# ELI5: Account Abstraction by Liraz | Devcon Bogotá

- Channel: [Devcon](https://streameth.org/devcon)
- Date: 2023-10-09
- Duration: 50:24
- Topics: Science & Technology
- Watch: https://streameth.org/watch/yt-QuYZWJj65AY
- YouTube: https://www.youtube.com/watch?v=QuYZWJj65AY

## Description

Visit the https://archive.devcon.org/ to gain access to the entire library of Devcon talks with the ease of filtering, playlists, personalized suggestions, decentralized access on Swarm, IPFS and more.
https://archive.devcon.org/archive/watch/6/eli5-account-abstraction/

.

Speaker(s): Liraz
Skill level: Beginner
Track: Developer Infrastructure
Keywords: account,abstraction

Follow us: https://twitter.com/efdevcon, https://twitter.com/ethereum
Learn more about devcon: https://www.devcon.org/
Learn more about ethereum: https://ethereum.org/ 

Devcon is the Ethereum conference for developers, researchers, thinkers, and makers. 
Devcon 6 was held in Bogotá, Colombia on Oct 11 - 14, 2022.
Devcon is organized and presented by the Ethereum Foundation, with the support of our sponsors. To find out more, please visit https://ethereum.foundation/

## Transcript

foreign [Music] I know your name and we've been working on account abstraction in various forms for a few years now and you have ice who is not here is from the ethereum foundation and he has been our colleague and um yeah we're working on this in collaboration with never mind and originally we were working as the open GSN team on meta transactions and all of it kind of merged into account abstraction the erc4337 standard but we're going to talk about how account abstraction is generally useful regardless of the implementation so let's talk a little bit about counts and ethereum So Right Now the default account well the what you get is an externally owned account and it is a one-size-fits all EC DSi key and the other option you have if you um if you want to have like your own smart contract that would be controlled by code but you would still need an eoa to interact with it probably your current wallet is an externally owned address there are some limitations with that the first one is maybe some of us have gotten used to it by now but Key Management can be hard and there's a little bit of a paradox with key management intrinsically because there's a secret that you're on one hand trying really hard not to lose because if you lose it you can't do anything with your account anymore and when you don't want to lose something what do you do well you create many copies of it and you make sure the copies don't get lost but on the other hand you also don't want the key to be stolen and we don't want something to be stolen what do you do you you know try to hide uh h-prom adversaries uh make it really hard to access mini minimize how many copies there are to be still in the first place and the engineering strategies when you just have this one uh key that you're trying both to prevent from being stolen and from being lost they're kind of uh mutually exclusive or they they pull in different directions so Key Management is hard even if you know what you're doing the other big limitation of EOS is just Access Control whether you're just playing a blockchain game or you're a multi-billion dollar Corporation if you're using an eua it's like the same Access Control policy so that's very limiting you don't have uh granularity um you wouldn't have multi-sig support there is no ability to implement roles um you can't have spending policies it's just one size fits all mechanism based on this access to a secret key the other big limitation of EOS is Gas payment so you're paying gas directly from the eua and it has to be an East the native token um now potentially there are many other ways that you might want to be paying gas if you hold some you know if you're a fan of dog coins maybe you want to pay gas in that you don't want to hold these um and uh the other limb big limitation is that well if you do have to maintain an eat balance in every account it's hard to split up your activity between a lot of accounts because then you have to top off faith and all of the accounts and the eat that you top off it has to come from somewhere that's probably attached to your identity I mean kyc at exchanges um and our identity is known to at least some players and then yeah it's not too hard to track and even if we split up activity oh this is all coming from One Source so that's a privacy problem the other big limitation of EOS is just in terms of efficiency many times what you actually want to do it spans more than one transaction so let's say you want to approve um approve and then you want to transfer those are two transactions but um really you you are thinking of them as one operation with the OAS you would have to send two separate transactions and um you know that that might not work out um it also might be a little bit more expensive because of that but there's no optimicity um um built in and if there is a revert on chain that's pretty expensive um so and that happens sometimes especially with things that are very time sensitive so what is the alternative what is account obstruction well basically what we're saying is you know eoas are um they're the one size fits all past and we want to move Beyond them to a world where um we can really manage the logic uh that controls our account and the logic can be pretty arbitrary and then that opens up a lot of possibilities and that is the future of how accounts are going to work so what are the use cases for this um you know the one that's been talked about since the beginning of social recovery like let's say something bad happens um to that secret that you're trying to both prevent from being lost and from being stolen well maybe you have some friends that you plugged into your account and three out of five of them can help you restore your wallet um and you can do even fancier things let's say your friends suck and they're trying to steal your money well you could have a delay mechanism built in so if your friend suddenly try to recover your wallet when you don't need a recovery you as the account owner would see that in advance and the operation wouldn't happen immediately so you would have time let's say maybe a week to respond to that the other really interesting use case is a dead demand switch nobody lives forever unfortunately and then what happens if what happens if you know you you need to to pass on um your your crypto but you're not around to help do that so with the dead man switch you could have a mechanism like okay if you haven't accessed your account from your main key for the last year then another key that your family members have access to um it would become active so usually you're in control full control of your account but then a year later you haven't been active something wrong happened your account automatically passes to multi-set controlled by your family I just one thing I want to explain these are all features of that account abstraction can give but it it's not that everything is Comfortably baked in these are counter possession or accountability opens the the door for wallet creators to add all those features none of those features are something that will be on day one maybe a limited set these are options that are open to do more oh now we can both talk even if we can talk over each other if we need to okay that's a good point I was talking to you off about that when you're preparing for the talk um this is oh yeah maybe that would be nice so that's a really good point um this talk is intended to inspire hey like we can do these things now um but we have this um this infrastructure built in but um this all depends on the code that's running um your your your smart Wallet account and that's up to wall developers that's up to anyone that wants to get involved and most likely there are many use cases um that are interesting that will not be covered by this talk so use you know the sky's the limit in terms of imagination um but we would like to talk about some of the use cases that we've been thinking about so one of the cool things that happened uh during each Bogota is there was a project that implemented an idea that we've um I've been thinking would be pretty cool is um well there's there's this signing mechanism in most mobile phones the problem is the signing mechanism doesn't support ecdsa curves so it's it's using some other mechanism um but because account obstruction doesn't lock in ecdsa as the only way to sign what you can do is you can have pair device Keys um that essentially you know you get your phone and you link your phone to your account now your phone can authorize operations on your account um and and a team actually implemented that in the recent hackathon which we thought was pretty cool multi-sig is a very obvious use case so you know now um you'll be able to just um get that out of the box um another one is BLS aggregation which requires a little bit of explanation BLS is a mechanism to aggregate signatures um and that's cool because it reduces your gas costs and because we have the ability to implement any signature verification mechanism in the account then now we can have more efficient signature validation and save some costs the other big one is eventually we know that the signature mechanisms that we're using are going to be broken it's only a matter of time Quantum Computing is making uh significant progress it's no longer theoretical the qubits are getting up there eventually we will I'll have to stop using ecdsa but there are alternatives um and hopefully by the time this becomes you know a pressing problem we will just be able to use any Quantum resistant signature mechanism that we want that is maybe like gas efficient and secure um and you can you can swap that so you have a lot of flexibility once you abstract away how you do signatures and it's not hard hard-coded so um other interesting use cases are spending limits I mean you know if you if using your wallet for small day-to-day things maybe you're paying for coffee for a meal or for something you know um that is not usually life-changing then it makes sense for that to be something that you can do very easily um that shouldn't be too hard but then if you're making you know very big investment um then you want you you might want to add uh add like a more uh secure way of doing that so you go get your Ledger um really depending on on what is the spending limit that you're comfortable with for different levels of security uh and you can you would be able to implement that because again you can have arbitrary logic inside um inside a contract that is controlling your account um the other cool thing that you can do is well you can have you can have uh different roles and you can delegate specific actions uh to those roles so you know if you're if the account is a corporate account you can give payroll uh authorization um to people who are in charge of that and you can specify the spending limits um another example would be if you're you know if you're um you have a legal department or you just want to delegate voting power for certain tokens you could do that in a way that still limits your exposure just to that um so they would be able to vote with the tokens and perform governance actions but not you know like transfer the tokens for example um and uh usually in companies I mean this is something that's very common in corporate bank accounts you give you give different uh roles in the company different powers so you know the CFO might be able to transfer a larger sum but there's still a 24-hour delay um and then you know maybe other c-level Executives could could veto that if maybe the you know the cfo's key was hacked uh or their computer um was hacked or the CFO is untrustworthy for some reason but um another example would be well you you want you want to give an auditor um the power to monitor payments and then the auditor has responsibility if if something you know they have the responsibility to reach out to authorize parties and then verify that what they're seeing happening on chain is what is supposed to happen so this way you can tap like maybe like a third party accounting firm maybe the one you already work with say hey like you know we we owe it to our shareholders at our treasury management um it doesn't get compromised and how about like we plug you into our our treasury in a way where yeah you can't you can't perform you don't have to rely on maybe your computer security or your staff we don't have we want to you don't want exposure from to them in terms of being able to um to transfer funds but we do want to give you the responsibility of whatever you see happening you know talk to the people who are in charge make sure that it is an authorized action and you know within this delay limit uh if it isn't just veto it just like cancel the transaction so you could give them the power to cancel but not initiate transactions and I think the one that's going to be like very useful very soon is session keys so in fact I think this is something Argent recently implemented for um for their uh Stark Stark net wallet so the idea is well you want to play some blockchain game and you don't want to have to to click approve every time like a message needs to be signed that that would really you know destruct from the gaming experience but you also don't want to give this this game full power over your wallet that would be crazy like you know what if there's a security bug in the game so ideally you want to create a session key and the session key has the ability to do whatever is needed but limited to the gaming contract nothing else and then the worst thing that can happen is I don't know maybe you're like your your game sword your nft items are stolen if the game gets hacked but everything else is safe and you could you could just store this this key inside your browser so as soon as you you authorize this as soon as you generated the session key um from that point onwards you're not even uh you don't have to be in the loop in terms of authorizing transactions and just becomes transparent so that's a really nice that's a really nice one um well the session Keys also you could you can mix you can mix what we were talking about before so you could have a session key not only has maybe limited access um to uh to a specific contract but you know maybe you have multiple devices and you want very easy things to be uh like very not not secure thing things that are not very risky um to be very easy from your computer so um maybe if you just want to sign into an event or or um you know something that doesn't have world changing implications if your computer gets hacked so you authorize your browser um and you you would still you you know you would still um have cap control um over your account in case something goes wrong browsers are not the most secure platforms um but you could just limit how much power the browser has uh over your assets so um there's also a big ux issue with gas where you know the traditional web tool world where most of our users will be coming from a gas is not really a concept so if you have a server you're not really thinking about who's paying to run the server and uh how do I share in the costs it just somehow works behind the scenes and users who are new to the space will not be really familiar um with the concept of gas and it's also a hassle even for people who are familiar so with with uh abstract accounts um you can you can have a lot more flexibility in how you handle this so for example if you're a game especially if you're running on a cheap uh cheaper layer 2 you could decide hey it's worth worth just subsidizing the transactions as part of my onboarding process um it's very common to to have to like um budget some amount in order to acquire users so if it's not prohibitedly expensive maybe you want to do that and then you know later um that's worth your while um another very obvious one is just being able to pay gas in any erc20 token anything that has value so if you're if I pay you in usdc there's no reason that I would also need to send eth into your account um for you to move that usdc and maybe pay your bills with it uh it's it's nice if you have the you know if you have an account you whatever whatever you have a value there um like when you know someone sends you a PayPal um PayPal transfer they don't have to send you like some other PayPal tokens so you can move you know your US dollars in PayPal to somewhere else um so payment becomes a lot easier um but also you know maybe you're participating in in some governance and uh the the governance of that project doesn't want you to have to think whether it's worth your while to pay the gas for the interaction or not um so they just decide to subsidize that um or you could maybe you know you have some allotment and if you go beyond the allotment you have to pay in the in that uh project tokens the nice thing about being able to interact with uh with a blockchain without needing eth is it also has some privacy bonuses so because if you go through the usual kyc process someone knows who you are and their companies that are dedicated to linking all the information uh together um so you don't have privacy but um yeah that would that would be that wouldn't be an issue if you can just um receive payment in any token and also pay for the gas in any token because let's say I'm a contractor and I work online for you and I'm pseudo Anonymous I do the work I give you an address you pay me and you know I would be able to pay my bills with that or or or buy something online or whatever and I wouldn't need to also acquire eth um and I could also separate let's say I'm working with with different customers I could give each of them a separate address and the customers wouldn't necessarily need to like they don't need to peek into my bank account or my equivalent of bank account and see like how much I'm making and what is the balance there that's like we're we're doing that um and maybe some of us have gone used to it but it's really weird where you're working for someone and you you give them the the like the number of your bank account and they can see exactly what's going on there um that's that's not very private um and and once we have once we have gas abstraction as a built-in feature then we don't have really have to worry about that anymore because we could just generate easily generate new addresses for for uh everyone that we interact with if we wanted to there's also an interesting um interesting uh use case for enabling cross chain operations there are many ways to do that um and uh yeah that's just something that becomes a lot easier once once you once you have um this functionality built into the protocol um there's also there's also uh some savings that you get from from just being able to batch uh different transactions together um also from being able to guarantee that the transactions are going to execute uh with optimicity so for some things it's useful if everything happens together and if everything doesn't happen together then you don't want it to happen um because there you know it wouldn't make sense uh for the transactions to execute separately um and you know whether this happens um whether this happens like in a gaming environment or you know there are some you know like Financial scenarios I mean imagine the simplest one would be you have to you have to approve you have to give some authorization to a contract to perform an action on your behalf um and then then you want to perform you want to give them like you know another transaction to actually perform the transaction and those both need to execute together otherwise like you just wasted gas um and there is there there's um there's a sort of General use case that's interesting where um you can actually uh Implement these um uh time time delay uh flow so for example let's say I want I want to be um I want to be uh uh selling my eth when it hits uh 5000 but I don't wanna I don't know when that's going to happen I don't want to sit in front of my computer um so it would be possible for me to just pre-create the transaction but make it conditional on certain certain things happening like okay only if the price of eighth is 5000 or only after this time or whatever the condition is and um so my wallet would agree to execute and pay for the gas for that transaction and also maybe a transaction fee and you would put that into a registry and or the registry of uh future time delayed or event driven transactions and Searchers they would be able to monitor this registry and see oh like these are transactions that their conditions have just been been met so they can compete on executing it for you um and that opens up it opens up a whole range of of uh interesting um use cases because now you know it doesn't have to be you that pulls the trigger at the exact moment the conditions are met and if the conditions are met um the transaction will be executed for you by Searchers and it could be time delayed it could be based on essentially like whatever uh whatever the conditions make sense um maybe another example would be you know there's an nft series and you have to be or I don't like you um an event that you have to subscribe to for in a certain time window um okay so a little bit about erc4337 which is so um this is the standard that we have been um working on and uh Paul here is is the guy who implemented the the contracts um so yeah please if there are any technical mistakes correct me okay so this is the first step um towards protocol level account of distraction the nice thing about our approach is that it doesn't require any change to the rules of consensus so we can kind of experiment uh for free and we don't have to solve governance in advance um the way we're doing it is okay we essentially create a mempool new type of mempool for anyone that wants to participate in this and uh a single you don't you don't need more than a single Network this this mempool it's um uh essentially it uh accepts it accepts uh something that is essentially a transaction we're calling it user operation but a user operation is equivalent to a transaction but it's a it's a transaction that works with with um with these account contracts um so what that does is it makes a contract wallets a first-class Citizen and it totally does away with the need for having um anyway you don't need a new way to uh to control um to control this this account um and I mean the way the way that works about maybe a little bit about how that works um you have it's kind of similar to how um uh you know flashbots Mev private mempools like the principal behind it is you can have a mempool where bundlers they are provided an incentive to submit your transaction um and essentially uh we're just we took that that idea or originally this was vitalik's idea and we added the gas abstraction part to it to make it a more general purpose and then the bundlers you know they their their job is to just uh take these user operations and eventually they they they bundle them together and they submit them um as when they're creating when they're creating um um uh bundles that that are go to the uh to actual uh blocks okay the the um the the other advantage of doing things this way is that we're separating validation from execution so you can have you can have [Music] um so so if I'm if I'm a bundler and I am paying for the gas of your transaction there's there's some risks involved for me because um you know what happens if the the you know like you know if you don't pay me what happens if I execute a transaction on chain it turns out that it um and it ends up uh um you know um ends up uh well at the very least you expect to be repaid in gas because otherwise um you know why why would you participate in the scheme um so to make it make it very safe for bundlers to participate what we've done is we've provided um a contract level guarantee that you're always going to be paid back regardless of of what happens with your transaction when it's executed on chain so we've separated validation from execution um and what the bundle needs to do when they accept the transaction they're just verifying that they're doing this off chain initially just okay if I if I accept your transaction um and I'm calling this huge function and uh am I going to be uh paid back for the guess that's that's all they're verifying so it's it's pretty cheap for them to do that and um later the totally separately when the transaction is submitted and gets executed but by then you don't really care um as a bundler um even if it reverts it's your problem just like with a normal transaction because you're still going to get paid and the the you know without this it wouldn't really be possible to um um the crate a permissionless uh pool of bundlers that are participating in in this um protocol because the bundlers would have to trust that um they're not going to get cheated just one we use the term brundler which I node Valley debtors just like any node Valley that also supports account obstruction and not one of them is enough to run a network the more they are the notebook is more resilient to a censorship and other things but the bundler eventually and eventually all validators will hopefully will be also bundles right so the right now we have so exactly you don't need a consensus change you don't need like 51 percent um of validators participating in this scheme because ultimately you're generating just uh um Regular legal bucks um and we have we have another mind um as as an implementation that is supporting this um the more the more clients support this the faster your transactions are going to get executed but um yeah Okay so so with erc4337 then you know once we have this scheme um we can also use it to make uh uh Roll-Ups cheaper because you can batch transaction you can aggregate signatures so um so yeah that's that's another advantage and like I said it doesn't require any protocol changes so on any ebm chain um you can we can start experimenting with this is there anything you want to add yeah okay technically a bundler can run as a separate entity it is much much better for you to be a node to be more resilient so the way to add it right now the way we currently adding it ungirly because it's still on test is elegant as a separate server a the wallets don't care it's implementation in order to be highly scalable and to be able to batch more it has to be a node in the network Never mind already have a node that can run a nethermind and there's a work to edit also with a get code so what's what's next well yes we can start experimenting and have um account abstraction and any evm compatible chain without consensus changes but the goal is to do away with the OAS eventually we don't we don't need doas and uh we know we're going to have to move away from them um eventually at some point and there are various ways of of thinking about this um so we want to want to want to have a count of abstraction um as a basic feature of the protocol but we want to do it in a way that doesn't enshrine any particular wallet or gives an unfair advantage to any particular wallet um and because the eoas are effect and they're very very common and they probably will be um until we uh you know we move away from them um it will take a few years we need a way to convert EOS seamlessly um to smart smart contracts so there will be there will be some default implementation where yes everything in the future is an abstract account including EOS but if you haven't upgraded your UA if you have an inserted code into your eoa if you haven't like activated it in some way then it just you know behind the scenes um continues behaving like in your way but it has it has the um the the functionality requiring allowing you to to upgrade it of course this would require a consensus change and there there's various ways this it can be achieved we're um we're discussing um you know one way would be there's if we create a new transaction type and then you can set the code for your eoa this is this is um this is now the code that's running your UA um or you know there's also I've been been a suggestion um EIP 3074 maybe that in combination with another EAP we could also set uh default proxy contract for all addresses um I thought you want to add anything Yeah by default that basically two options one of them to let a user decide the exact point of time where he wants to upgrade its eoa into a a contract wallet either using the transaction type or new OP codes on such the other way is to decide that at one point of time all Eos start using some default implementation we've deployed and tested thoroughly before which behaves exactly like anyway so all users will not notice a difference except that from now they have a way to modify the replace the actual implementation so they have a smart contract that just behaves exactly like an eoa until they decide otherwise right yes the basic Contour doesn't offer any of the Advanced features we described earlier except the one feature which is replace implementation the user can replace implementation once it is replaced the sky is the limited yeah all the use cases that we've yeah use cases in it all the uses that other people really try to find okay okay I think that is if anyone has questioned well we still now we're going to talk about well how do you join this okay yeah yeah okay [Music] um so how do you join this account abstraction Revolution you can start experimenting with the rc4337 right away and uh we had eight wonderful submissions um and uh this is something that's um you know already working so you don't have to wait um you can add useful features like the one we discussed you know batching or key recovery um or any any of the things that that we've been talking about you could build features that were totally not possible with the uas that we haven't thought about and if you do if you if you're building anything cool um then you should definitely apply for an EF Grant because we want to see this we want to see this [Music] used and adopted and we want to see the experimentation and want to update this presentation with more interesting use cases um so definitely apply for an EF Grant if you have a cool idea that builds on ERC on the CRC the other thing is if you are building adapt you have to think about a future where contract wallets are first class citizens I mean contract wallets are already pretty common especially for teams um you know multi-cigs are an example but still many dapps assume that they're going to be interacting with an eoa and that that is just um an obstacle for us to move forward it means your adapt already can't interact with things like you know it's a safe wallet um if you're assuming you're making assumptions such as um you know the the the like you know the how how signatures are validated so there are easy ways to make your dap compatible both with smart contracts right away and account abstract accounts in the future and that's ERC 1271 with just it checks um it checks if the caller has code and then there's a mechanism where it can just invoke uh invoke a function and instead of assuming that there's they can rely on this EC this DC DC DSi key the other one is if you can if you can um benefit from batching in your user interface and many many dapps especially games can then you should check if you're connected to a contract wallet that supports it um and that will that will create a better experience for your users it will save uh uh gas costs the other thing is um with how gas is paid so if you have adapt you should think about different types of uh Gas payment models I mean you know a big an easy example is if you have a token then it makes sense perhaps that your users should be able to pay uh for the transactions in your token when you're using your dap um and if you don't have a token or you want to subsidize your users that's easily accomplished um with with the with account abstraction um you you set up yeah you know you set up a a contract that authorizes to reimburse your users for whatever criteria you feel comfortable with maybe the onboarding process maybe they they have to perform some action but um it's uh it's it's something that's possible now and a lot of the a lot of the improvements that we're gonna get for for that usability is going to also require some you know wallet support so wallets are an important part um of uh of usability for depths and as a depth developer you have some influence by collaborating with wallet Dev saying okay this is this is something that you know would would be beneficial for my use case uh um and you can you can have a have a bit of an influence by just saying okay this is this is useful for me I need this feature in the wallet um for example supporting account abstraction so other than talking with us um we're happy to help anyone that's implementing uh different use cases we do have an SDK up uh later you have will share on his Twitter um the the links but there's there's a there's an SDK there's an SDK you can come up to us and we'll give you yeah or we'll just give you but I can I think I think I just like open this right now and show like where this is pointing to wait this is very small why is it so small yeah yeah so we have our SDK it's on GitHub eth infinitism account obstruction it's not dedicated contracts oh where is the SDK is infinity okay oh this is the link is broken then all right we'll fix that that is um okay oh this this is the SDK okay so the reference implementation of a bundler simple bundler and the SDK oh cool okay so that's the SDK we will fix the link later um you can also read up on the ERC uh and oh maybe that's the right link no that's that's really uh yeah that's your seat itself yeah yeah it's the ERC itself so you can read the ERC it's very detailed uh but you can get very precise understanding of how it works there's also a discussion on the ethereum magician's uh forum and and of course um it's nice to be able to talk with people so we have a Discord now we have a Discord server and uh you're very welcome to join and ask questions and even after this event like if there's something that you don't get to ask us in person um and yeah maybe now we will just take some questions yeah yes over there okay the internal development roadmap what we've developed are the interfaces and the core contract that performs this magic we call it entry point it was audited but then was extensively modified to support the l2s that is not yet audited we still have some work on it the API probably won't change so a wallet will be able to work so it's not deployment deploy the mainnet only on testnet currently but you can create and start experimenting with Wallets on top of that the interface of a wallet the change of wallet needs to be it's quite minimal I can go through it adding a single method or two uh yeah we have a sample uh yeah a sample that uses that adds account attraction support to ignore the safe you add the module and you basically get make a okay a single owner and you can also safe and account obstruct the compatible um so yes you can create and there are some work on creating wallets today in terms of applications yes it's chicken and egg application it's a wallet in order to work there is a way no it's for a basically a hackathon an application can work without a wallet but it's not something else you want your user want to sign blindly a hash if you're good with that it's also possible to with an application can work with accountability in the day the way gas abstraction works is that when you submit a user up I said the contract itself validates itself the signature and its nonsense and in order to accept the request but there's also a pointer to what we call a paymaster a payment is a contract that before submitting the transaction has a chance to decide whether it agrees to pay or not if it says okay that is it doesn't revert it will use its own balance its own stake and everything to pay for this transaction now what this payment does on chain depends on the paymaster so the most obvious example of a paymaster is that the validation will be I will check that this user has a balance of enough die and has approval for me to use this die and I will grab enough die to cover this transaction so and at the end I will refund it with the excess So eventually I pay the user pays with tokens for the transaction this models it's most obvious use case of a paymaster but there are other use cases if you want a voting data and you don't want the user to pay anything okay so the check is that the user is eligible to vote if a user is eligible to vote and didn't vote yet I agree to pay it's an another example of a paymaster other examples are still open you can write whatever you like just just a contrast right now if you're using Nosa safe then someone on your someone on your team has to like pay from their account right even if the agnostic safe has plenty of ease um someone still needs to like pay from their own accounts just to have that transaction uh finalized um so with account obstruction that wouldn't be necessary your account will be able to pay for yourself and this doesn't require a paymaster but let's say you're you're safe for um going back to the safe example if ignosis safe doesn't hold any heat it doesn't hold sufficient eth where you don't want to have to care about the balance and continually like exchanging and topping it off and whatever you have a balance of assuming it's enough to pay for the gas um you would essentially include in your transaction a reference to something we're calling a token paymaster and token paint Master can be completely autonomous contract on chain that it just um it accepts tokens it and it it would pay for the transaction in eth and then it would settle in some way so well an old reference like an older reference implementation recreative which is using swap using unit Swap and add single transaction which was kind of expensive either cheaper ways of doing it but just it's like very simple and one single Atomic transaction it gets paid it gets in a sort of an allowance in whatever token you have it pays for the gas it charges the transaction fee um and then it gives you back the the remaining uh tokens so with account abstraction if you want to use if you want to use that for example then you would just include oh the um The Entity that is paying for the gas is a token paymaster but there is a way for the token paymaster you know to to receive a commitment from your account to pay it back so it's not just subsidizing the transaction that's also possible it makes it possible right now you have to split different concerns if you need some corporate level security using a Galaxy safe and if you want a game you will use this metamask and fuse your private you can use trezor now if you if you want them all in the same single address yes you will be able to do it with a counter fraction probably you still might have multiple abstracted accounts for different purposes but for different reasons if if the reason to have multiple addresses is only security then yes you will be able to use a abstracted account to find something that can cover all the all bases yeah because you can just limit your risk exposure to each device depending on how much you trust it again I'm not saying that there will be one wallet it will give you all these use cases you'll find a wallet that give you all the user's case you need and use that and you always can switch I think the the first use case of a signature is chain signature just think of it you start using metamask and after a few years you collected a lot of nfts and a lot of money but you can't change the security model your browser holds your private key so you have no idea if anyone hacked into your computer and grabbed it through a copy of your computer without changing the address you can change the security with account obstruction with single operation of change owner now the trays are on the same account even if with the basic simple account I just changed the owner and now I am really secured because the previous private key is no longer relevant yeah even before you get into the really fancy stuff that you can do with the abstract account abstraction the basics are actually pretty um pretty useful just by themselves because right now if anything goes wrong then it's really hard to transfer everything from from one eoa to another I mean you would have to create separate transactions for each asset that you hold that could be pretty expensive um and if your computer got compromised you might need to do that in a huge hurry you know so it's it's um it's just not not the best situation to be in even very simple improvements like this will make a big difference