========================================================================
ENSv2: Identity for Apps, Agents & Beyond
Speaker: Kevin Werkemann — ENS Labs
ETHGlobal Tokyo 2026 · Workshop · Friday, September 25, 2026
Video: https://www.youtube.com/watch?v=CtCoMHf9R_w
Transcribed with aloha by the AI ʻohana 🌺
========================================================================

[00:00] So, my name is Kevin Werkemann with ENS Labs, and over the past year we've worked on ENS
[00:18] V2, which will be like the new version of ENS to be released very soon, but you can
[00:25] already test it on Sepolia right now.
[00:28] And today I will give you like a brief introduction of ENS itself, if you haven't heard about
[00:32] ENS, and also go a little bit more into the details of ENS V2 and what kind of tracks
[00:37] we have at the second one.
[00:40] So maybe just as a start, what is ENS?
[00:46] That's kind of the definition that I'm kind of like, that I've memorized, like the efume
[00:51] name service, the centralized permissionless name protocol that lives on Ethereum, right?
[00:56] It's like the, if you want to have an analogy, it's similar to DNS for V2.
[01:03] And the original idea is quite easy, actually.
[01:07] It was about turning complex HEX addresses into like human readable names, right?
[01:13] Again, like similar to DNS on the web, where you turn IP addresses into domain names.
[01:21] So I'd like to think ENS is a user experience protocol.
[01:26] It's kind of, for me, like infrastructure on Ethereum, and in some of its best use cases,
[01:32] I think you don't really like realize it's there, it's just like working.
[01:39] Like in this example, like just to kind of make the point of why it's useful to have,
[01:44] right?
[01:45] If you want to send five bucks to me, then you could use this address, or you could
[01:51] just send it to my ENS name, which I think is much easier to memorize and like a very
[01:55] helpful thing to have.
[01:59] But that's just how it started.
[02:00] At this point, ENS has become much, much more, it has grown over the years.
[02:04] And ENS essentially also allows you to have like an on-chain user profile for yourself.
[02:10] This is a photo of my profile on the ENS Manager app.
[02:14] As you can see, there's my X account is connected, my GitHub handle is there as well.
[02:19] You can have multi-chain addresses.
[02:22] So for different chains, not only EVM, also like, for example, Bitcoin, as in this case,
[02:28] you can store your address there and the wallet or some other kind of application can
[02:32] just fetch this information from ENS on your mainnet.
[02:37] And then like a wallet can use it to, I don't know, send Bitcoin to this address, right?
[02:43] So this is one of the things that you can do.
[02:45] We will get into some other things a little bit later, but just to give you an idea of
[02:49] like what it has grown to.
[02:53] Maybe like a few examples of what a common integration of ENS looks like.
[02:58] If you connect to some D app, the typical kind of like default thing that you would see is
[03:02] something you connect your wallet and then if you're connected, you can see like maybe
[03:07] a small avatar thingy with your address, right?
[03:11] With an ENS integration, this could look something like this, right?
[03:14] You take the avatar that is stored on chain or like referenced on chain in your ENS name
[03:19] and your ENS name to present like the user, like to provide like a more personable experience
[03:25] to the user in your app.
[03:28] Two examples for apps where this is already live.
[03:33] Snapshot.org, Marcher X, Y, Z, but there are like a lot of other apps that are doing this.
[03:39] For example, here, this is etherscan, you can see kind of the difference of having
[03:45] like named, like having ENS names being used or like just having addresses.
[03:53] In one case, you can see I kind of can understand, okay, like it's coming from here, it's going
[03:57] there and so on and so forth.
[03:59] For addresses, you don't really understand that, right?
[04:05] So in general, the idea is that users of crypto part experiences should never have to
[04:09] interact with a hexadis, right?
[04:10] And like going back to the ENS user experience protocol at heart.
[04:16] And yes, anywhere you read or write a hexadis, you should be able to use an ENS name.
[04:22] So how does it work?
[04:23] I've already kind of teased on like technical side, there are two processes.
[04:30] One is called forward resolution, so that's kind of the established name in the ENS world.
[04:34] You start from a name like nick.if and you resolve this into an address.
[04:39] It can be an Ethereum address, it can be base address, optimism, it can be Bitcoin.
[04:44] In this example, unfortunately, the addresses are the same, but they can be different, right?
[04:49] You can have a totally different base address, for example.
[04:53] So it actually, when the address resolution happens, the forward resolution, you would
[04:58] provide like the information, like let's say the chain ID, it's not the chain ID,
[05:04] but essentially the chain ID of like what kind of address do you want to get back
[05:08] when you do the resolution.
[05:11] Then there's also reverse resolution, which is just a reverse direction, going from an address back to a name.
[05:18] And if you combine both, this is what we call a primary name, so it's like a bidirectional thing.
[05:24] And it's essentially also what is being used in the examples I showed you before,
[05:30] where after you connected your wallet, there was your username and avatar,
[05:35] because the wallet obviously is going back from the address to the name, right?
[05:41] So, yeah, let me talk about DNS V2 a little bit.
[05:44] DNS V1 has been like quite popular and just, I guess, quite established in the ecosystem.
[05:50] V2 is a whole redesign of DNS V1.
[05:54] And it became kind of necessary because DNS was very early, I think,
[06:00] like one of the big projects in the kind of childhood of Ethereum, I'd say.
[06:07] And over time, the applications, like the things that people have used DNS for,
[06:11] wanted to use it for, kind of changed.
[06:13] And DNS had to grow with this, but of course, at some point,
[06:17] because it was not designed with these kind of applications in mind,
[06:20] like there are some shortcomings, right?
[06:22] And DNS V2 kind of takes all of these experiences and puts it into like totally new structure.
[06:29] I guess one of the most obvious differences is that in DNS V1,
[06:34] you essentially had like one flat registry structure,
[06:39] which means even for like subnames, you can see here on the left side,
[06:45] like there's Alice.Eve, which is maybe like a name that you would buy.
[06:49] If you have a subname for that one, like sub.Alice.Eve,
[06:52] all of these things were living in the registry, like one smart contract on Ethereum.
[06:58] And actually, the actual like parent-child relationship was more or less encoded in the name hash,
[07:03] which is like a certain algorithm that is being used to hash together like these kind of names.
[07:10] But in DNS V2, this is something that is made explicit on chain.
[07:14] So on every level of the hierarchy, you actually have smart contracts that are registries in their own right.
[07:21] And I guess that's something to get used to.
[07:25] Also, these registries technically don't know, per se, like where they are in the hierarchy,
[07:33] because they are never storing the full name, as it was the case in DNS V1.
[07:37] They are just storing labels.
[07:38] That's why you can see like here, these small labels, Alice, Bob, whatever.
[07:43] So the registry itself is kind of mounted in a certain position in the V2 hierarchy.
[07:49] But you can also kind of unmount them and mount them somewhere else.
[07:54] And then all of a sudden, like a different name would kind of start to exist, right?
[07:58] So it gives you much more flexibility in what you can do,
[08:02] because you can take an existing subtree of the whole hierarchy and just cut it off at some point
[08:07] and mount it somewhere else without anything being lost in terms of data or infrastructure.
[08:14] Then the general setup that you would do when you use DNS V2 is that essentially,
[08:26] there are two kind of smart contracts that you would use.
[08:29] One is a resolver, and the other one is a sub-name registry or user registry.
[08:35] So the resolver is something that you would, it's like your own contract.
[08:39] You would deploy this on your own.
[08:41] In the NSP1, it was the case that by default, you are using what's called the public resolver.
[08:46] The resolver is the contract that actually stores the data of your DNS name, right?
[08:51] So when I showed you the profile before, there were like these,
[08:55] my accent, like you can put a website, your avatar, and so on and so forth.
[08:58] These are all data that are stored in a resolver contract.
[09:03] In the NSP1, by default, this was like one contract, a public resolver,
[09:08] and there was always the option to kind of use your own custom resolver,
[09:12] but that was not the default, right?
[09:17] In the NSP2, it's now the default that you will deploy your own resolver instance.
[09:22] So it's like deployed through a factory proxy.
[09:27] And so you can very cheaply deploy this resolver, and it stores all of your data.
[09:32] And this enables like a lot of different possibilities in terms of configurability in the NSP2.
[09:40] Similarly, if you want to tokenize subnames of your name,
[09:46] then you can deploy a sub registry.
[09:49] And this is the kind of situation where actually you would kind of extend the NSP2 hierarchy
[09:57] that you saw over here.
[09:59] For example, here's Alice's registry.
[10:01] So if you want, you can deploy such a registry contract to tokenize subnames.
[10:09] Yeah, to tokenize subnames.
[10:11] It's actually not necessary.
[10:12] You can also just use your resolver to create subnames,
[10:16] because once you write records into your resolver, they start resolving.
[10:20] So if Alice decides to kind of, she wants to use sub.Alice.EVE,
[10:28] then she would just tell the resolver,
[10:30] hey, set like summary theorem address on this subname.
[10:34] And it would resolve in the NSP2.
[10:36] So it would kind of like, it comes into existence.
[10:40] The only thing is that it's not tokenized yet.
[10:42] So kind of the decision in the NSP2 is for you, do you want to tokenize your subnames?
[10:48] If you just want to use subnames, for example,
[10:50] to get paid on a different chain or whatever,
[10:53] and you want to have like a short handle for that,
[10:55] like, for example, base.pay.Alice.EVE or something,
[11:00] and you just store your base address over there
[11:03] to make it easier for people to discover,
[11:05] you wouldn't necessarily need to tokenize something like that.
[11:08] You can, of course, you can deploy a user registry.
[11:12] And in this case, you also get a token.
[11:16] But generally, I like to think about like registries
[11:21] as something that only makes sense
[11:23] once you want to hand out subnames to other people.
[11:26] And specifically, NSP2 has a new system in place,
[11:34] like a permission system, like we call them worlds.
[11:38] You can define certain worlds on names,
[11:42] on the resolver level and also on the registry level.
[11:46] And these worlds permit the user
[11:48] to be able to do certain things.
[11:51] For example, there's a role for a user
[11:53] to allow them to renew the name or to set the resolver,
[11:56] so to point to a different smart contract
[11:59] or to set a subregistry on their own
[12:02] so they can also issue subnames, for example.
[12:05] All of these things, when you have a setup like this,
[12:09] for example, where you put like a registrar contract
[12:12] on top of your subname registry,
[12:14] you could automate this process.
[12:16] You can say, for example,
[12:18] I don't want people that register subnames
[12:22] under my .if name,
[12:25] I don't want them to be able to transfer them
[12:26] so they're kind of like so bound.
[12:28] That's something that you can do.
[12:30] Or you can say they cannot set their own resolver
[12:33] or they cannot set their own subregistry,
[12:34] all of these kind of things.
[12:38] When you design this kind of system,
[12:40] you can think about what kind of permissions
[12:43] make sense for you, so you can furnish
[12:44] the whole subname space to whatever
[12:47] you want it to be, essentially.
[12:49] But you can also be extremely permissible
[12:50] and say they have essentially the same permission
[12:53] that you would get if you buy a .if name.
[12:58] You can do it whatever you want.
[12:59] And maybe you even want to, like for the parent name,
[13:01] you can register it for, I don't know,
[13:03] like 100 years or something
[13:05] so that it essentially never runs out.
[13:07] And you give the subnames away for free
[13:10] so they become kind of forever names.
[13:12] So there are like a lot of different things
[13:14] that you could do that I think are interesting to do.
[13:18] But generally, the thing that you have to think about
[13:24] is kind of do I want to tokenize subnames or not?
[13:28] Do I need to deploy a registry?
[13:30] Or if I only want to use subnames
[13:32] kind of as like data-only entities for myself,
[13:36] then you can do this with a resolver already.
[13:39] Another thing that is new in ENSP2,
[13:43] which I guess takes a little bit of time to get used to
[13:45] is that in ENSP2 tokens,
[13:49] like tokens, like meaning names in these registries,
[13:54] they have a mutable token ID.
[13:56] So as they are like ESC1155 NFTs, essentially,
[14:03] but with a token ID that is changing.
[14:05] And the reason for that is that this permission system
[14:08] I was talking about,
[14:11] you need to make sure that, for example,
[14:13] if a name is being sold,
[14:16] like you're not getting scammed essentially, right?
[14:19] So you want to make sure that, for example,
[14:21] on grant and revoke of these permissions,
[14:23] a new token ID is created.
[14:25] So what is happening behind the scenes
[14:26] is that the token is regenerated.
[14:28] The token is destroyed and minted to the same owner
[14:31] without them really noticing it,
[14:33] but the token ID changes.
[14:34] So if they were to kind of put it on OpenSea or whatever
[14:38] for sale, and then afterwards change the permission somehow
[14:43] to scam a buyer, then the token would be destroyed
[14:47] and minted again with a new token ID
[14:49] and the offer would go stale, right?
[14:53] So most of these things, like mutable token IDs,
[14:56] you usually don't touch this, I guess.
[15:00] But it's good to know about this right because you cannot assume that an ENS name has a fixed like a constant token
[15:06] token ID it will change upon like different
[15:11] Actions, but if you for example if you just like transfer the name
[15:15] From you to a friend or whatever then the token ID stays the same
[15:19] So it's it's really just a thing that's kind of related to expiry of a name and like this kind of world permission model
[15:27] So yeah, like why should you be why should you be excited about ENS V2 so ENS V2 I think mostly for me
[15:36] Brings like a lot of flexibility
[15:39] To the table as compared to ENS V1
[15:43] Like a lot of things that you had to kind of where to deploy custom contracts on ENS V1
[15:47] You just kind of works out of the box of ENS V2
[15:50] it sounds maybe a little bit technical right now because I'm trying to explain the technical side of it, but
[15:56] In reality actually if you kind of create your if you buy any like an ENS name and
[16:02] Like deploying the resolver and so on so forth
[16:04] There's a front end for this and you just have to push a button and everything happens on its own essentially
[16:08] Right, so there's not not a lot of things that you have to do unless you want to obviously so in the ENS V2 dogs
[16:14] there are
[16:15] Code examples for all of these things so how to deploy a resolver how to deploy user registry how to
[16:22] Delegate permission and so on and so forth there are code examples for all of these things
[16:26] So you had a permission model actually I think it's very interesting especially like in the context of this hackathon because we can do things like okay
[16:35] Yeah, I have I have a resolver and on this resolver there are I don't know 10 people and
[16:42] My product is I have a payment processing agent that
[16:46] Makes sure that for all the people on my resolver they each have their own Bitcoin address for example
[16:52] And whenever they get a payment to this address
[16:55] They might receive like a new address like kind of rotating the Bitcoin addresses, right?
[17:00] But automatically essentially so and you can give this agent permission to access really only this one record
[17:07] Only this one address record and he can do nothing else. He can just swap
[17:11] These addresses, right? So I also would like block a tick article recently on this kind of topic
[17:17] That's just an example. Obviously, but this permission system
[17:22] Excuse me. It's very strong in what you can do
[17:27] Like what you can for example like an agent allowed to do, right?
[17:31] And then also there's aliasing, which is also quite interesting
[17:36] Aliasing is essentially you can share records between names. So there might be like a default record
[17:43] That you
[17:44] Want all names to share once they are kind of like newly registered?
[17:49] And then maybe on later on they kind of can use their own records or whatever
[17:53] but in the beginning but by default they all share the same records or
[17:57] You have like some setup for your brand and the and you want certain names to also share records or whatever
[18:05] So like it's an interesting feature. I think as well
[18:10] Yeah, maybe we quick
[18:12] Have a little some time left here. So
[18:15] Agents as namespace is also interesting. Try to use sub names for your agents
[18:20] There are also a few
[18:24] Specifications already in place that links different standards in SAP 25 is linking 804 and DNS
[18:34] Maybe we quick
[18:35] So what can you build AI agents a lot of stuff with our agents use usernames text records very important also data records
[18:42] subnames places in general if you build a wallet and obviously like usernames avatar primary names are things that you can use or
[18:49] You build like a social kind of platform again, like you know, it's also like about identity, right?
[18:56] So usernames avatar and these kind of things you can use it over there and maybe like one underexplode areas content. There's also a
[19:03] Fiat caught the content hash that you can use like for decentralized websites
[19:07] so maybe
[19:10] Yeah, that's kind of like the credo or like anywhere your app shows a hexadell should instead showing us name
[19:16] Let me get to the prices we quick so we have two tracks one track is the best use of the NSV 2
[19:25] 1000 for the third price
[19:29] 2000 for second and
[19:30] 3004 first place and then for the continuity track also like 1k for first and second price and then 2k for the
[19:40] Finalist yeah, then this is my telegram and there's also an ENS developers telegram group
[19:49] Yeah, feel free feel free to just come to the booth or text me on telegram
[19:55] Yeah, should have any questions
[19:57] Happy to answer them. What is the continuity track?
[20:05] I mean you are aware of the general system or you're asking specifically about DNS
[20:12] Well, the idea is you so maybe to make it clear. It's good that you're asking actually
[20:18] In SV 2 is not live on Maynard right right now, right in SV 2 is on Sepuliar
[20:23] So for the continuity track you would extend an existing application or project to integrate in SV 2
[20:30] Which is a little bit hard because in SV 2 is not on Maynard right now
[20:34] but then again, it kind of depends on like what project you are working on and
[20:39] You might be able to integrate in SV 2 anyways to just get it kind of like to a state that when it hits
[20:45] Maynard you can just kind of
[20:47] Like flip the switch essentially right and integrate it and have it live
[20:51] With the continuity trick I suppose we are like a little bit more like lenient on what we allow for you to do
[20:59] All right, because the NSV 2 is only on Sepuliar right now
[21:02] Yeah
[21:04] Yep, what work remains?
[21:17] Where they have been like multiple stages of like auditing also, right and then there's also
[21:23] migration work necessary
[21:26] Or like like there are different stages to the migration and so yeah, that's kind of I guess doing the most
[21:33] The biggest part of the of the work, right?
[21:39] Anybody else? Okay. Thank you