extra long to will make our lives cycle crazy. How does it look like from an infrastructure company when you have hundreds of things going on? Um, it's a good one again. Um so as as some of you know we've launched our own layer 2 um earlier in like late last year and we've took a lot of lessons from it.

So it was a very exciting and fun uh experiment for us. Um that is still ongoing and uh um yeah if you want to watch a fun launch video you should definitely check out our YouTube uh we we've had so much fun like filming it in in RIA. was a was a great launch video but um zero network it just to recap what it does uh we basically tax the toxic order flow so basically all the chains that are existing right now like layer 2s they don't differentiate between different types of users so the fee is pretty much the same the gas fee is the same across all the users versus in zero network because we know some past history about the wallet we can actually subsidize these kind of transactions coming from more real users and uh have a higher fee on on the bots and and transactions that are basically not originated from uh I guess non-toxic um flow uh and uh from that perspective like it's been pretty sustainable uh zero network but I think what we definitely underestimate is how much work it takes to um build an ecosystem and especially if you're trying to be in the generalized layer 2 world and I think all the chains that we're currently seeing they're still trying to be a very generic generalized um EVM approach and I think first advice is definitely specialize. I would say that u lower down like the smaller the smaller number of use cases that you support ideally one the better. Uh so if your chain is trying to to on board pretty much every piece of infrastructure you you're going to suffer from paying every vendor and then you you'd also need to incentivize the someone else to come in and build and this is also not sustainable.

So I think uh go to market is uh is key and obviously distribution is the the hardest bit. So if you have some uh I guess uh advantage in that that's definitely uh worth utilizing. Yeah. Uh but maybe right now don't launch in L2. Wait a bit.

Yeah. uh says the the team that also just recently I guess you had to kind of go through that cycle and then realize uh yeah I mean we yeah sometimes like underestimate or overestimate things but it's not the end like we have the plans for for the zero network we also like I'm just following my own advice to be honest with the zero network strategy we're going to dumb it down that's amazing um maybe I guess as a as a closing thought Um what do you think is missing from uh the world of uh crypto infrastructure? Anywhere from things that are either expensive that you like to be maybe bring down the cost for to things that you want to use but you don't want to do that work yourself and you would rather pay somebody for it. What are kind of some of the interesting uh feedback or observations you've had? It's a great one.

Uh I think right now uh just overall the infrastructure is is very nent still across the board if we're talking about uh and it but it's it's getting built but it it's pretty amazing to watch. Uh but it's only been recently that it started to emerge. So for example if you if you're just trying to bootstrap your own consumer app the complexity of that is is enormous. So I think like it it comes from where do you store the keys? How do you do uh security?

How do you do compliance? How do you do data? Like all of these questions are currently answered by a million different providers with various and very complex uh sales cycles. And it just uh it becomes for for someone just trying to hack things together. It's it's pretty complex.

So I think that's something we would want to um you know see solved uh and their progress in that regard. But I guess in terms of what I'm I'm excited about is the AI part of things. It's not directly related to our current cost, but we've been uh plugging in Zerion API into many of the AI tooling out there and it's been we're seeing some early success in that. Um and how are you trying to use it? So there are MCPs obviously.

Uh so there like people were helping us out to build plugins for different kinds of frameworks such that Zer API can be used as part of the agent flow. Others are using Yes. Okay. I see. Yeah.

So Zerian API is wrapped into a plug-in. So you can just come in and like if you're trying to fetch a balance of some kind of wallet, it talks directly to the Zerian API MCP fetches that. But I think the cool part would be once we have even um the transacting endpoints. So we're leaning more into that as well. So you would be able to use Zion API to transact and it's because it's so uh standardized.

It's the same like flow for agents. And once that gets out there, I'm I'm very excited about the that's actually very cool. Yeah, because then you have a very way standard way of just passing in commands and Exactly. So birectional. Yeah.

Fetch what they're about like they can interact and transact. I'm honestly very bullish on the AI uh agent on top of uh basically the thesis that Joy wrote in 2023. I recently checked he wrote that in Joel Monro from voice orderer. Uh so yeah, he wrote this uh thesis about AI agents using permissionless networks um in like October 2023. So it was I think uh way ahead ahead of many getting to the same conclusion, but I'm still very bullish on that one.

Amazing. I think that's a lot of things to look forward to. So I want to thank you for uh taking the time and uh can't wait to play with some more AI agents and Zerion data. Thank you so much. Thank you.

[Music]