[trustless://interop.landscape] Orest Tarasiuk - Cross-chain apps powered by real-time proving

Transcript

Our next talk is going to focus on crosschain apps powered by realtime proving. And get excited. We're going to see another demo. We love to see a good demo because it shows us exactly how things work. In this talk, we will dive into X Yield, a crosschain vault optimizer built on T1's realtime proving based on crosschain application infrastructure. Exile enables yield optimization in the multi-chain world by automating deposits, withdrawals, and bridging across yield sources on multiple chains based on verifiable strategies proven onchain by TE's. So that was a mouthful. I'm sure that it will be it will make a lot more sense at least for me once we see the demo. We

solve interrupt challenges by eliminating the need to switch networks and bridge funds because at the end of the day things have to be easy for things to be usable and not just if we want more adoption but for ourselves as well. Please welcome to the stage or thank you so much. Hi guys, welcome and thank you so much for staying to uh this late hour. Uh I appreciate that very much. I know it's been a long day and it's super cool to see so uh many of you actually care about uh not just interoperability and the interop landscape but also the prefix of this event which not for no

reason is called uh trustless. We have the trust manifesto over here and uh I'm very glad that uh we've just heard a talk by uh Steven from offchain talking about how interoperability should actually be reflected in the wallet. um because I can call out a couple of elements in that approach that are not exactly trustless like we can talk about RPCs but we can also talk about uh proving and uh following the different chains but then again there's also the user experience element and this is something I would like to focus on the most today so what I would like to present to you today is a very quick uh recap of uh my rough thinking about uh

creating ing interoperability in a world where there is 15B on arbitum and 15B on base and other rollups that are incumbents and that have the staying power etc. H so I would like to give you a recap of how to create interoperability without changing them. I think we align with the previous speaker on that. Um, but I would I also like to show you how creating programmability in this paradigm unlocks new use cases. Um, and by programmability I mean not having to actually go ahead and click a button in your wallet every day. And and finally, I would like to show you a demo

of an app that is made possible by infrastructure of this sort. Okay. So how what's crypto like today? I guess it's obvious to anyone here. We have a lot of islands that have pretty good um intracluster same island interoperability like the aforementioned um rollup clusters but and and and one could state well this all those intent bridges and if I want to have five E or a thousand dollars from arbitum base I can easily bridge. However, um I would argue that this is not what interoperability is about. I would say that interoperability is about

composability, not just bridging. And co composability is the idea that you can actually have this program that does the swap on arbitrum against your ease gets you USDC and then you could have another program on scroll that consumes the output of the other um arbitrum uh based program to do something um yet again and its output can be used by yet another rollups dab and so forth. So basically the idea of the original Linux CLI vision with one tool to do one highly specialized job and that job alone but combined in a paradigm where it's very

easy to just pipe the output of this highly specialized piece of software into the input of this other specialized piece of software. And that without um those individual constituents having awareness of each other, without them um necessarily having been programmed to be aware of how the inputs and outputs could be composed. And so if you combine programmability with real time improving, I I would argue only then is it that we get to real composability, to real trustless interoperability.

um and and this long for user experience of a click and forget um user journey. Now in terms of uh numbers status quo is we have optimistic rollups with multi-day delays to finalize on the L1 and we have ZK rollups with maybe an hour of a delay to finalize on DL1. And technically you could argue we have composibility already. It just happens to be super slow and in fact it's as slow as though to prevent anyone from practically leaning on that um paradigm.

You don't see applications where you do something on arbitrum use the canonical bridge to go via DL1 and then deposit that uh maybe output into scroll and and and do something on scroll. So yeah, that doesn't happen. Now with an infrastructure like T1's what you can do is you can take a network of uh security hardened TE's run follower nodes for all of those existing rollups sequences that are actually consuming the state as it gets produced and exposing the ability to read that state from any other rollup or

from the T1 native block space. Now with this you already get certain benefits such as the ability to quickly prove intent fields and therefore provide better quotes. But what's really interesting is enhancing this infrastructure with the ability for such a crosschain DAP to request a read happen in one of the uh incumbent rollups. So what you can do then is you can react to a state change in arbitrum. You can maybe withdraw your your uh position from a vault on arbitrum and you can uh instantaneously bridge the um output tokens and open

another position on base and all of that would be done with this read and write capability. It is a best effort base but a very versatile one within the harden TE infrastructure. So basically in uh developer terms what you can think about as a parallel is the asynchronous programming paradigm with callbacks where you may as a developer um request that some transaction be put on on on base and uh at some point in the future will hopefully the success call back be called to which you can

then react in your application logic or a timeout will hit because there is no guarantee that the success will ever happen and then you can revert and unfortunately you must indeed revert the previous legs of your metat transaction of your super transaction to restore a consistent state. Now this is what uh and in in and our thesis hopefully is the best we can get to in the world as it is today with the incumbent rollups to approximate the original vision of the money uh building blocks that um DeFi was enabled by in the first place but in this crosschain interoperability compatible paradigm

the team behind this is uh my co-founder John uh who previously co-ounded secret network using tees also for privacy and learned his lessons there. This is one of the ingredients for the design we've been uh working um on um because we we we learned the lesson of uh securing tees and identified how much easier it is to actually be able to just reproduce computation within a TE and then have attributability in offenses that the TE might have committed versus leaking secrets which is not attributable.

And then there's me. Uh I was pre I previously co-founded two non-crypto startups and then worked at scroll again learning about the limits of practical zk applications learning about the strengths but also the weaknesses of a zkon uh system hopefully distilling the lessons and hopefully um putting them to good use. Now we are talking about crosschain applications. We are here to work like go from the user perspective and uh why does any of that protocol design stuff matter in the first place? Well, it matters because in a paradigm with interoperability being easily available,

the user experience actually improves significantly and the user experience um can indeed even be quantified in terms of financial gains or savings um as relevant in DeFi applications. So, one example I'd like to guide you through today is what I call yield. We literally just came up with this name uh this morning. So uh I hope no one will uh frontr run us to the domain. Uh the idea is how about instead of the user journey that uh the previous speaker just described where the wallet does things for you. How about you um express the idea of not caring which

Oiler deployment you want to put your funds to in terms of chain. You don't care if it if if if the Oiler vault is on base or arbitrum or scroll, but you care that it's the Oiler vault such and such with the risk profile such and such. So how about you express that desire and then let a uh sec security hardened TE network run this application run this program that tries and uh materializes your desire using the ability to talk to sequences of the of the chains involved and so what it would do is it would check okay what's the

yield on Oiler base what's the yield on Ola arbitum and it would go and bridge as necessary and open a position on the user's behalf. The idea being that um this yield, this API could change every day. And so the ag [snorts] sort of agent running in this TE secured programmable paradigm could uh be checking APYs continuously and it could even be running way more complex trading strategies.

So if the demo gods are with us today, ah great you we will be able to see a live demo of this. But to give you a quick gut feel on why we care, we ran a couple of uh we we crunched a couple of numbers and we found out that even at a very uh like short term a couple of bridges involved potentially we are able to generate a yield which is up to 20 what is it 30% higher um in a world where bridging between the different um instantiations of a given USDC volt as possible versus just staying on one chain. So this is the money that you are

basically forgoing by not continuously opening your wallet and and uh moving funds manually. Whoa. [snorts] Okay. So let's see if demo gods are with us and we will be trying Oops. we will be trying to do the thing that's that the TE is supposed to be doing automatically for you live. So let's see if uh the Wi-Fi is with us. It seems so. So what you can see here is the uh interface of yield. Um the interface basically talks to a back end which is hosted in a hardened TE. Um and we can see okay there's this ola golet gauntlet

USDC vault. Let's go ahead and fire it up. And whoops. And so we see okay uh we have uh a total deposit number of 140 bucks. Let's connect my wallet. A couple of days ago I deposited some money. I deposited something over 100 bucks. And I can see okay cool. My position is worth 126 USDC currently. I acred some yield like this the total yield acred and interestingly I can also see the rebalance history. So you can see okay um November 6th two weeks ago um arbitum was the chain on

which the yield was highest and so from November 6th to November 17th did the TE agent keep user funds in that uh deployment. um getting an average APY of a stunning 9.5%. But then the market moved whatever and on November 17th um the the agent decided to move funds from arbitum to base and it remains um the most attractive vault deployment to this day. And we can confirm that by going to oil scan and we see okay uh the average volt is at 3.6% 6% and the uh base vault is at 8.7%.

This looks right. So, how about we do the hard thing? Let's assume a user Alice is on the chain that does not have the high yield. So, in this case, let's assume she's on arbitrum where the high yield is on base. And let's see what the UX of uh T1 enabled infra looks like. Let us try and deposit $40 into this uh crosschain vault. And and you can see there's no permit required. We are we will be sending 40 bucks to an EOA that was generated and is managed by this dedicated TE.

Let's do that. And uh we can see then on Arbitum. Let's open Alice's account. This is Alice's account on Arbitum and we can see token transfers 10 seconds ago. Um Alice moved 40 bucks um out of her account to this dedicated uh EOA run by the TE. Let's open that account and see what's going on there. We see 25 seconds ago, Alice deposited into this um dedicated account. And a couple of seconds later, the um agent um decided, hey, I I actually want to bridge over to the other network. In this case, it it decided to go with

across um and it sent $40 to the other network. So, let's see what's going on here in our front end if our deposit arrived. And there we go. We can see okay we have $166 um total deposit value it is acuring interest on base and should the market conditions change will the TE agent automatically move the money to yet another chain and ape into yet another vault um to earn yield that was the demo um now how Is this even safe? How is this even secure? You

could ask. We've heard about all those TE exploit and TDX fail and TE fail and whatnot. And I keep saying um security hardened TES. I keep saying that for a reason because again we we all learned our lessons and uh TEES can only be trusted with certain things and should not be trusted with others. In our situation, there will be significant funds secured by the TE. So, it's important to take precautions against untrusted node operators, etc. And so what we elected to go with is a tiered security architecture that I call defense in

depth which starts with fast and efficient TE proofs that are running computation uh TEES that are running computation and creating proofs that are verified onchain and then in the next step will we will we be adding multiple TEES uh coming to consensus about this compute So that they need to basically vote on the output of the program after it ran and uh reach consensus there. Um possibly an optionally secured by restake stake.

But then you could ask what if there's like billions of dollars being secured by this system? How can you ever reach this level of stake? Um and the answer here is um we have a the idea of keeping track of what the value that is being secured by this um consensus of TEES amounts to and as it approaches the cryptoeconomic security limits of the system. Would we encourage the creation of an ad hoc zero knowledge proof of all the state transitions up until now? And with this, should this ZKP be able to be verified on chain? Would the system be

able to securely reset the value at risk counter and therefore continue operating with uh with low latency and and finalize on uh um on on chain um without issues. And then in the extreme scenario where single transaction or like a very small amount of transactions were to exhaust the cryptoeconomic security budget instantly, would we need to actually hold finalizing and require a zero knowledge proof to continue finalization on onchain?

And so overall uh we have a couple of more ideas around creating censorship resistance in the system. On top of this, not everyone will be able to easily get a hold of a TE. Not everyone will be able to actually run several rollups follower nodes. This is quite hardware um uh quite beefy hardware that you need etc. And so to create censorship resistance, we are also in introducing in our long-term vision an additional role um that I call the sequencing as the sequencing um actively validated service here um which only is tasked with um ordering user transactions like things such as the uh

metamas transaction that I just uh showed you into bundles, signing off on those bundles and only then um passing them over to the execution AVS which is running inside the TE which is able to decrypt those transaction bundles and and and and uh only then execute them and which is then um actually running the uh state transition and and the new state root submission on chain. but they do not have the ability to reorder user transactions or to exclude users, sensor users. This we hope will create a system which hits a good balance between uh

latency and censorship resistance in in this case. Finally, quick road map to to to mainet like this is infra so of course things take time but we elect to take an uh um an approach that's quite lean and we start with a very uh lean version of our mainet which we just announced a couple of days ago on which yielded is running already like you can what what we've seen before was real funds on arbitra mainet and base mainet using real USDC and this is life already. Um, next up, will we be securing this network and uh adding permissionless block space in

which any um application may be able to um benefit from the ability to interact with uh other rollups? And finally, will we be opening up the network to towards a decentralized architecture? And so my call to action for you today is if you are interested in taking yield for a spin, please hit us up. This is a um a link to a very short form uh where we will be selectively inviting people who want to risk their funds on mainet um in exchange for maybe high yield. so far has been significantly higher than uh staying naively on one chain and

definitely a lot of satisfaction of finding out how a post interop post fragmented uh Ethereum world might look like. So please uh hit me up if you have any questions and if we have time for questions I would love to take some. Do we have time? >> Okay so thank you very much guys. Thank you very much.