Fireside Chat with Kain Warwick (@kaiynne) and Sarang Parikh (@sarangparikh22)

Transcript

Hello. Hello. GM. Awesome. So, yeah, let's get started. I mean, most people here would already know you, but if you want to go ahead and, you know, introduce what new thing you're doing with, uh, Infinex and, you know, Sure. Sure. Um, yeah. So uh you may know me from synthetics days um the DeFi summer days uh but I've been building for the last couple of years a new project called Infinex um which is designed to improve uh some of the UX issues that we have in building D5 the things that I encountered in the synthetics days um and and yeah so you know we've got a lot of uh a lot of solutions that we're working on to improve UX in deep. Oh that's that's awesome. So let's actually

start diving with the elephant in the room which is the EOF. there was a lot of Twitter uh drama around it and like a ACD you know decision that took place that it's finally out from the update but what are your thoughts around like the pros and cons of it and like how does the security change and like what what could the community improve because this directly affects the UX of Ethereum right so uh yeah I I mean DevX is important of course right um you know especially at this conference I don't want to get booed off the stage um so so I I think it's not incorrect to say that we should be doing things to improve developer experience that we um have uh aspects of uh programming and solidity that are sub-optimal for sure like

everyone understands that um the issue I think and and how this whole thing is kind of symptomatic of uh the dichotomy between builders um who are building at the application layer and the researchers who are working at the consensus layer is that oftent times the unintended consequences uh of a change like this are um just really ignored, right? Um even to the point of you know like in the last few days people were like how the did we not test this?

Like this is insane, right? Um and it's like oh our job is not to test things like that, you know, our job is to optimize for this, you know, particular outcome that is uh idiosyncratic to our like research mandate, right? um you know uh your gas estimation nonsense is not our problem, right? Which is which is pretty wild when you think about it. It's like guys, we've spent like four years getting better at uh at you know estimating gas and now it's just like ah start again that it's just it's kind of crazy. So I'm I'm glad that and this is the one thing that I will say is most of the time um and we saw this you know back uh many many years ago um with uh with one of the changes um that the

Kyber team called out that uh created you know a huge issue with uh with gas like you know massively impacted gas on uh um one of the op codes and again no one tested it because it was like oh yeah like running smart contracts like that's not our problem, right? Like we're working at the consensus layer. Um but in the end uh you know the the rough consensus mechanism that we have for making changes to Ethereum does work. It might be slow, it might be inefficient. In this case it was extremely inefficient. Um but we do get there and so enough social pressure is able to kind of overturn this and so we got the right result I would say. Um but you know uh it's a bit of a known goal.

Yeah. I mean like for me it should we should take like you know like some pages from how the semiconductor indust industry works because they are like the machine code and the chip manufacturers and chip manufacturers never say we we don't care about how the machine codes powering rocket is going to get impacted. It would be wild to to think that right like that's not our problem like it's every like everything above you in the stack is your problem on some level right so yeah that that makes sense and uh what do you think about you know the space in account abstraction we have like few EIPs like with 4337 uh and like 7702 getting activated like for me like I was always more leaning towards 7702 um and we have seen like from the past

okay there has been some adoption for 4337 but I personally feel it was like more forced than you know actually organic what are your thoughts around it and you know what's the way for forward I think at a at a fundamental level um they're they're both directionally correct right like both both approaches are directionally correct um with different trade-offs uh I think it's interesting if you talk to the infra providers even the ones that have implemented um 4337 they haven't implemented it to the standard they've kind of gone like they've like done some stuff and and then gone around around uh the back on on other things and and haven't actually adopted it um you know as it's specified. um which again kind of speaks to this uh mismatch between

the intent of the EIP and you know the the usage of it right if you're if you don't talk to users and find out how they will implement something then it's very easy to implement something in a sub-optimal way um and and this is one of the other challenges I think with like generalized solutions you know many many of the solutions in the Ethereum ecosystem are extremely theoretical right um they don't actually contemplate the usage of the platform and and how you know uh a smart contract engineer will be building a set of smart contracts. Um that said, we've chosen 772. Um you know, we're where as in Infinex has chosen 7702. Um that's that's the direction we're headed. Um I

think it is a very good improvement to to UX and it's one of the components that has allowed us to kind of uh build this very abstracted um uh omni chain view of the world. Um and so yeah I I think you know again it's not a huge issue uh if 4337 doesn't get adopted. It it was directionally correct and it's kind of like okay you know that one didn't work but 7702 does. Yeah, I mean 437 is like a glorified smart contract with like a network, right? And the issue was like they're like 95% of the apps I've ever used doesn't have 437 wallets. The ones which have have like work probably alchemy as one provider like some other providers. You put money

in them and then you have to resign up with another provider and kind of put money there and then you don't see them on DBank and problems like this, right? Yeah. There's a there's a fragmentation issue with it. Um and you know there's also like some jank around like an alternative mempool and stuff which is just not ideal and and you know again like directionally correct. I don't want to criticize uh people for writing EIPs. I think that uh you know anything that is focused on UX is is bullish right for Ethereum. Um but again it it can't just be a theoretical solution has to be practical as well. Yeah. And what do you think about the because you you guys heavily work on 7702 right? Do you think there could be any security implications that we probably haven't explored yet? U and but we might have to be careful in

putting all of this in production because it's it's like a big change at the EOA level for every account that exists, right? So yeah, I I think there are already chains that have implemented it. Um so you know we have seen uh what it what it does in the wild to some extent obviously not on mainet and therefore you know there's there's a number of things that um that I think uh are potential implications um but you know we use zero dev um and they're already very much leaning in that direction. So you know I I think we can reason about it fairly well um and uh and I'm I'm not super concerned um but obviously you know we'll see. Oh yeah, that that makes sense. And the other that leads to for the question of you

know like we're leaning more towards the standardization here. Like if you look at some of the web two industries with even like half of the market cap of what we have in crypto, they have very robust standards, right? Uh for me personally, I don't see a lot of those standards being followed onto like wallets or infrastructure that deals with users money and you know custodial in general. um what do you think we can do to change it and can we make like universal standards and you know like push EIP such as account prefixes right like you know like eat and then the address a and then the address you know 15th standard problem huh um so so look I I think um I've been thinking about this a lot because we've got an announcement this

week um about uh how how we've approached this problem um and you know I I've talked to Pedro and you know I've talked to a bunch of people the Rainbow Wallet guys and and you know various people that built Infra for wallets and and have adhered to the standards and the you know there is definitely a push for standardization. Um the problem is that I think you end up in this state of uh stalemate between application developers and the infrastructure providers. Um and and what I mean by that is as an application uh builder, you don't really have control of the infra, right? You don't have control of the interfaces. The interfaces are are fairly uh well kind of structured and

and oified at this point, right? You know, it's basically metamask. The meta mask API is is kind of, you know, what dictates how we pass the signatures through and and interact with contracts um at the DAP layer. Um and everyone else is kind of built around that same that same architecture. Um, and so if you are an application developer that wants to do something differently, and I'll give you a very specific example. Um, today as an as a DAP, you can only have one chain in context at a time. Yep. Right. So you go to Ave and there's like this big drop down which is like which market are you choosing, right?

Which is a bit insane if you think about it. The only reason that is the case today is part dependency, right? We had one chain and so we built everything with this expectation of one chain and then we had a bunch of L2s and you know Salana and a bunch of other things that have now created this weird thing where like we're locked into one path and and you know it's very unsatisfying and we all know it um but we can't really think our way out of it. So as an application developer, you are building an application and you've got a wallet that can only have one chain in context at a time, right? And so you have to build your application with that in mind. You can't change that. You have no control over the infrastructure. Um then as an infrastructure provider, you need to be

backwards compatible with all the DAPs which are also built to have one chain in context. And so the only way to break out of this stalemate and and to you know uh kind of this inadequate equilibrium is to have someone who's willing to go and build the infra and the application at the same time and historically no one was willing to do that. And then Infinex was like actually YOLO we will um because we're trying to compete with centralized exchanges and centralized exchanges are full stack applications. They control the infra they control everything. there's a little bit of a trade-off you have to make, which is, you know, giving up custody. But if you're okay with that, you get a full stack application that is multi-chain. Um, you don't need to care about gas. You don't need to care about

any of the things that we care about when we go into decentralized um application land. And so we said, actually, no, we're going to optimize for UX and we're going to optimize for um, you know, abstracting away all of these things. And so we're going to build this full stack application, which means rebuilding the signing infrastructure. And so we don't have a browser wallet. We don't have, you know, any of these uh things like a WagMe connector. We don't use wallet connect. We don't use any of that infrastructure. We use some modern stuff that's available like zero dev uh turnkey, but we basically re rebuilt the entire stack. And the only way to break out of that stalemate is to be willing to do that. And most people are not crazy enough to do that. Now the question is

then, okay, you've done that. Congratulations. you built a little silo and there's like 20 people using it. What does this mean for me as an application developer? What it means is we need to make that available to you. Yeah, we need to get we need to take the thing that we've rebuilt and we need to say, okay, now you can use it as well. Um, which is maybe what's coming. Yeah, I was sold that we don't have browser wallets. You know that still annoys me like why why when I was at Sushi we expanded like 40 chains and we had to add like a search element on the drop down so that people can easily search the chains right it's crazy I mean you look at MetaMask and you go okay you know there's no chain concept at first right it's so slow after a point because

you have so many chains dev nets test nets exactly like there was a period where you'd add test nets and it's like okay you add test nets and you have your like dev environment that has test nets but you would never have that in your like production environment that you're using for your normal transactions and now we the drop down is like 8,000 pages long. It's like just insane, right? So, um so you know we need to solve that. Um we need to we need to rethink how the infrastructure and application interact. Um and unfortunately uh the standards path for doing that is just too slow. Um and so we just said yolo, we're going to do it ourselves.

Yeah. I I mean like if if probably what I would say as a devil's advocate if metamas would have pushed it it would have been way easier to get a standard because they already had like so much of distribution right like we don't even have portfolio proper done like have you tried adding a token on metamask but you know half of the people don't even do it yeah I I think the interesting thing about even metamask unfortunately now is trapped right they're trapped in this like uh this kind of yeah like you know you look at rabi right got adoption by literally pretending give me metamask that's its solution is like we'll use the exact same APIs they have no control over even they can't even make their own APIs right so this kind of uh this oification of the standards means that

even MetaMask is now trapped by this they can't get out of it the only way they can get out of it is if they were to build their own application and they they've built a mini app right they've got swaps inside of MetaMask and they you know tried to build snaps and stuff and it was kind of I didn't think the snaps were also insecure in a lot of ways it didn't really go anywhere but like they they they even saw this. They're like we cannot break out of this kind of uh you know gravity well that we've created, right? Um and so you know what's the solution? The solution is sidestep all of it and just start again and build you know uh from first principles a new thing. It's the 15th standard problem unfortunately because now you've got your own thing and you

know um everyone else has to adopt it. Um but if you build it well enough and it is successful enough and you have distribution, you have a platform on which to kind of use to to uh leverage um your distribution, it can work. And so I'm I'm very confident that we can actually uh break out of this equilibrium that we're in today. Um you know, maybe that's we'll see. Yeah. And that it's actually a good segue to the next question of you know, the entire crosschain UX like we just spoke about the multi-chain UX and it's so bad. Now the crosschain UX is much worse, right?

Um so like and and we need platforms like Infinix that basically like kind of put it all under one umbrella so that everything is abstracted but um how do you think we we solve this in a more scalable way like like because you guys kind of like are like a decentralized exchange way right so you have a huge order flow so will it move towards more of intent side or like you would still find people lping onchain and you know so it'll be like more like cow swap plus across or it'll be like layer zero plus unis swap so when we started out to solve of swapping and bridging. Um we we found LiFi initially, right? And we're like, "Oh, this is a solved problem.

Cool. Okay, we'll just like integrate their API and then we're good to go." Um and immediately we found that there are huge gaps in the the kind of implementation that they as a single aggregator have been able to do. Um, for example, they don't support Stargate Oftens, which sounds fine until you accidentally airdrop a bunch of ENA tokens onto Arbitum and all of your users are trapped there because there's no liquidity to get off the chain. Um, so we've been forced to basically um aggregate all of these aggregators to become a meta aggregator, but then also go and disagregate them as well. Um, and so we now have I think like 15 different bridge and swap providers um both single chain, crosschain, intents, etc. Um, and

even then I would still say we have like 75% coverage. So, you know, we're we're probably going to have to get to like 25 different providers, including um, you know, providers like Ugabooga on Barchain, right, which we just integrated. So you're going to end up having to go all the way down to the DEX layer and then all the way up through the aggregation layer and intents if you want to do it properly, which means make it as seamless as possible so that a user never needs to know what chain they're on or never needs to know what route is optimal. It just does it for you. And even once that's done, then you need to actually be able to do multihop transactions because often times there are multiple routes from one uh you know path from point A to point B. Um, and it's not obvious which one is the best

route necessarily, right? So, if you're going from Salana and you have Solon Salana and you want Ethon arbitum, there are probably four or five different routes that any given time are are more or less optimal. Um, and so you need to be able to orchestrate multihop transactions that stitch together multiple aggregators and bridges in order to be able to optimize uh for the user because the way that we think about it is imagine you had like this um, you know, omniscient omnipotent LLM, right? Like an AGI and it would know at any time the state of every single blockchain and it would know the optimal route for every single order that anyone could potentially propose. And that's the state that we need to get to. We

need to be able to get to that point where we have a thing that is so um certain of the state of all of the different blockchains. And this is like a deterministic problem. You can solve the problem. Um but it just requires really really good orchestration. Yeah. I mean like data is like like I feel like crypto is like one of the fields where we have everything public and available. It's all there but we still don't have data for it. You know, you just you don't want APIs and everything to search. You need really good indicers. is you need you need a bunch of different things um in order to do this really well. Um it's a not it's not a trivial problem but it is solvable.

How do you think about like you know like these new AI updates coming in could shape the way how we interact with crypto. Uh and you know because sometimes some apps are complex as users say uh and how would AI change this because sometimes I I started using build my own small LLM. I actually don't use unis swap or any other interface. I just created like a small small wallet plugged it into an A amm and I can say oh Vitalic is poor the AI already understands that we need to donate E to him you know so how do you think the entire UX is going to change and how would impact let's say Infinex or the space in general so I think that there's uh this kind of movement whenever there's a new tool um where you know everyone wants to use it right so I saw

uh I saw like this proof of concept that someone had created um and tweeted about yesterday um where they'd built this agent that was supposed to be able to do all these actions for you. And the first thing that they did was type what's the price of Bitcoin. And I'm like that's the dumbest like we have a solution for that. It's called Coin Gecko, right? Like the idea that you would need to like use natural language to like type out like you know what's the what's the price of Bitcoin? Um seems a bit silly, but you can imagine saying you know what's the optimal route from going Salana to Bitcoin. Okay, now that's actually more interesting, right?

because it can go and check all the routes for you and you know run the calculations and come back to you with a path and then you click the button. Um so we have a a proof of concept that we've um built which is a very rudimentary orchestration layer where um inside of Infinex you have an agent that will not completely orchestrate the transaction for you but it will basically give you it will render in real time for you all of the buttons that you need to click because you need to actually uh you know pass key sign all of the transactions. So it'll say okay you want to you know um get yield on bear chain um in the form of uh you know some WBTC ETH Kodiak pool or something like that right um so you've

got USDC you need to swap 50% of it into WBTC 50% into ETH you need to bridge and it'll give you the sequence of transactions and then you imagine the next step is okay put all of that and orchestrate it together um but the reality is for today you shouldn't actually trust the LLM right? Like they they you know are not perfect, right? So it's actually easier to see all the sequence of steps and go yes okay I want to do that that that that which is not normie friendly yet but we're getting closer. Yeah. You just need like a tenderly verification as well. You know before you do that make sure you just check the output of this. Yeah. Uh and I and I also feel like I think like let's say now you said that it Infinix got integrated into Barachchain with

Ugabooga right? Ugabooga is the V3 folk I assume or whatever like like whatever the V3 native DEX is and most of the chains are just built on folks there'll be some a compound moro or you know like V3 folks uh and I think like if if AI is good enough to understand what the user wants can understand the addresses on chain right it can actually do the integrations without you guys actually have to push the integrations on the front end right so now you would be like okay like now I want to go on beta chain right and but infinex let's say haven't like manually in an integration, the AI should be smart enough to kind of, you know, figure out, oh, it's a, you know, a V3 clone. You can get that data from D5 Lama. It like, oh, these are the addresses and this is how a swap should

be done, you know. So, yeah, the the issue that we have is that the list of infra for us to support a chain is now very long. It's like a whole page, right? And there's some like, you know, um, alternative solutions we can use, RPC providers, indexers, etc. Um but to get to stand up a new chain insideex is you know bit of a challenge right like there's there's engineering work um I certainly wouldn't give it to an LLM to go and integrate a new chain for us um one thing that I have I have found incredibly interesting this year like just in the last four months I would say um is the uh you know we have uh 50 plus people um in infinex and we're all in a single office right so we all spend all

day together. Um, and I've really noticed that the people who have deeply gone uh down the path of um how to use LLMs and how to optimize, how to, you know, optimize prompts um are really starting to distance themselves from the people who are fading it. We're like, "Ah, it's not ready, whatever." Um, like we have one front-end engineer who basically runs a team of LLMs that like do stuff and he's like, it's like he's running a team of engineers. It's crazy. Um and so you can you can the people who are early adopters who have really leaned into it. Um there's a a big big gulf between them and the average person. Um so that that's been very interesting to me to watch and that goes

for like people in marketing, people in um you know in every different aspect because the especially in crypto there's so much data out there like there's so many references that you can pull for things um and you know if you use deep research and get it to go and you know do a report for you on like the evolution of wallets there's just like millions and millions of pages because there's so many people producing content in crypto um that you can like pull together incredible reports. Um, and it also minimizes hallucinations as well because there is so much source material to go to. Um, so yeah, like I I'm I'm incredibly bullish about the ability to kind of optimize things and and uh increase efficiency. Um, because you

know, in in a D5 project, you need to be as efficient as possible. It's it's hard enough to do your day-to-day job. Um, you know, you need to maximize efficiency. Yeah, that's that's great. Um and what do you think like you know is like the holy grail for Infinix that you that you see you know like what is like when you sleep and when you wake up the dream of Infinix comes to you like what is that the best product you want to build um and why do you think like you you know you will probably be the ones who are like pushing for it and you know contributing towards it. So, so you know, we took our life into our own hands and built this siloed um product that uh that we had full control over, right? Um and it's not perfect. It doesn't uh have all of the functionality

that we would like. Um there's still a lot of work to add yield to add um you know, per spot trading and and a bunch of other functionality that um that we need to be able to compete with centralized exchanges. Um, so we're very happy with that, but it is still an unsatisfying state of the world if you're in a silo. Um, because trying to get people to adopt it is um is challenging. Um, and you can just launch on Solana and there and there's there's always going to be uh some new thing every single day that someone has released that you have not integrated.

you know, even and and if you run the thought experiment, even if we had a thousand engineers or 10,000 engineers um who were working every day to integrate every new thing um as quickly as possible, you're still going to be behind the state of the world, right? Um and so the only way to fix that is to go and expose the things that we've built um this new infrastructure that we've built um to external parties and and let them build it. And and so I think the the most exciting thing for me is we believe we've built a new wallet form factor and this new wallet form factor will enable new applications to be built. Um and so you know we're not ready for this hackathon but hopefully by the time we get to uh France in a few

months we'll be ready and it's going to be very interesting to see what people in in a weekend can hack together um when you give them some new infra to play with. Yeah. If if you give them an SDK and a ETH global hackathon with 50K, you'll you'll find like thousand integrations things done, you know. Exactly. That's the plan. Yeah. Uh so imagine you're like Vitalic for a day. Uh what change would you like to make on Ethereum instantly to you know make it great again or you know push any EIP that you had in mind that ah this should be you know in prod but it's not. Um yeah I've I've talked about this a lot.

Um, you know, as someone who uh worked very hard to meme L2s into existence. Um, and and you know, in fairness, I did try and meme a single L2 into existence. I did try and get us to all coordinate around optimism. And I think if we had all been on optimism, we would have avoided some of the challenges that we've had over the last few years, but unfortunately that was not to be. Um, so in the end, we ended up with 100 or 200 or 500 L2s. Um and the problem we have is that while that has massively increased block space um it has uh created this situation where um the the L1 is no longer profitable right like all of the margin has been captured by

the L2s um and we do not have good economics around blob space and so if I were Vitalic for a day I've said this you know publicly before um I would find a way to increase the price of blobs um in a way that was, you know, that L2s couldn't route around because at the moment there are escape patches they can use. They can just go and and submit um you know directly outside of the blog blob space if they needed to. Um uh so find a way to put pressure on L2s to uh you know decrease their margins and bring the the margin of block space back to L1. I think that is the the single most important thing we can do in the short term um to to rebuild the

narrative for Ethereum. There's a whole bunch of things in the long term like scaling L1, but before you scale L1, we've got enough scale, we just need some more profit. We need to burn some more ETH. So that would be that would be step one. Yeah, makes sense. And I think for the blob, I think there is EIP which tackles this kind of like dynamic markets and pricing for it like even like if L2s want they can hedge it for it and stuff. So I think that's pretty interesting. So maybe if that gets pushed so what you're seeing happens in real life. So yeah. Um so yeah, do you guys have any question? or are you guys happy with just the MetaMax UX? Anyway, since we don't have any questions, we'll wrap it up. Uh, thank you for your time. Yeah, it was great

talking to you. Thank you very much. [Applause]