Exploring the spectrum of intent-driven designs in DeFi - Adam Gągol

hey everyone I'm Adam U from alzero and I'm going to be talking about um intent driven designs so maybe first question who have heard the bzw word intent in the let's say uh d space recently and who knows what it stands for like who could Define I mean I'm not going to ask for it but um okay seems like half of people so uh what is intent intents are well basically glorified limit orders uh and I could end the presentation here because like really this is what the intents are they are limit orders uh but usually when people are talking about intents they have some very specific narrative on mind and uh I'll try to show what this narrative is about so um classically in defi how everything looks or I'm going to be focusing on token exchanges because this this is like the St of of of defi so how it usually looks in defi is that someone let's say wants to exchange one if to usdt sends one if to some smart contract then some logic in this contract happens and then usdt is sent back to the user the user usually have some idea about how many usdt he's going to get in return there is a parameter called tolerance so like he's not going to accept too low number but generally the idea is that he wants to get as much as as possible and there is no strict requirement when it comes to intents it looks somewhat different so what user is actually sending is a very specific in intent or actually you could say a limit order so user says that well uh he owns one if and he wants exactly 2,400 usdt or more in exchange for this if and where this intent LS is like a big pile of intents it's not a smart contract that's going to execute it by itself it's just kind of like a heap of of intents uh that are put there by by various users and then there are guys like solvers they have teeth because this is very competitive market of this solvers they are kind of aggressive but very useful so what happens then they are fighting with each other to get this intent and make it filed and here like what they are usually doing is kind of their their own strategy so they can do whatever they want usually they root this intent through several smart contracts maybe they exchange it on different exchanges maybe they use their own personal fund to funds to uh to feel the intent anyway they are trying to find the best way to fill the intent using entire chain and all the smart contracts that are on this chain so this is like a very competitive market where to feel an intent instead of going to just one smart contract like all chain is there and there is a competition among this this competitive solvers to use entire chain and everything was there to to f the intent or maybe even get more so usually uh like the idea is that the user should get slightly more than he asked for so for example here it's uh 2,48 and the nature of this competition among solvers should be so that the solutions which are giving more to the users are winning so the key question when it comes to this intent driven designs is how do we actually choose which solver gets to fill the intent because what we want is some kind of a game that on one hand will promote the solvers with better Solutions and on the other hand will not um incur too much of the additional gas costs I mean what we want is we want as cheap transactions for the users as as we can so first we want to choose the best Solutions because that's how uh how user will get chip transactions and the other thing is we don't want to user to spend too much on the gas cost for the solver game itself because then we that would counter uh the this optim alties that we're fighting for so there uh and by the way uh currently this intent driven designs there are two which are well known Unis swap X and CI and they indeed offer much better prices than than any single smart contract like like for example Unis swap because uh this game related overhead is really small and well like uh the solvers can be using all the available dexes not not just the Unis swap one okay so there are two types of games that the solvers are playing and uh well maybe that's because there are two well-known protocols that are intent driven right now uh so offchain open bid auction this is the way that CI is is going about it and onchain ddge auction this is what Unis swap X is doing right now so offchain open bat auction means that there is an intent and the Sol ERS are without using the chain playing beats and saying how much more they can give to the user and the one who wins the bit gets the right to actually feel the intent and uh one can think well why is this game play offchain not onchain so like this is ethereum everything onchain is super expensive we we want it to be cheap so the solver which wins the game then executes a normal transaction on chain so like very is basically Zero game related overhead uh so uh like there are several goals here are going to be comparing here so the one of the goal is to be cheap like we are fighting for this uh small small fractions of per percentages of the price so we don't want to lose it all on the gas uh gas prices so when it comes to uh offchain uh auction this is like absolutely optimal gas price because all this over game does not incore any additional gas prices when it comes to onchain D Auction on the other hand it has somewhat suboptimal gas price in the sense that well there needs to be a smart contract which which is orchestrating this Dutch auction and while Dutch auction is generally blockchain friendly uh still this is the gas cost is slightly more than it would be if the transaction will be just executed the other thing uh it's important so like when in this offchain open bit auction when someone wins the auction it happens offchain and no one else has the right to execute the transaction so there is no gas beeding War possible like there is no one else who could claim to fill this particular intent the game is finished off chain there is one person uh which can can legally feel V intent and like no one should be front running this person because very is no no sense in that it's already established who will who will fill V intent when it comes to onchain DCH auction beaing wordss are absolutely possible so dut auction usually works in a way where in each time point we ask uh auctions to uh auction uh to to come with a Solutions with a specific price and it very much can happen that there is more than one solver with the very same solution so what they can do is they can try to front run each other so so like so they get first and they actually get to feel the intent why is this bad well because if solvers are spending some funds on uh on this gas beaing course this means that the user does not get optimal solution I mean optimal solution would be if solver would not spend any excess excess funds for for any kind of fighting any kind of games but only uh only optimalized for the user price if if solver is spending funds on anything else that means that well the the the actor who's going to be paying for that is the user because of course no solver will be paying from his own pocket for for this kind of [Music] games so what we would like it to be also is relatively trustless or well actually fully trustless and decentralized so here the situation flips because when it comes to offchain auctions they are not trustless in the sense that whoever wins the auction well this information about the winner is not onchain so so the the solvers needs to respect the game and they need to respect who won the offchain game uh that's why in CI uh solvers needs to have collateral substantial collateral collateral actually right now I think they are they want half million dollars of collateral just to be a solver uh and that's because if a solver would go malicious he could stop respecting the this this auction game the solver could just onchain try to fill every intent without respecting this offchain game on on winning uh winning the the intent to be filled so if someone would do that of course the collateral would be slashed uh but generally it's somewhat problematic I mean this collateral is substantial it limits the the uh let's say let's say the the um number of solvers available so when it comes to onchain Dutch auctions this is fully permissionless so no no collateral is is required for solvers and is fully trustless and it's also much more simple like this um this offchain game that happens in CI is generally pretty complicated and unfortunately um CI needs some time to to do some manual action when it comes to to uh figuring out who is playing first and who is play who is not I'm not going to go too much into details but this is pretty the rules there are pretty complex so it varies I think D over but definitely that requires manual decisions uh so next goal is me solvers always submit the best trade possible uh and uh when it comes to onra Dutch auction this is partially true so for this who of you who knows how Dutch auction Works uh or for this who doesn't Dutch auction is the type of auction where the with in each moment we are looking for slightly worse solution so uh as someone who wants to participate uh in the auction I'm waiting for the moment in which my solution will get accepted finally because like the bar for acceptance of the solution uh goes down with time so in case uh and like this is very as blockchain is not continuous is it's rather discrete like the blocks are happening every 12 seconds like this is granular it's not continuous the uh decline of this of this bar is pretty granular so like if the bar goes beyond what I have as a solution I would probably submit not my best solution but something slightly worse which still goes above the bar so yes it's pretty incentivized to give the best Solutions not exactly in this regard the the offchain auction is is is doing a better job so just also like the comment on on the state of business right now uh open uh offchain open bit auction is definitely better on ethereum I mean CI has the volume I think 10 or 20 times bigger than Unis swap X so it seems that this gas prices uh aspect on ethereum definitely wins none of this protocol is deployed on any other chain but ethereum like my uh my prediction would be that whenever they get on l2s the situation can flip because uh on L2 the gas price is much less relevant so I would predict that on whenever this two are deployed on arbitrum uh Unis X can actually start winning with CI due to this factors being much less much less relevant and the last thing which is kind of our Twist on it so what we're working on is a model with what we call reveal conditions so now what happens is Joe places this intent let's say or limit order but it's encrypted and there is a specific reveal condition attached to the order so then periodically all the reveal conditions are are checked and the orders with the reveal conditions which are satisfied are generally uh decrypted and are are becoming active so why we think it's nice is because first uh you can create Iceberg orders so you can put your order on chain in such a way that it's encrypted and it automatically gets decrypted part by part so it's an invisible order until it gets filled when that's when it becomes visible so this is AK to what what's happening on traditional markets where vast majority of liquidity is actually hidden so in crypto to to make it so right now you actually need to have some kind of a bot that would be resitting transactions so this is a cryptographic way to avoid having this offchain actor you can make your action onchain once and it's going to be acting as as an iceberg order so this relates to the second point that like Traders without automated systems like like Bots can automate something by sending transaction on train ones uh the other thing is that using this system I mean I'm getting a bit out of time but using this system you can implement this uh this solver games as well so solvers can also put this encrypt in intents uh which will be decrypted at once and evaluate at this specific point so you can write arbitrary rules for this solver game and like it basically fills this framework so we see it as a programmatic framework for uh for kind of private trading um but the last slide also we're out of time so uh thanks a lot for listening I think if anyone gets a question it's better to ask off or you have like two minutes question well perfect so uh yeah if anyone has a question I'll be happy to answer now