Keynote: Unifying Ethereum Through Intents and ERC-7683 by Hart Lambur | Devcon SEA

Um so, my talk here following up on that panel, an amazing panel, and let's give a thank you to uh to Vitalik, to Yeah, well I thank Vitalik, but also Ben, Jesse, Steven. Um that was an awesome talk. So, uh we're going to talk continue the conversation about how we unify Ethereum uh with intents and ERC-7683. Um and so, I'm Hart. I am the uh co-founder of Across Protocol.

Uh we are an intent-based interoperability and bridging protocol. Um but we're going to talk about the standard and how we developed it and how we built support for it, um which we talked about on that panel, too. Okay. So, I stole some of the slides for my intro to the panel uh because they were so good. I I needed to use them.

But uh I'm going to replay quickly some of the slides just to set the framework for for what we're talking about here. So, 4 years ago, Ethereum had the scaling problem. Uh we had 15 transactions per second. At peak demand, Ethereum was so expensive, it was basically unusable. And Vitalik helped us align the Ethereum community around this roll-up-centric roadmap.

And this worked. Um I walked through these slides uh on the panel we were just on, but like look at what we did here, right? In October 2020, we aligned on the roll-up-centric roadmap. Uh Optimism shipped a couple months later. Arbitrum shipped a couple months after that.

ZK Sync shipped a year after that with ZK roll-ups. Uh we had this like massive innovation in this roll-up-centric architecture. And then of course, blobs came around just this year, so March of this year, and I don't know if you guys just saw the charts or I I I honestly should have included another chart in here. Uh the cost of transactions on Ethereum just plummeted. Um and now on an L2, you can do a transaction for pennies.

Um we actually saw this in a crosses data. The median transaction size dropped from about 500 bucks going L2 to L2 when you're bridging, it's now 50 bucks. And we have real consumer use cases because these these L2s are now so cheap and usable, normal people, not just whales, can use them. And so, this is a beautiful chart. Like, we really did scale Ethereum thanks to the roll-up centric roadmap and some of the people that are just on stage with me.

Here we go. We scaled Ethereum. And this is something that I don't think we celebrate enough. Uh, we solved the big hairy problem and uh, we should be proud of that. New problem.

Ethereum's fragmented. It was so simple back in the day when we had one chain. It was like such a beautiful thing. I had MetaMask, I had one chain, I hit one button, it did the thing on Ethereum, and it worked great. Um, but uh, it got more complicated.

So, our roll-up centric roadmap now means we have lots and lots of chains. Um, and we got to like do this network switching thing. And sure, some wallets are beginning to hide this for us, but man, like switch between all these chains, add a new network to MetaMask, it's a mess. You got so many versions of Ethereum. There's just so many different ones.

Uh, it's it's a total mess. And I I think it's just worth us um, admitting that we have a major UX problem here. This is This is not something that the next billion people are going to be able to use, even though it's fast and cheap on any individual roll-up, this is this is a mess for the the the users. What we got to do, we got to unify Ethereum. This is This should be the rallying cry for 2025.

This is what I want uh, Ethereum to be focused on for the next year. I hope it doesn't take us four years to unify Ethereum. I want it to take one. Um, Vitalik seems optimistic. Uh, I hope I hope we're we're all optimistic here.

And we want Ethereum to feel like one chain again. Um we want it to be like, "Hey, I click one button in one wallet, and it does what I want to do on whatever chain I'm on." Um and Jesse was speaking about this on the panel we just had. Um so pretty bullish that there's like people running big wallets like Coinbase Wallet that are aligned and thinking this way, too. And so there are a few components, I think, of how we actually make you know Ethereum feel unified.

It's the better wallet and app UX. Uh I think there's a necessary component around account abstraction, specifically EIP-7702, which I'm going to talk a bit about later on. And then there's speed. Um to to make Ethereum feel unified, I think interoperability needs to be fast. Like really fast.

And for me, fast means um 2 seconds fast. And it kind of seems like I'm making up this 2-second number. Like where is this coming from? It isn't totally made up. Um there's lots and lots of Web2 UX research that says if you wait for more than 2 seconds for a website to load, users just bounce.

Like they just don't want to use the thing. And I don't know if you guys have like used airplane Wi-Fi when websites load forever, it just makes you not want to surf the internet. It makes you not want to use the thing. And there seems to be this psychological barrier that I can I can afford to wait 2 seconds, but any longer than that, I want to go do something else. So if we want the experience for the next billion users, we want to compete with Web2, I really truly believe we have to hit this 2-second interoperability time.

And intents, what what we do with Across, uh it has a different tint icon. That's kind of cool. Um intents, what we do with Across, uh are the only viable mechanism to deliver this 2-second experience. Um and we do the I'll go into how we structure this, but like right now today in the Across protocol, the median fill time between chains is about 3 seconds. We'll go into some data on this, but it's fast.

And and why is that? So, it's worth walking through what the traditional bridging flow is when we're connecting blockchains and why this intent architecture is different and faster. Okay. So, traditional bridge. I got two chains, A and B.

I got Arbitrum and I got base. Um step one, I want to deposit my assets on chain A. And then I wait. I wait for finality. I can't the protocol can't do anything in the traditional architecture until it knows that those funds are actually on chain A.

And this waiting is a a problem. Once I'm done waiting for finality or for a sense of finality, then I can go and send my message to chain B. And then I can release my funds back to the user. And so this flow is all fine and good. Maybe I have some security assumptions around how I'm sending this message.

Maybe I don't love those. But generally speaking, like this architecture is like conceptually simple. The problem is it's just slow. Um I have to wait for finality or at least a sense of it. I can't get the two-second experience with this traditional bridge architecture.

So, how do intents work? How are they different? Intents, I still have users deposit assets on the origin chain. I need that. But when they're depositing assets, they go they get escrowed into a special account that broadcasts the intent to a network of solvers.

And this network of solvers are going to then compete to fill the user on the destination chain. Um importantly, they're filling the user with their own funds. So, what these guys are doing is there's actually a marketplace that's taking on the finality risk of the origin chain and they're competing to see who's going to be the first one to fill the user um in exchange for the the transaction fee or the fee charged here. And this architecture, if you notice, there's no message that gets sent between chains. Um and the finality risk is then being taken on by the solver network.

So, this is fast. We can fill the users in in 2 seconds. And the user can move on with their day, and we achieve this fast inner up that can help us unify Ethereum. Of course, there's still one step after this. After the user gets filled, we still need to verify that the fill happened and release funds back to the solver.

But we have time to do that cuz the user got their fast fill. So, Across, you know, I'm slightly biased, but I do think Across is the leading intent-based inner up protocol. Today in production, how many people in the audience have used Across? I'm actually kind of curious. That makes me feel good.

All right, okay. So, Across, we are bridging protocol. Um if you go and those that haven't, if you go to across.two and bridge between Arbitrum and Base, your fill time will be less than 3 seconds. Our median fill time is 3 seconds, and we're keep lowering that.

As block time block times on L2s get faster, it's going to get cheaper and cheaper. For transactions less than $10,000, our average cost is less than 4 cents. It's extremely cheap, and there's room for this to get cheaper, too. And we've done billions billions of dollars in volume. You know, today I think we did 50 million in volume with 50,000 transactions, and it's growing.

So, this works in production. This isn't a theoretical architecture. This is working. But, you know, we we're looking at Across, and I don't believe that interoperability is going to be solved by one protocol. Like, sure, I might like that, but I don't think that's the path.

And go back and think about the roll-up-centric roadmap that Vitalik set for scaling Ethereum. It wasn't solved by one team. It wasn't solved by Optimism. It wasn't solved by Arbitrum. It wasn't solved by ZK sync.

It was an architectural decision that many teams innovated on. And I think the same is true for inner up. And so, to make this intent architecture really like win, we to define a standard. Um And so, we worked together, the Across team worked with the Uniswap team to develop ERC-7683, which is a standard for cross-chain intents. Um it's not owned by Across, it's not owned by Uniswap.

It is a true, uh ERC, uh that's a public good. And it's worth going into like, hey, what what is this thing do? What is this ERC? Um And my mental model, the way I I kind of think y'all should think about this, is it's like an order ticket. Um you can almost imagine this if we go way back, like 1980s finance, like Wolf Wolf of Wall Street stuff.

You like had a paper order ticket that your broker filled out with like, I want to, you know, sell Apple stock and buy Google stock. Although, Apple and Google didn't exist back then, but you get what I'm saying. And then they'd like send it to the floor floor, and some guy on the trading floor would make like hand signals to go and like execute the thing. But they had a format, they had a standard for how they express this trade ticket. Um and so, you can think about ERC-7683 as defining a common format to express this user intent, this trade ticket.

And what you do, what a developer would does, is they go and they fill in their trade ticket, origin chain, origin amount, destination chain, destination amount. Um And then, uh we got a couple of the fields here that I want to talk about. But, you fill this out and you send it to the solver network, and it just gets done. It just works. So, I got these two other fields in here, fill instructions and settlement system.

Um and this is a little bit stylized, the just how this works, but fill instructions, you can think of as like a memo field, where I can include extra information about what I want to have happen in this intent. Um and uh remember, the solver is the one that's moving the assets to the destination chain and sending them to the user. So, there's nothing that prevents that solver from executing arbitrary code. So, the fill instructions can be say, "Hey, let's bridge one ETH and then swap it for frog token once you get to the destination chain." Totally possible.

That's actually something you can do with the cross today. And it could be bridge one ETH and buy an NFT. And it could be bridge one ETH and deposit in an Aave or whatever your DeFi protocol is. And this can happen in one click, remember. So, in one click on my origin chain, let's say it's Arbitrum, one click on origin chain on Arbitrum, I can have this action happen inside 2 seconds on my destination.

Which I think is pretty cool. Um and I think this little really does lay the groundwork for how we unify Ethereum and make these chains feel like one thing. Um but I mentioned this earlier, one of the things I get super excited about is once we combine this with a ubiquitous account abstraction. Um this would be an entire other talk. Uh so, I'll just keep it high-level.

Um but EIP-7702, for those of you that that aren't familiar with four-digit EIP numbers, uh it's actually a EIP that Vitalik wrote, an improvement on some other technologies like 3705, I believe, um that will uh bring smart account uh capabilities to everyone's EOA. And this is going to ship. This is scheduled to ship as part of the Petra hard fork in the first quarter of next year. And so, what this means now is that anyone's account can you can send a message and somebody else can execute that message like it was being sent from a EOA. Um and what this means is that I can now remote control any remote chain from my home chain.

And how this would work is I would sign an intent, a 7702 message for the action I want to have happen on my remote chain, and I would include it in as a payload in this fill instructions field um in our 7683 standard and then I would use the solver network, the intense solver network, to basically bring the gas money over to go and execute this arbitrary transaction. Um and again, we'll be talking more about this. There's a probably a whole other conversation or talk to have, but there's this very cool way that we've got all the LEGO blocks and pieces when we add together four-digit numbers like 7702 and 7683. We add them together and we're going to have this like I really do think of it as like a remote control where from my home chain I can do an arbitrary action on the destination chain. This standard also modularizes, it's a hard word to say, it modularizes the settlement system, too.

Um so, what do I mean by that? If you guys go back on the other slides where I was kind of walking through the intent architecture, user funds get escrowed on the origin chain and then they only get released to the solver when we verify that fill happened. Um I'm calling that the settlement system. The system that it it serves two purposes. It It keeps user funds safe where it doesn't release user funds to a solver uh until the fill happens and it also keeps solvers safe where it doesn't let solvers get rugged by users.

So, in this standard, that settlement system is is a module and you can plug in what you want. Uh Across has a version of this that we've been operating for 2 years that works pretty well, but it's by no means the end state. And what I think is super cool about this and and Jesse on the panel beforehand was was talking exactly about this idea. If I have two chains, if my origin chain and my destination chain are both superchains, I could use a settlement system that uses uh optimism uh superchain native interop um and that could be faster to repay the solver faster, which would make the intent cheaper. So, I have all the flexibility in the world to choose to to to just optimize uh my settlement system based on the origin and destination chains.

And this works with um Agler, uh Polygon. Can we can use the Polygon Agler system where if I have my origin and destination chain is both Agler enabled chains, I can write a settlement system that uses Agler to to uh to speed up solver repayment. And Jesse talked about uh Coinbase's push. Coinbase uh Wilson Cusack and Jack Kuma from Coinbase are doing this really innovative push for RIP-7755, which is like a trust-minimized nested storage proof messaging idea. We could plug that in as a settlement system, and we're actually working with the base team on a proof of concept to do just this.

And so, uh what I find super powerful is that 7683 supports any settlement system, uh Superchain, 7755, Agler Elastic Chain, chain clusters, succinct SP1. If I want to write a settlement system that uses ZK proofs to prove that our intents were filled, I can do that. I- If you want to use one of the other generalized messaging passing frameworks like LayerZero, Hyperlane, Wormhole, Chainlink CCIP, completely compatible, too. Um and so, this standard transparently upgrades with whatever the latest and greatest interop techniques are. And uh Ben Jones was talking about this being a mullet, right?

We put the standard in the front, and then you got this party in the back with whatever whatever um infrastructure best serves verifying that intent, we can seamlessly upgrade and utilize that. Why is this standard helpful? So, developers for both the app and wallet, we now have a single common format to do anything cross-chain. Developers should implement the standard. Just works.

Um and again, it is going to be transparently upgradeable as we get better technologies to verify the settlement layer. For solvers, solvers now um don't listen to like one common format without needing custom integrations for each new protocol. So, solvers just listen to the 7683 order standard and any intent-based protocol that builds on top of it um uh is is they're able to serve. Okay. So, um Across and Uniswap worked together to develop this initial standard.

Uh we got a lot of great feedback um from many people in the space. Uh that feedback was actually merged in about 2 weeks ago. So, anybody who's read the standard, go read it again. It's been updated. Uh and if you want to read more, erc7683.

org has all the details. Um but we've been trying to build support for the standard. And uh Optimism um was a an early adopter. Uh they uh they understood that even though they're working on superchain native interop, they don't just want interoperability between the superchain. They need interoperability so like users on Arbitrum can come to the superchain and what have you.

And they also understand that the standard, you know, it can use superchain native interop on the back end to allow for faster intent verification. So, they thought this was a good idea. And the Polygon team thought this was a good idea. And I'm going to read out some names for a second. We got Linea, we got Alchemy, we got Blast, we got Conduit, Manta, Grelot Pie, Mantle, Mode, Taiko, Superseed, Worldcoin, Connext, uh Scroll, Nevermind, I'm almost done, don't worry.

Um AltLayer, uh Aori, Fluent, Subwallet, Open Social, Zero Dev, Rhinestone. We got a whole bunch of teams coming to announce support for for standard, which honestly, guys, like I didn't really think this would happen that well, and it did. It was pretty cool. Um Oh. I will tell you guys, so we had this hole here, right?

So, this is a hole here as of like 2 days ago, and it kind of it made the chart look more annoying. And this should also all be in a big tent. I want these guys all in a big tent, but it didn't fit on screen, so um I'm What's cool though is you come to a place like Devcon and you you meet people and you kind of get them over the line and explain what you're working on. And so, now we have both Arbitrum with Offchain Labs and ZK Sync uh Matter Labs ZK Sync also supporting the standard, which means we have like complete coverage of the major L2 ecosystems saying, you know, "Hey, we think this is a good idea." Just like ERC-20s defined a token standard, we think this is a pretty cool way to define a standard for intents.

Um Okay, so I'm almost done. So, what's next? Um we have broad support. Um I've never tried to build support for a standard before. Working with Uniswap, it's been great, but many people in this ecosystem are just like, "Hey, yeah, there's good social capital to try to unify Ethereum, and this is a lightweight standard that lets us do this."

So, we've got broad support um from the L2 ecosystems. We need to figure out better dapp and wallet UX patterns. Um that's like the next step here. And then we really want to do this like full integration with 7702, what I was talking about before, for this beautiful chain abstraction where I I use a 7702 uh uh uh uh kind of user op payload to have uh the ability to do a a remote control where I can remotely execute any action on a destination uh chain. Um And so, guys, that's that's my pitch for how we're going to unify Ethereum and how we're going to do it quickly in 2025.

It's not going to take us 4 years. We're going to use intents and the 7683 standard to unify Ethereum. Um you can follow me on Twitter at Hal 2001, and thank you very much. Thank you. You want want me to do it?

Yeah. I think it's running out of the battery, maybe? Yes, can you check it? Thank you. I'm personally excited about what you just sharing.

It's huge. It's huge. All right, let's talk about Q&A. Will this new standard support NFT? Okay, I got to I got to go deeper on some of this.

Will 7683 support 721 and 1155? Guys, I can't remember all the EIP numbers. I try really hard. Um I got to go deeper on this. So, um we will get back to you on Twitter with that and or I can talk to you.

Oh, you fungible NFTs? No, we can't support NFTs. I'm sorry. Um remember that the intent architecture relies on the solver filling the user on the destination chain with their own funds. And NFTs are non-fungible.

So, the intent architecture relies on fungibility. Um that's actually one of the core insights that allowed us to be so fast. So, um non-fungible assets do require a different thing, and I think that's uh that's important to to kind of understand. All right, thank you. Next one, can you compare and contrast maybe like a general multi-chain intents that we existed on the market with the 7683?

Um so Suave so I'm a huge fan of Suave and what the Flashbots team is working on. Um I look at Suave as like computation for like a TE trusted enclave to compute all sorts of interesting and private things. Um lost the screen, but um they're not in competition. I actually think Suave is something we have looked at quite carefully to build better auction mechanisms to let us have like trustless auctions to auction off intents. So, I think these things are very complimentary where you can use 7683 as a common order standard for an intent and use something like Suave to to achieve optimal execution of that intent.

Okay. All right. Next one. Can we use this new standard for the non-defy protocol like ENS, social fi? What do you think about it?

So, I think this goes a little bit back to like the NFT question. So, um the intent architecture is not for passing generalized messages. That's the back-end part of this, right? The intent architecture is I want to move funds quickly and maybe I want to do something with those funds on the destination chain. Um so, if you're trying to uh move NFTs around or trying to move like ENS names that are sort of that are conceptualized as an NFT, you need something else, right?

We can do things like, "Hey, you want to buy your ENS name, um you can go and pay for it from any chain and make that be one click." So, I take my Arbitrum funds or my Optimism funds and I go and I buy my ENS name on like Eth mainnet and it's one click to go and execute that. But, these are slightly different concepts. And what what Sorry, my question's like why should why cannot we just like merge it together into the same standard? So, um go back to the idea of how the intent architecture works.

You are going to have these solvers front the user funds on the destination chain. I can't front them an NFT that's on some chain. I need to create another representation of it like trustlessly. So, fungibility, the fact that we have fungibility, is what allows the intent architecture to work so well. If I'm trying to do something that can't be fungible, like money money is fungible or tokens are fungible.

If it's not fungible, then I'm going to need to like send a lower level message to make that happen. Okay. Uh we got 10 15 seconds, but yeah, can you can you answer the the top one? Uh how can 7683 prevent devs from messing around with the intent the user would have specified those? Um so uh So the the um the user and the solver both have to agree and trust the settlement system.

Okay. If you write a dodgy settlement system, um in my example there, yeah, you could steal user funds and that's bad. Um do I expect the user to to sanity check their settlement systems? No, but I do expect wallets or other applications that implement this to sanity check that settlement system and or, you know, we'll create