Jito BAM and Solana Market Structure | Lucas Bruder

Transcript

What's up, guys? Just wanted to let you know about the Block Works Digital Asset Summit, which is back in New York March 24th through 26. We'll have top speakers from leading asset managers, financial institutions, DeFi protocols, crypto companies, and policy makers all under one roof. Think of Black Rockck, Coinbase, Robin Hood, and more. Follow the link in today's show notes to buy tickets, and make sure to use code 0x200 for $200 off at checkout. See you there. Nothing said on Zerox Research is a recommendation to buy or sell securities or tokens. This podcast is forformational purposes only and any views expressed by anyone on the show are solely our opinions, not financial advice. Pocachio Ryan and our guests may

hold positions in the company's funds or projects discussed. What's up guys? Welcome back to another episode of Xerox Research. Today we're joined by Lucas um Buffalo from Jira as well as Carlos and Sam from the Blockworks research team. Lucas, how you doing? Thank you for coming on. >> Pretty good. Thanks for having me. >> Perfect. Um I guess just to start us off, we recently released a report on Solana block building and obviously Jetto is very very involved. Um, I guess just so like people have a bit more context, do you want to chat a bit about how Jito's involved in block building specifically with the BAM? And then we

can get into more in depth. >> Yeah. So, yeah, backing up a few years, we created Jetto Salon and the block engine to try to make the chain more efficient. I think it was pretty successful in 2024 and 2025. it was pretty hard to land transactions and a lot of people relied on us to kind of filter the spam and land transactions and um a lot of that just was like performance bugs in Agave that they fixed. So the chain's way more efficient now and we were kind of thinking like what's next after this like um think we built some some valuable products here

and uh what can we do to make Salana a better place to trade and a better place for applications and what are some of the problems that we're running into today. So I would say like that's kind of when we started thinking about BAM and um essentially BAM tries to make the way Salana works super transparent and verifiable. So uh you know how the market works, you know how blocks are built and you can verify that they're being built that the way that they claim to be built. So we do that using trusted execution environments. So uh you can basically go there's u this BAM node

which is kind of like the block builder and it runs in this TE so it's all private all the transactions and contents inside of it are private and you can verify what's going on through that to stations. So uh yeah basically you could for validators that are running the BAM node it's not open source today but you know we're aiming for within the next few months here after audit and all that uh you can go to the BAM node say hey what version of software are you running go to GitHub see what what's running there and then you can kind of uh compare the two and see see what's running and then yeah you also um BAM has this concept of plugins so basically applications can kind of build some customiz izability or

features into the block builder and those get the same benefits of the privacy and the verifiability and all that. So yeah, we're working with a few applications on that and we should have some more info on some of the first stuff that we'll be releasing pretty soon here. >> Perfect. Um, one thing you mentioned is that I guess the initial goto agave client was to fix some of the issues that existed back then, but since then agave has improved enough that it's not really an issue. Um, would you say that the current market structure on Salon is still broken to an extent and that's what BAM is aiming to fix?

Uh, I wouldn't say it's broken. I just think that there's there's more that can be added and I think that the I think there's more that needs to be added than just IBRL. Uh, like there's a lot of applications that are requesting ACE. So being able to support that at the block builder layer is something that's really important. And uh I would just say that there's um there's this kind of trend where like things are becoming more black more of a black box like how does this validator order transactions and uh when that when the way the market's working is changing every 1.6 six seconds then uh we think there's

kind of some room to uh improve that and try to provide a little more consistency. >> Yeah, that makes a lot of sense. Um I guess from a user perspective um particularly looking at per on Salana, how um like what BAM is doing with ACE and those sort of things um benefit um those applications and users there. >> Yeah. So yeah, we've seen a lot of uh there's a lot of chains that are implementing ACE. Most uh the most popular version is Hyperlid. Actually, there was some discussion I think yesterday or the day before on I don't know if you guys saw this, the Poly Market news. They had a taker speed bump of 500 milliseconds that they just removed and a bunch of people were angry

and like we'll kind of see what happens there. But um yeah, I think people the perplexes and the prop ammms want Ace. Um I think ultimately like users will get better pricing because of this. So that's kind of the outcome that we're looking for is like better better user outcomes and more customizability for applications. Um I guess I have a question in that regard. Like it's great to enable application controlled execution but the road map to get there seems unclear to me as of today. Like do you think the timelines what do you think the timelines look like for really having a scale like on Salena? Uh

because to me it seems like right now ordering is more or less trying to be standardized maybe by priority fee per CU. uh a seems like the end uh term goal. >> Yeah. Um I think that BAM uh we are focusing a lot on um like kind of a version of ACE right now with the first plugin that we're working on and um I think BAM can provide ACE to Salana like today without any protocol changes and so today BAM's at like 25% of stake and I think it's it's a good time to start testing that and iterating and

talking to apps like, hey, what do you guys want? How can we make this better for you? And then, uh, implementing that. And I think in like the short to medium term, obviously, we're optimistic the BAM stake will grow and that will just make it the ACE that's running inside BAM happen on a larger percentage of blocks. But uh yeah, I think the the priority fees uh CU sorting is kind of nuanced here because um a lot of it depends on how the scheduler is actually built inside the validator. And uh I want I I'll try not to get in the weeds too much here, but um like previous versions of Agave, they

basically would process transactions as fast as possible. So when there's no load on the network and there's like not a ton of transactions, there's basically no priority fee by CU ordering, it's like, hey, whatever I have here, I'm just going to process it as fast as possible. In 3.1, there's some uh version of kind of pacing here, which uh would help where like it will um I guess rewinding to the previous version where it just like fills up as fast as possible. uh on our validators, we're seeing that we could build a block, the entire block, and fill it up in like 150 milliseconds. So on the other 250 milliseconds, it's kind of just sitting

idle. So with the pacing, it'll kind of uh linearly release the cost units throughout the slot. So halfway through the slot, you'll have 50% of the block full. And uh that kind of gives you some ACE primitives where like you get enough time to kind of collect these transactions and actually do the sorting where in the FIFO like Agave is essentially FIFO before. So it's just like there's not really any sorting there. Um BAM today we're running 50 millisecond batches and we want to get down to 20. Seems like a pretty agreed upon number by the the people that we've talked to market makers and um other people. Um and then yeah longterm I think the best

solution is probably MCP. We'll just see what the exact design is for that but that will uh change how the blocks built a lot in that the sense that uh you um all the the reordering like the priority feed of I see you ordering will be done in replay. So you get basically your batch size is like the slot size or your slot length. >> Perfect. I um Carlos, before I let you go, I just wanted to clarify for listeners that ACE is application controlled execution just because I feel like we chucked it around like 50 times in. And I'm pretty sure we were referring it to it that as a completely different acronym like three months ago.

So >> yeah. Yeah. And I guess like to add on that like the most popular version of Ace is just like a taker speed bump. So that's what some like Hyperliquid and Poly Market had until a few days ago. But um basically you give market makers a time advantage to update their orders and the um they can basically quote tighter without having to worry about toxic takers uh picking them off. Yeah, I I saw that poly market thing and I saw a lot of people crash out in their Discord because they didn't announce it any changes to their API which to me seems a bit to to go that um again you mentioned MCP. What does how does what

does BAM look like in a postmc world? >> Yeah. So yeah for the listeners MCP is multiple concurrent proposers and today when you are uh basically every well there might be a few networks that have like various forms of MCPU but there's not that many essentially what it is is today there's one block producer at a time on Salana you get four slots in a row every slot's 400 milliseconds so you have 1.6 six seconds where you are the only block producer. And so essentially have a monopoly on block space at that time. And so um you know there's the opportunity for the block producer to

censor transactions during that 1.6 seconds or reorder them however they see fit. Um and just kind of like play more games. In MCP you have multiple block producers that are basically each one of them produce a produces a block uh at the same time and essentially they get like stitched together. So you have like you know us four are block producers at this time. We all produce different versions of this block and then they get stitched together and the protocol would reorder by the priority fee. And so um yeah, this the goal of this is to like provide

more censorship resistance. So if if one validator is going to censor you, you can send it to another. And um there's some other benefits as well. Um like you can have mult you can have redundancy on block producers and things like that. So uh if a validator is offline, you still get a block made for that slot. So yeah, that's kind of how MCP works. Um, it'll definitely change the way that BAM works, but I think BAM will still have a lot of value in this world. BAM's kind of a few things. It's uh provides privacy. It provides verifiability, which I think market makers want. And um, there are still ways to influence ordering in MCP world. It will just look

a little different. I was responding to someone on Twitter about this, but um today the block producer doesn't really need to respect the priority fee cudering. Um you can kind of just stick stuff in wherever you want. And with MCP you uh you will have to the um the code that's like running the block builder if they want to stick things in or uh whatever they're doing they they will just need to be a little smarter about how they choose their priority fee. Like you can imagine if if Carlos makes a trade and pays like 100 priority fee and I want to stick my transaction behind him, I can do 99 and

uh it would end up behind when the the reordering happens. >> That makes a lot of sense. Like to me it seems like MCP brings a lot of consistency that is currently lacking uh in Solana's like block building landscape and it does that with an in protocol solution. That said like I think Alpenlow is one requirement, async execution is another requirement. Uh >> like and so like the timelines for MCP vary depending on who you ask but usually people say like 12 to 18 months.

Uh but Sena needs a solution today, right? Uh, and so I I guess my question is can like can Sana get better market structure without MCP that is competitive but with like app specific like chains like Hyperlid. >> Yeah, I think so. That's like the entire BAM thesis. So, um yeah, we think we can get there today and uh yeah, we'll have something that kind of like is like a demo application of this um um pretty soon here. But, um yeah, I think it'll be able to get there

and then we'll see on the like the timelines. I think you know 12 to 18 months seems pretty reasonable to me. But, uh we'll see what happens. And um on that like in the interim period before MCP do you and with BAM itself do you see there being like a good enough adoption threshold for BAM in like stake share per se where you think like market makers can largely start pricing in leader risk and apps can have essentially an optimal environment to build on like is it if we go from 25% to 75% is that pretty much there or do we need to go all the way to 95 100% like how do you view this? Yeah, >> I think 25 is like the bare minimum.

Obviously, the more you have, the better. Um, because you just know how the system works and that the apps will be treated favorably uh when there's more stake running this. Um, but yeah, I think we're like at the point now where it's a two it's a two-sided marketplace, right? like you want applications to build stuff in BAM and you want them to be able to kind of have a little more control over the way the the system works and you have the block space which is like the validators and so uh like if you have 5% of stake it's hard to get the apps on boarded and like bought into this and I think 25%'s like the sweet spot where we can start

to show value and then you know obviously we're optimistic the the BAM stake will continue to And so, uh, yeah, I think like 50 75 80 90% is pretty solid. And, uh, yeah, I think it's like back in the day, um, when we were first like bootstrapping Judo Salana and it was at a small percent of stake. Um, it was like it was just a little additional complexity for people that wanted to use bundles. It's like if leader running Jetto Salana I can submit a bundle else like submit a transaction and then now it's like you can basically submit a

bundle across like 96% of stake I think right now um so yeah there'll be like a little additional complexity on like if user is running BAM or if the validator is running BAM I can do this logic and if not I can do this logic and that's something that we'll uh try to simplify for people with some of the stuff that we're working on so that they don't really need to think about it. >> In the question of like network stake, uh DAM has surpassed 25% of stake, but also Harmonic has been growing quite fast and they are at like 16% of stake.

Like it seems unlikely to me that we get to a state like we got with the previous JTO agabic client where you had like 97% uh adoption at some point. Uh it seems like harmonic and bam will have to coexist like in the interim while MCP like arrives and probably even pose that. So how do you view like can these two systems essentially coexist and still push market structure forward? >> Uh yeah I think so. Um we are at Jetto we're very focused on like the privacy verifiability and consistency angle. Um but also we want people to experiment

with the system. So uh you know I think we will provide the the best uh kind of platform for this verifiability and like understanding how the market structure works and the goal is that it will be consistent across block space as opposed to like this uh algorithm that can consistently change blockto block and people maybe don't the the additional cognitive overhead of like trying to understand what algorithm is running when you have multiple builders competing is a lot. And so, uh, we're focused a lot on consistency, but at the same time, we want people to experiment.

And so, you know, we're very, um, we're very serious about open sourcing this code and having other people working on it and maybe trying different schedulers and whatnot, but, uh, we just want it to be open and verifiable. And I think that's where we, uh, that's where we spend a lot of time thinking on and like trying to optimize for. That makes a lot of sense. Um, and on that topic, so Harmonic has said that they essentially are selecting the highest feed block is like can be as a proxy for packing efficiency because their three internal builders are optimized for different network conditions and the auction has um strict rules in place. I guess do you buy that idea that highest fees can be a

good proxy for efficiency if there are certain rules in place or do you see that there are like unavoidable tensions where any type of feebased selection could drift towards um the wrong way and not towards execution quality? >> Yeah, there's definitely some tensions that exist. Um yeah, like uh the thing that makes the most fees is probably not the best for the network. Hopefully we can like figure that out somehow. Um but uh I mean we you can see that like the the validators are making the most money are just kind of delaying transactions to the very end or uh they have uh some some custom strategies running or like

running arbitrage bots on their validators and stuff like that. So, uh I think like the yeah um there's definitely a tension there and I think figuring out how to mitigate that will be super important because uh there's there's a lot of lessons to learn from other ecosystems and uh we can't make the same mistakes and I think that um when you have a lot of competition at this layer it's how do you avoid the uh like playing games to make higher rewards. So, validators choose you and yeah, you'll run into like order flow and orderflow games and integrate searcher builders and uh like

a lot of application rebates and um things like that. >> I I think that's a fair point. I um I feel like we this is not on this line, but I feel like we kind of touched on it. BAM at 25% is at a correct at a good state. Um are you guys thinking of implementing incentives or what's the game plan for growth on that side? >> Yeah, there are incentives today. So the the DAO is sending a lot of the protocol revenue to validators to uh that they can go and claim to run this and uh there's some like tapering off of it as the uh I think as the state grows and there's like time time based tapering

off and all that but it's just to like get the the system bootstrapped and I think it's worked pretty well. we've gotten good amount of stake from it. And so now it's like now we're we're obviously focused on stake growth, but now it's like proving value and how do we add value to the network today. So uh that's the thing that we're focused on a lot right now. And um yeah, I think we'll have some some cool stuff to share on that soon. I just want to ask about this like one idea of adding value versus I mean you already said there's some tensions in the harmonics and I don't want to push too much into this harmonic side because I I don't see the point of it but can you I guess like expand on this

thought process of you guys want to focus a bit more on like being good stewards of the network. Is this the idea of BAM? >> Yeah. Um yeah, we are not really um like not really into playing a lot of games here as far as like trying to really maximize rewards. There's a lot of things that you could do that would just not be super great for the network. Um and so yeah, we are our main customer right now. Obviously validators are important, but we've been talking to a lot of applications. uh we have you know I have at least three or four calls with

different applications every week u mainly focused on the trading side like hey what can we build on the network for you that would make your application 10x better than it is today and so that's the angle that we're approaching it from is like talking to a lot of applications figuring out what they need we have a few ideas on things that we could build internally and starting to build those out internally while the code is still closed source ource and then you know once it's open source then people can kind of build what they want inside there.

>> I'm curious like when you speak to these applications like what are their biggest pain points today is it like mainly on the block building side or do they also have issues with the fragmentation that's going on with the transaction landing services and the uh like fragmented fee stack between out of protocol tips and and priority fees and whatnot. Yeah, it depends on the customer. Um, a lot of the prop AMMs, it's like we're seeing since we released IBRL, that uh that website that shows like how people are packing blocks, we're seeing a lot more people looking into how blocks are packed. So, I think sandwich.me the

ghost logs team teased some data that they were looking into that I think they'll release pretty soon here. a lot of the prop amms are looking into data and finding some very interesting things that are going on. And so, uh, yeah, I would say it's like the transaction landing. Um, and there's just so many services that exist to try to do this. Um, how the actual like ordering is taking place in the block for like the protraders, like the prop ammms or pro traders and the arbitrageers and things like that. And um yeah, I would say that the like a lot of the AMMs and market makers want ACE. So that's some that's

kind of a common theme. I think the question is there's a bunch of different ways to build ACE. So what are the different ways that we can build it and how do we serve them? And then there's some other interesting use cases like what if um if my application was more tightly coupled with theuler and um like if I could run my application in theuler then what types of new things can I build or what types of new experiences. So that's not necessarily like a problem, but it's like a new design space like um yeah, I can't say too much, but uh there's yeah, there's some interesting things that you can do if you're like running um your application inside of the block builder.

I guess one that we like touched on is like this idea of like just in time oracle updates. So like uh put pulling in fresh Oracle updates before a user's transaction so they get the freshest price. So, uh, that's kind of like one example of something that you can do inside the the block builder that, um, just gives a better user experience. I have um I'm gonna take us back for one second to the BAM a little bit more just because just going off that idea of being a good source through the network. You guys said you don't want to do um well you shut down the private mele a while ago.

It's been two years now. Um and then you're kind of giving up >> um some jet like other jet revenue sources for BAM as well. Is that >> Y >> um is that how how was that decision made I guess a good question like as a because you guys run a company as well like you need to make revenue how yes approach the spa process. Yeah. Well, today all the revenue from all judo products except for like some validators that we run at labs goes to the the DAO. And so it wasn't just a decision from us. It was a governance decision and the people the JTO token holders and delegates

decided that that was the uh right move to make. But uh obviously like someone has to write the proposal and we we're trying to decide like a few things to do. But um I think we're kind of like re-entering this new growth mode. Like we went through this Jetto Salana growth. Literally launched it like right before Breakpoint at 2% of stake. Grew it to over 90 something% of stake. And we're kind of re-entering this growth mode with BAM. And uh there's a lot of applications that want BAM to succeed.

So uh it's kind of like, hey, we should really try to throw fuel in the fire here and get this to grow. And uh yeah, that was kind of like the the thought process is like we really need this thing to grow fast and um I think it's been helpful from like getting applications more serious but also uh the more block space you have the faster you can iterate on. So like collecting data we have a ton of data and ton of performance data and like ton of block data. So the more stake that BAM gets, the faster we can iterate. And uh yeah, I think that was kind of like the the

thought process behind that. >> I was just going to say no, that makes a lot of sense. Um like growing the pie per se. Um and I guess in this interim period, what applications or teams are you most excited to be um working with or you're most excited about in general um that are utilizing BAM? There's a few that we announced in our uh original blog post on BAM that we're working with, but uh you can assume that we're like talking to a lot of the prop amms and um some other application some perexes and things like that on Salana and uh yeah, right now we're like mainly focused on propms because that's where a lot of the activity is. Uh but yeah, we'll start to kind of slowly expand out

from that into perexes and other things. But perexes aren't the primary um focus like because to me >> not at this point >> looking out from a non >> I love Salana I love GTO but I would say I'm less Salana focused than let's say like a Carlos um >> so to me it always seems like from an outsiders's perspective it's like prop amms are great but Salana will probably have to focus a bit on perexes eventually and I assume that BAM would Um, and would be like playing a very key role and it is something in plan but currently like first iteration is props.

>> Yeah. Yeah. We have limited resources so we kind of went through like what's the state of the world today and uh yeah we decided it would be good to focus on prop amms and um we do talk to the Ellipsus team a lot and talk to them about Phoenix per that works and how we can support them. So, um they're still I think they're like kind of still in beta and like rolling out um whitelist codes and all that, but uh yeah, we we want to help them and we want to help Drift and um the other perplexes as well.

Um yeah, on that on that topic, I saw the Finenix team mention something on the timeline like a 30x reduction in the cus which would be uh quite significant like uh to enable much better like uh ordering uh within the block. Any thoughts on that? Yeah, that's how um I think they have like a prop amm inspired design. And so the beauty of the prop amm compared to traditional order books is that the updates are super cheap for market makers. So I think a lot of the prop amm today it's like 50 100 200x more efficient to update their uh the

price on the the uh AMM compared to uh someone taking and trying to swap on it. So uh I think it makes sense. Um, you do still want some type of like pacing or batching for that to take place cuz uh if you remember earlier I was talking about like if you have FIFO it doesn't really matter that much. I mean it it is the the cost savings is nice because they can just stream in updates super fast. But uh yeah, if you really want if you kind of want that those ACE like properties, you need some type of like pacing or like batching or something so that you can get enough time to collect everything and like let it sort out uh

and like all the the maker updates sort to the top and the taker stuff sorts to the bottom. I guess on that point like in terms of like if we zoom out to network upgrades like if we think about bandwidth like the block limit has been increased to 60 million CUS it's going to get increased to 100 million CUS but one thing that I think everyone is excited about is the eventual reduction in like slot times from like the 400 like millisecond threshold it is today like in your opinion what what needs to happen for us to get there. Uh because you know BAM runs 50 millisecond like auctions, harmonic just announced they are switching to that as well but it

seems like a reduction in slot times would really be beneficial for the network. >> Oh 100%. I think that uh a lot of us have disagreements on like what where we should go in Salana, but I think we can all agree there that lower slot times is better. It reduces the uh the proposer monopoly time and um it reduces some of the timing games that you can play. So like instead of late packing and packing all your stuff at the very end of the slot every 400 milliseconds, it would be like every 200 or 100. And so uh yeah, I think everyone's a fan of that. And uh my understanding of what needs to

happen for that is just performance improvements. So there's uh one performance improvement that will be in 3.1 uses uh this networking technology called XDP. Basically means you can send data way faster like 100 times faster or something. And so, uh, that will be enabled on validators and so they can propagate state out faster because they're they're spending a lot of time like distributing the blocks to different parties. So when the XCP is enabled, they'll be able to send it way faster.

Um, [clears throat] and then yeah, there's just a lot of performance. Every time you create a new block, there's some like fixed cost to that. like I have to, you know, do my my house cleaning to like produce the new block and all that. And so there's a lot of performance optimizations I need to happen around that and probably some other stuff like leader handoff and other things. Um, but yeah, I think like someone else like Brennan would be the right person to talk to about the specifics there.

>> No, that's true. It was also my understanding that uh because voting costs were relatively high that if you reduce like slot times like essentially if you have the slot time you're doubling the cost for validators and that will be really bad. >> Alang will fix that. >> Yeah. And alpo will fix that. Exactly. But yeah, it seems like Q3, Q4 this year, like a lot of the pieces could probably be in place for for a reduction in slot times. But yeah, this is a better question for the ANSA team, I guess.

>> Yeah, I do know that like if you follow Aleandro on Twitter, he's got some crazy profiler that he built that uh um a lot of times when you're like profiling high performance software, it actually slows the software down. So it slows it down to the point where like you can't use the software anymore. Uh so he has some crazy profiler that you probably see him sharing screenshots of and he's just like going through everything and like it's like oh we're spending 15% of our time here. What the heck is that? And you just like keep finding those things. You make it 5% something else bigger pops up. Shrink that down. Something bigger pops up.

Keep shrinking it down. So uh I keep bothering him to open source that tool. We'll see when he when he does it. >> Yeah, that that's a great point. I guess like on the matter of Agave, like it seems Fire Dancer was kind of like super hyped like two like one year ago and now that it's like launched it has launched like with just revalidators I think from the jump team. Um, but Agave has made like immense progress like in the past like year, year and a half. Do you do you think like Fire Dancer can still be able like to compete with a gap in terms of stake eventually?

>> Yeah, 100%. I think that um last I talked to Fire Dancer team like kind of around breakpoint time, it's pretty good. there's still some uh there's still some like consensus related mismatches going on in place. So, I think they want to keep the amount of stake small, but uh I know that they're working every day to to get those uh those mismatches figured out. And so, yeah, I think um it'll definitely be able to keep up it. I'm pretty sure it just runs all the execution on one thread and where agave

is like 8 24 or more. So it's it's like way faster but same time you have core team and ona working hard to to make it the ona client faster. That makes a lot of sense. Um and I guess on this like on these competitive dynamics I guess um do you think I guess in aggregate the internal competition um in block building has been a net positive for Salana so far or and maybe forward looking as well. Do you think it could potentially create too much fragmentation and coordination overhead um from I guess at different ideas and incentives in the long run to actually get get what is best for the network?

Yeah, I think it's too early to tell at this point, but there's a lot more people thinking about this, which is good. Uh, we're seeing more people looking at data on salon and block building. people are thinking hard about it and um yeah I think it's too early to tell but that's definitely something that we are trying to avoid with BAM and that's kind of why we built it the way that we did is to try to avoid this fragmentation and like these uh black boxes that are building blocks on the network. Um, so yeah, that's kind of what we're optimizing for. And um, I really hope that we don't go down

that that uh that route on Salana. I I guess a bit more specifically on that would is there any um realistic path or I guess some communication efforts at the moment uh to coordinate with um say like a competitive block builder like Harmonic in regards to trying to get a a few things on the right page um for the best I guess for the best perspective from the network. Uh yeah, there's some conversations going on, but um yeah, I think we talked to a lot of parties, uh core devs, market makers, applications, and um try to try to figure out what they want. So, um yeah, that's kind of uh that's kind of what we've been spending

most of our time on. >> Yeah, I I kind of want to move away from BAM a little bit. like it has like dominated the conversation like in the public for the past few months, but I think like still a relevant part of like jetto is like the goto soul part of the business and it seems like you guys had a lot of efforts like in terms of like ETF approvals to enable like gotool as an underlying asset for more operational efficient like uh staking rewards for ETFs like could you walk us through like what those efforts have been looking Yeah.

>> Yeah. Um, I'd say it probably goes back to around a year ago. We were the first team to go talk to the SEC crypto task force and we talked to them about liquid staking and ETPs. So, we kind of started thinking about it a little bit before then, but we saw that opportunity open up and it's like, hey, this administration wants to talk to builders, so let's go in. So Rebecca, myself, Greg, and Kyle at Multicoin went in and talked to them. And uh that was kind of like what really kicked it off.

So yeah, we spent a lot of last year like understanding what do you actually need to launch an ETP. And uh I don't spend most of my time thinking about this. It's mainly like Brian and Rebecca and others. But um uh yeah, you need like market indices, you need custodians, they need to like be able to mint and burn the asset, uh you need like security audits and all these things. And so we spent a lot of time last year doing that. And uh we announced the VANC jet soul ETP. Um there's still some work going on cuz uh on that because LST there's there's no LST ETP in the US that's live today. So

there's a lot of things to think about like what is the liquidity on dexes where sex is? Um like do you can you only rely on DEX liquidity or does there need to be sex liquidity? Like how do you build an index for this thing? um how do you do the mint burn all a ton of stuff like that and then um we're able to move a little faster in Europe. So there's the J Soul ETP which is 100% Geto soul that 21 shares launched a few weeks ago and then uh yeah there's a few others that we're working on in like UAE and uh Asia and things like that but um yeah the the benefit of the one in

Europe specifically is that they need to do a lot of um they need to be able to meet redemptions on stake like when you are redeeming coming out of the ETP. When you have stake soul, it's really hard to do that. So, you need to have some buffer. Like, I'm going to stake 90% of it. Assume I've run the numbers and I don't expect more than 10% of it to be withdrawn in a day. So, that's we're just going to stake 90% of it. If you want to redeem more than that, then you have to get like a line of credit and that has a lot of cost to it. So, it can actually eat into the the rewards of the ETP a lot. So Jeta soul you there secondary markets for it. You can get in

and out of it and so uh they are able to stake 100% of that and uh get slightly higher staking rewards than some of the other products and that just means better a better product for customers. >> Really funny to hear you say Europe is moving faster on this than other places. >> Yeah. >> Switzerland though. [laughter] Yeah. >> Yeah, we were moving fast and then the government shutdown kind of threw a wrench into things and um but uh yeah, there was a lot of work to get these things set up. They just uh crypto you can just like create an LST, launch a website and you're good to go. That's not how things work in the the TreadFi

world at all. But, uh, I think that's kind of what our strength is is like chewing glass for this like six to 12 month period to kind of set these things up. >> Carlos, do you have a followup? >> I I mean, I have a question, but I don't know if it's a question like for for a lawyer, so no not worries if you can't answer it, but like one thing that I've wondered is that you have all of these soul ETPs and ETFs, right? Uh, and it seems like you want to launch like a jetto soul like ETP, but wouldn't it be better like because if you think about institutional investors and they look at like an ETP product and it says like Jet Soul, they're probably going to be like what what the hell is

this, right? Like whe whereas if you just like allow for Jeta Soul to be used as a underlying asset for the existing sole ETPS, I feel that would be much better. But I don't know if that has come into the conversation. >> It has. Yeah, there's uh we spent a lot of time on that and I think Rebecca would know more, but um I don't really know where that conversation ended up. I think it's a little tricky, but uh yeah, we definitely agree with you. People are going to be like, "What is GTO soul?" Um so if it was just like a sole ATP, then that would be much easier to understand.

>> Yeah. I I guess we covered a bit of the institutional side. In terms of the DeFi end, are you guys still very focused on growing integrations for Jedos? >> Yeah, for sure. Um, yeah, our main focus right now is like Jetto Soul and BAM. um spend most of our time thinking about BAM and then we are starting to expand into other areas with some other new products that are a little more consumerf facing as well. So, that's something I've been spending a lot of time thinking about in between BAM and uh yeah, we should have more to announce on that later this year. But, um I think there's a lot of room for us to grow in

that and a lot of things that we can offer that other products aren't offering today. And so, yeah, we're hiring a head of product mobile. If you are listening and that's you, please apply. Just a quick show there. Very good. Um Sam, Carlos, unless you guys have anything else to ask, I think I'm going to do my wrap-up questions and then we're good to go. >> We good? >> Go ahead. >> Okay. Um Lucas, I have two things. Number one, um what is one thing that you guys should have done differently over the past 3 to 6 months? And number two, what are you looking most forward to over the next 6 to 12 months? This is probably a cliche, but I think we could

have been using AI more recently over the past six to 12 months. I fell down this like deep rabbit hole. Uh I guess it was like two weekends ago. I got open claw set up and uh you know I it's uh I've been using it more for BAM stuff to like build me custom dashboards um and [clears throat] like write features and all this and I don't know anything about how to work on frontends but these things can just like crank out stuff that's like good enough for me to like debug stuff or view metrics and things like that and it's just saved me so much time since I started using it. And so I in addition

to all the other stuff and like thinking like how do we leverage AI more and uh one thing I kind of realized is we have uh there's just so much knowledge inside of these different LLMs like our conversations with it like maybe Carlos is a chat GBT person Sam's like Claude and your Grock um and you're all asking it similar questions and uh like getting similar responses so I've been thinking a lot about like how do we unify that and use it as a learning lesson and uh like make it basically like keep feeding that in to like kind of recursively make something better. So that's a kind of like a a side project and then uh so

yeah I think we'll start using that way more and then um 6 to 12 months I would say pretty optimistic on BAM and some of the applications that we're working with and then um excited to reveal what we're working on with some newer products as well. Um, we're we're pretty excited about that. >> Very good. Very exciting. We actually just had maybe an hour before this conversation about um how we all have different contexts within like our separate agents and our separate um AI tools and that we need to unify them a bit because it is kind of a massive

thing like the fact that I don't know like me and Carlos can't interact our contextes with one another like our agents can't talk about it which is really messy. >> Yeah. I mean, you can imagine uh I'm assuming you guys aren't in person, but like if you're in person like the overhearing other conversations and like getting at least for engineers like debugging stuff and getting in the room or um you know collaborating with people it's uh easier but right now you're just kind of like collaborating with LLMs and uh there's a lot of stuff locked in there that I think is really good. Yeah, I think I imagine the next step is like co-c collaborative co-works and claude codes and stuff that are just going to

get much much better over the next year. >> Yeah, for sure. Yeah, it's crazy. The uh the pace of the pace of innovation AI is pretty insane right now. It's like a new model is being released every week. >> Yeah. And somehow they're all better than one another. But >> yeah. [laughter] >> Um Lucas, thank you again for coming on. This was very, very fun. And um for all of our listeners, make sure to check out the IBRL.wtf website to see what Lucas is talking about with regards to validators and what they're doing. And Lucas, if there's anything else you want to make people look into, >> uh check out BAM. It's BAM.dev.

That's it. >> Okay, perfect.