PropAMMs and WET w/Kevin at Humidifi

[Music]

[Music] Hey.

[Music]

[Music]

All right, welcome welcome to a Lightseed live stream. I'm joined by currently green screen uh Kevin at humidify

um the dominant I think leading maybe most dominant uh propm um which has become a pretty meaningful category um for

trading on Salana uh in the past 6 to 12 months. Uh

months. Uh I guess Kevin is your camera Okay. Can you see me?

Okay. Can you see me?

>> No. Is your Did your camera input change or something?

Let's see.

[Music] Don't think so.

H.

Technical difficulties here.

See if I can get you.

H.

It's interesting that when you are off screen, >> do you want to try real quick? I'll keep

things running just leaving the studio and then hop back in and we'll see if that fixes it.

>> It seems to understand that his camera is there, but uh it wants to be green right now. Um, and that does not make

right now. Um, and that does not make sense, but uh, excited to have Kevin on talk about all things humidify and prop

amms. Let's see if his camera will work now. Okay, nice. Streamyard is being

now. Okay, nice. Streamyard is being nice for us now. Awesome. Glad to have you on, Kevin. Thanks for for coming on today. Um, here to have you talk about

today. Um, here to have you talk about all things humidify propm, and related.

Um, maybe just to intro that a little bit. Um, I know the Jupiter guys did

bit. Um, I know the Jupiter guys did have a live streamer or spaces earlier today actually talking about their sort of the wet uh token launch and they're they're doing it on their their new DTF

platform. Um, and we won't talk too much

platform. Um, and we won't talk too much about the token details, but if you want to go check that out, go check out that that Jupiter um stream more and they talk about some of the details there.

Um, but I did want to talk to you about kind of all things propm. I think uh Dan Smith, shout out to him for being sort of, you know, the source for a lot of at least my initial information for this as

well as a new dashboard actually that we have at blocks block works which I'm not sure whether it's officially announced or not but we do have a humidify dashboard and it is on the public the

site so you can check that out a bunch of data about humidify um I guess before we get into anything um specific can you give some background like yourself and

the team at Humidify, like what what's sort of the background there, how you got into this space and and building a propm specifically and and what led you to kind of wanting to build this product?

>> Sure. Well, first, thanks for having me, Danny. It's great to be here. One of my

Danny. It's great to be here. One of my first podcast appearances, so we'll see how it goes. Um, yeah, I'm Kevin. Uh, I

kind of started my career out in trading. uh spent about nine years at a

trading. uh spent about nine years at a well-known HFT trading firm called Jump Trading and then decided to get into crypto about five years ago. Uh spent a

little bit of time at Paradigm and then at a trading firm uh called Symbolic Capital Partners. Um and with Humidify,

Capital Partners. Um and with Humidify, it really just started about six, seven months ago. Um you know I was trading on

months ago. Um you know I was trading on Salana doing some automated stuff and not really finding the sort of success we were looking for and got connected

with the temporal folks realized that we had a lot of complimentary skill sets and yeah decided to just meet up in person see if we could whip something together and after a weekend the first

MVP of humidify was born. Um and then from there on it moved pretty fast. I

feel like uh we weren't really expecting to hit the sort of volume metrics that we have, but that's not really been the focus. And yeah, hopefully today I can

focus. And yeah, hopefully today I can kind of share a little bit more. We've

been very secretive because by nature, like we are proprietary traders and there's really no reason to share any alpha, but we're getting into more of an educational mindset. We want people to

educational mindset. We want people to understand what's going on, how the microstructure is working, and we're hoping to contribute a lot to that. So

that's part of the the reason I'm on today.

>> Cool. I maybe to kick that off, if you can just start with sort of an intro to what a propm is and maybe like sort of generally how you or one generally would

operate um how you make markets, how you make money, kind of the the basics there.

>> Sure. So I guess I would start by saying the traditional AMMs uh on Salana like Radium, Meteora, Orca, they're built to

programmatically allow folks to deposit liquidity and to generate trading pool fees from those. Uh propm are a little different. Initially, they were called

different. Initially, they were called darkm on crypto Twitter because nobody really knew too much about them and importantly nobody really understood how they worked. Um these days people have a

they worked. Um these days people have a much better understanding. Um but the way that I would describe it is rather than taking customer deposits, we

currently use our own proprietary capital and more importantly we use a uh sophisticated trading algorithm that

updates our quotes in real time. So, uh,

contrasting with like the traditional passive AMMs where you kind of set your liquidity, uh, parameters and then leave it on ours, every single block, actually

multiple times a block, we're streaming in different updates to update the, uh, the prices that we're willing to trade at. And that's kind of what

at. And that's kind of what differentiates us from some of the other ones.

>> Gotcha. Um, and maybe one of the other sort of key I don't know shifts that's been happening at least on Salana is very quickly I feel like we went from all

trading activity going through sort of like traditional passive dexes to now the large majority over the past you know six to eight months has basically

quickly shifted over to going and flowing through prop a venues.

>> Um I guess can can you speak at all to why this happened? so quickly like was this an opportunity that was just sort of like right for the taking?

>> Yeah. So, I'd be remiss to not mention like the genesis of propm. We didn't

come up with this idea. Um there was an initial cohort of a few uh Soulfi, Ober and Zeri that kind of created this uh

this next generation platform and before them actually uh Lefinity kind of tried to pioneer something that utilized a lot of the same technologies specifically

using a off-chain oracle sender to update the prices in some sort of real-time method. I believe they were

real-time method. I believe they were using pith as a oracle and then with that first cohort they use their own models. Um

models. Um and so yeah the biggest difference is that um

prop ammms kind of require uh constant maintenance and constant updates to to the code. Um, and we kind of have been

the code. Um, and we kind of have been trying to add like innovative pieces to that. Either making our price oracle

that. Either making our price oracle better, making our compute usage more efficient, um, making, uh, the entire smart contract optimized,

um, all to kind of service customers, uh, to provide tighter quotes and tighter spreads. Was it a surprise that

tighter spreads. Was it a surprise that it took over so quickly? I don't really think so. like if um that's kind of the

think so. like if um that's kind of the beauty of crypto and especially on Salana where you know ultimately it's only performance that matters and if you're able to offer a better product uh within a couple months you could

actually acquire a lot of market share and become like one of the the top narratives. So uh for us you know we

narratives. So uh for us you know we were I think we did our first trade in June and I believe by maybe August we were already the number one propm and I

think in maybe October we became the number one decks on all of Salana. Um

and I think it's because we just built something that people wanted even if they didn't know what was actually going on under the hood.

>> Cool. That makes a lot of sense to me. I

think another big sort of like feature of of this development is that we're seeing you know a lot of volume on like new and hot pairs maybe more than you

would expect otherwise or like in the sort of traditional DEX landscape. So

recently I think you know maybe Zcash and even like Monad um are good examples of this where I think if I quote Dan Smith here something like 60 to 80% of

the total trading volume on the MON token was via Propm venues. So I guess as part of that maybe can you explain the thinking and process behind like

adding new assets like quoting new assets? um you must be sort of

assets? um you must be sort of constantly monitoring you know the market and what what goes into that decision-m.

>> Sure. So I think that traditional AMMs are still the optimal place to to trade longtail

tokens because um market makers simply can't spread inventory into like the top 5,000 assets on Salana. So we try to focus on the the short tail of assets,

the most common, the most demand uh driven tokens. So Solon Soul USC is our

driven tokens. So Solon Soul USC is our top pair, but we've added around 15 other tokens. Um and the story behind

other tokens. Um and the story behind MON is kind of interesting. Um

Wormhole had created a new initiative called Project Sunrise and they were kind of sharing their plans with us on Friday night. And uh I jumped on a call

Friday night. And uh I jumped on a call um with the team on Sunday night and they're like, "Hey, like in 8 hours we're going to launch. Would you guys be interested in trying to provide liquidity for this thing?" And we're like, "Yeah, why not? Like can wake up

at 5:00 a.m., give this a shot." We

weren't really sure how it was going to go, but we uh we kind of have a very stable setup to just set up new pairs.

And uh and yeah, we were able to get it running within like an hour of when it started trading on Monet itself and when it started trading on Coinbase. And uh

what was interesting was we didn't really know what to expect in terms of the performance but yeah by the end of the day I think we were one of the top venues in the entire crypto space. Um

within a couple hours we are offering the best prices of any venue uh including versus CFI exchanges and on some Dexes on Monet themselves. Um I

thought it was a very successful launch for them and I I was happy that Salana and Humidify got to play a role in it as well.

Nice. And and maybe somewhat related to that, um obviously inventory of assets is like a a piece of that like quoting

new markets or which markets you quote.

>> Um how does that play in? So like for example, if I look at some of the numbers, I know that your total sort of like inventory or TVL if we want to call it that is maybe only a few million or

maybe even like sub 10 million, but >> around eight I think right now.

>> Okay. Around 8 million right now. But I

think on most days, I mean, we're looking at 500 to maybe a billion plus in volume routed through you.

>> So, how does that like inventory piece work in relation to the markets you're quoting?

>> Sure. So, I think one of the biggest strengths of propm is um is their ability to use capital very efficiently. So with concentrated

efficiently. So with concentrated liquidity AMMs like Uniswap B3, we were able to see people place liquidity just

at the top ticks near the top of bookm kind of take that to the next level. Um

all of the liquidity is placed just for the next trade. So you know if people want to be able to sell all their Salana down to like a price of $1, like we won't be there. Like we have no

liquidity waiting to buy Salana at $1.

Um, however, if you want to sell Salana right now for, you know, as much as um, uh, if you want to sell one soul, then we're going to be able to provide you

the absolute best price. If you want to sell 10 soul, uh, we're going to give you the best price. If you want to sell 100 or a thousand, so we instantaneously give you all of our liquidity for just

that one trade. And so, because of this, we're the most capital efficient form of AMMs. We don't have any resting liquidity that is unused.

>> Gotcha. That makes a lot of sense to me.

Um maybe going into that a little deeper, how like when we consider some of these trades and the prices you're quoting, how do you then sort of measure profitability? Like what are then some

profitability? Like what are then some of the key factors that go into, you know, I'm making a market, I'm I'm filling some of these trades. Um how are we ending up, you know, net positive at

the end of the day?

>> Sure. So I guess uh maybe first alpha leak is that um for us the way that we measure profit is we send an oracle

update which is driven by a off-chain uh process that we run. This is a predictive price model. Think something

that like jump or tower would use in tradi listening to prices from Binance, OKX, Coinbase, all the major sexes, spot

and as well as some decentralized exchange venues to try to incorporate data that nobody else has access to. And

that is what we believe the fair value of the asset is at any given moment. And

we try to send that onto the Salana blockchain as efficiently and effectively as possible. Once it's

there, we give a price to users that we're willing to buy below that price at. That's our bid. And then above our

at. That's our bid. And then above our above that price at for an offer for any size trade they want to do for any asset that we quote. And then the difference between that oracle price update and the

price the price that we transact at, that's what we consider our profit. So,

you know, in theory, every trade we do um where we're able to land an Oracle update is a profitable one for us. Now,

we do have competitors and adversaries who are also very good at landing their transactions on chain. And so, sometimes they'll be able to beat us uh either with their trade or because um they have a better price model or they have a

better view on what the fair value of the asset is. And so, when we look at our total profit, we look at things like markouts, but we also look at the just total accumulation of the the funds in

our pools.

>> Gotcha. So then you know in in a scenario where say you have an adversary or someone else who's who's picking off your trades, what what is the reaction to that outcome? What do you do in

response and how do you maybe monitor for that or like think about it going forward?

>> So that's a fun question. This is kind of getting more into how do proprietary trading systems work and less about how does like a DEX operate or a propm operate and I'm happy to share a little

bit here. Um so yeah we we have uh we

bit here. Um so yeah we we have uh we have a huge amount of analytics and metrics on different participants in the market on uh all the traders that

interact with us um some directly some through aggregators and uh and yeah there's different flavors. There's arbitrage bots there's

flavors. There's arbitrage bots there's sexex traders there's retail traders and we kind of try to profile each one. Now

we've kind of taken a bold stance on what to do here. Um, unlike

some of our competitors who have decided to simply blacklist or refuse to trade against, you know, intelligent, informed, toxic actors, we up to this point have simply tried to improve our

system to make sure that we can remain profitable regardless of who's asking for a quote. Um, I wouldn't say it's a really philosophical stance. It's kind

of just because we're competitive and we don't like being beaten by anybody. So,

when we see that somebody's able to consistently profit off of us, we try to figure out why. And so far we've been able to kind of combat uh every adversary. I don't think this will last

adversary. I don't think this will last forever. Um but that's kind of like our

forever. Um but that's kind of like our mindset right now is like in order to offer the best service to Salana users to provide the best tightest spreads of any decks, you have to be able to

accommodate any trader. So you kind of have to be like the alpha on the chain.

And that's kind of like the standard that we're setting ourselves up for.

>> Gotcha. So, so almost using it as kind of a a learn and grow mechanism at the moment. I I would imagine if you sort of

moment. I I would imagine if you sort of just started one by one kind of blacklisting more and more potential flow you'd >> I don't know, maybe to your point, would would you inevitably sort of fall behind

if if there's more and more actors who are kind of beating you to these opportunities?

>> It depends. I think that discriminating between sophisticated toxic flow, toxic toxic actors and you know retail users trying to get the best price is not like

really black and white. I think that ultimately we kind of serve multiple purposes. One of them is to provide the

purposes. One of them is to provide the best prices for retail and if you know that you're transacting against retail, you can offer them better prices. Uh

this is something we naturally already do. Um, we've just taken the stance that

do. Um, we've just taken the stance that we also believe that there's other pools of liquidity on Salana, whether they're passive AMMs or other propms

and, uh, letting bots arbitrage our prices between humidify and, uh, and Orca or Humidify and Sulfi. Uh, these

trades are still profitable from our perspective. So, we're happy to to do

perspective. So, we're happy to to do them. Um, and this does lead to to more

them. Um, and this does lead to to more volume for us because we are helping Sulfi or Orca align their prices to what you know what what a what everyone

agrees the price should be. Um, and this is kind of just the current meta. Like I

I wouldn't be surprised if we continue to see advancements that continue to help retail users get better prices and maybe um make it more difficult for arbitrage bots. But for now, like we

arbitrage bots. But for now, like we we're not uh we're not taking a firm stance on that.

Okay, maybe one more kind of piece of this is uh and Dan Smith is is pinging me behind the scenes here to to ask about something, but uh it he really wants to

make a strong point here that we're actually seeing maybe price discovery move onchain because of how propm work in relation to like maybe some of these arbitrage bots and the pricing that they

see on maybe centralized exchanges versus what's being offered onchain.

What are your thoughts there? Do you

think that as a result of like kind of this buildout of, you know, more trading and and propm into the future that you just see more

and more trading come onchain and off of the centralized exchanges?

>> Yeah, that's definitely the goal. That's

the ambition is um if we're not able to quote every asset that CFI exchanges can support and that other venues can support, we're not really advancing the purpose of crypto to create

decentralized markets. And this is

decentralized markets. And this is something that we strongly believe in.

Um, I think we're really at the inception point of firms realizing that just taking, you know, Binance prices, just copying Binance Mid onto the

blockchain is not actually providing a service to anybody, but instead you need to help create liquidity where uh people like don't necessarily want to

uh to trade on a on a sex. And right

now, the passive Amms support this. They

always are ready to set a price, but they're doing it in a way that's not necessarily the most efficient. So,

we're trying to kind of marry what we're good at, landing transactions, having a good price model, and then also aiming that at like new assets, some that don't even have uh C5 listings. We kind of

have an inside joke. We call it abstinence trading uh because there's no sex involved. But the uh the idea is

sex involved. But the uh the idea is that uh there's no reason why we can't create a price model that only looks at decentralized uh trading activity. Um

you know all the other HFT firms have done it. There's always some source of

done it. There's always some source of price discovery and we want humidify to to become that for Salana for for as many assets as we can support.

>> Is one of the examples there like ore would come to mind? I don't know if there's any others that you're quoting today that are not on sexes. Yeah, Ora

was the first one we tried. It's a very challenging one because that thing moves a lot and um there's no C5 venues at all and there's actually not too many D5 venues for it either. So, what it takes

is um kind of some creativity, some innovation. We're able to set it up and

innovation. We're able to set it up and we're still experimenting with it. Um

but I think you should expect to see more of this coming from us. whether

it's uh whether it's um tokens that don't have C5 listings at all or hopefully tokens that you know do have CFI listings but the majority of the volume will happen on Salana on Humifi.

>> Gotcha. And so and so naturally maybe a a big part of that is if there's a lot of demand on chain for a particular asset. It doesn't really matter if it's

asset. It doesn't really matter if it's got a sex listing, you know, there's there's people that want to trade it. So

it it might be or probably is worth quoting.

>> Yeah, exactly. I think that you know Salana wants to be internet capital markets in order to do that you have to be able to support the assets that people want. You don't have to support

people want. You don't have to support assets that nobody wants but if uh if traders want to access or liquidity or if they want to access monad liquidity or if they want to access Zcash

liquidity and Solana wants to be the home of trading then it needs to support these assets and it needs to support them on day one. um needs to support them at minute one and needs to support

them by giving the deepest uh liquidity and the best trading spreads for users.

>> Awesome. Um let's dig into maybe some of the details then and anything you can share. Um and maybe there's limits to

share. Um and maybe there's limits to what you can share, but on sort of how you improve your systems. I think you know one of the big things you mentioned

is Oracle updates. Um and from my understanding like the amount of CUS that are used in transactions is an important piece of this. We have some data now on our humidify dashboard at

block works um that shows you've kind of decreased this this usage of cus from like the 300 some level to like under 50

over the past six months. I guess

ultimately how impactful is that to your operation and like your success and the quoting of markets?

>> Yeah, it's definitely a very important component. I think the way I'd contrast

component. I think the way I'd contrast it to traditional markets is different entities have their own view on price and there is no right price

like um but when you want to express your price you have to have some execution logic. you have to send an

execution logic. you have to send an order to the exchange and certain firms just have faster execution logic than others. Uh some folks now use FPGAAS and

others. Uh some folks now use FPGAAS and AS6 in uh in traffic markets. So I think the ability to send a transaction using

the lowest amount of compute units is effectively latency uh but in the blockchain sphere. So for us being able

blockchain sphere. So for us being able to reduce from 300 to to 47 uh this means that when the sequencer is deciding whether to put our Oracle

update ahead of some other popmms or Oracle update this will be one of the determining factors. Now there's other

determining factors. Now there's other aspects as well. Um I'm happy to share a little bit more about my thoughts on those but in this case our partners at Temporal you know are kind of like the

kings of being able to optimize for this. So yeah, shout out to uh KV and

this. So yeah, shout out to uh KV and Ben for for whipping up the new price oracle. They kind of see it as like like

oracle. They kind of see it as like like a like a contest. Um and they want to keep bringing this number down as as much as possible. But ultimately what it does allow us to do is to get our Oracle

updates and our prices onchain faster than anybody else, which means that we're instantly willing to trade with anybody at the tightest spreads. If we

were unable to do that then we may end up getting arbed against other AMMs and if we do that then naturally the response is to quote wider and then everyone suffers.

Gotcha. And then so obviously another piece of this is that not all of these updates not all these transactions are successful. So

successful. So >> like successful to reverted transactions I guess are the reasons for you know sort of having a an oracle update revert

something that is on your end like your model was off or you were too slow or you missed something or is that something >> like the trader you know the price was offered to them and they clicked the button and then it was too late like it

was already stale. How does that work?

So I think there's two separate questions there. For the first, yeah, we

questions there. For the first, yeah, we try our best to land every Oracle update and given Solana's microsstructure, there's, you know, 400 millisecond

slots, but within that there's a very complicated system of how transactions actually land on chain. Some combination

of, you know, TPU as well as through jetto. Um, and so we try a lot of

jetto. Um, and so we try a lot of different methods to land our oracle updates. Sometimes they do revert. Uh

updates. Sometimes they do revert. Uh

worse sometimes is if you send a Oracle update because the price has moved up, but then the price has moved down and you need to send a new one. And if you only get the first one to land, then now

you're like willing to buy at a very high price and the price is completely reverted. You know, you can find

reverted. You know, you can find yourself uh losing money in those situations. So yeah, we I think the vast

situations. So yeah, we I think the vast vast majority of our oracle updates land and that is like the full intention. Um

I think your second question was about user transactions. Could you say that in

user transactions. Could you say that in a different way?

>> Yeah, I guess understanding if or what is the difference between for example, you know, a a quote or you know a transaction not you know succeeding due

to it being >> sort of you know on the protocol side or being on the user trade side. I know

just like for example like if I'm trading on Jupiter and maybe I don't have my slippage set properly or something like that maybe my TX gets reverted for that particular trade. Does

that impact some of the data that you look at and how you have to kind of interpret you know how you're being successful?

>> Got it. Yeah. So I think it's unrelated to our Oracle updates. We do have some logic so that we're not continuing to quote if we haven't received Oracle updates in the, you know, very recent

past. The entire premise of why we're

past. The entire premise of why we're able to be successful is because we're able to always have up-to-date prices.

Um, but when users transactions fail, it could be just because their slippage tolerance was set too low and the prices moved away from them. And from the moment that they received a quote to

trade at, you know, $130, now Salon is up to $140. So because of that, we had an Oracle update sent between when they saw what's on their screen and when they clicked, you know, sign. And so if that

happens and their slippage tolerance is too low, then the transaction won't be able to execute. Um yeah, all of our technology like our CU optimizations, our landing technology, which we use uh

Temporal's Nomi product for, is to allow us to quote safely and aggressively. And

so we can kind of do as few unprofitable trades as possible and then pass that benefit on to uh to users. We're also

driven by competition. We have a lot of formidable competitors that you know kind of keep all these spreads as tight as possible. But the micro structure is

as possible. But the micro structure is kind kind of always evolving as the client teams make improvements. The idea

is like you have to find ways to to play the new game.

>> Yeah, that makes that makes a lot of sense to me. Um maybe coming, you know, looking at the market from like the recent kind of 1010 event that played

out. I know that um some data was shared

out. I know that um some data was shared around where there were like certain periods during that super volatile activity where some prop amm stopped quoting prices

>> and I'm guessing in that case it's you know maybe you're not confident you can like offer the right price or something along those lines. I guess some of the

followup to that was, you know, maybe there will always be a need for some passive liquidity venues to to take flow if there's periods where you just wouldn't be comfortable kind of

providing a price. Do you think that's the case? Or is there some scenario

the case? Or is there some scenario where you could have continued quoting prices like if you had some more information about the market or something along those lines? Yeah, this

is a funny one because um I don't really engage too much on crypto Twitter. I'm a

big lurker, but seeing some of the theories out there was um it it was uh bothersome, let's say. I can kind of share what actually happened there. Um

the truth is like during that huge amount of volatility, our Oracle sender uh our price prediction model fell over.

Um as did a lot of other HFT strategies, as did our C5 strategies, as did you know um many others that just couldn't handle the unprecedented volatility. And

I happened to be at lunch uh interviewing a candidate. So as soon as I saw the messages, I was like, "Oh no."

and had to run back to the computer and like basically restart the model because we had never faced that since the beginning of Humidify.

By missing that hour or so of trading, I think we lost, you know, more than $500,000 of profit. Uh those are exactly the conditions in which we thrive

because of our ability to have a better price model than competitors and our ability to land transactions in all s sort of um market conditions. So yeah,

it certainly wasn't us intentionally trying to stop quoting. Um the way that humidify is set up, it should be able to continue to quote profitably in super

high volatility conditions. We just

hadn't made our system robust enough. So

that's what we're continuing to try to improve. As to would I recommend

improve. As to would I recommend humidify be the only venue for trading?

Like no. I think there should always be additional uh resilience. I don't think any sort of market microsstructure should rely on one party. I think that kind of defeats the whole purpose. So I

think that having other propms you know available other AMMs in general available and other you know central order books that are evolving there is a is a very healthy thing for the

ecosystem.

>> Gotcha. Yeah I I do think that makes quite a bit of sense. Um I was just I was digging through some threads um just looking for some maybe some interesting points of discussion and that it it did

seem to get a bit derailed there about um you know >> I get it bad actors versus good actors right people want to draw their conclusions but >> I think it's natural for market makers

to reduce the amount of liquidity during high volatility conditions um for the really uh astute you know data analysts they'll see that we do a little bit of this as like when there's higher

volatility, the quotes widen out and the liquidity is a little bit thinner. Um

those are simply the incentives. You

can't ask people to lose large amounts of money when there's high volume. Um

what you can do is you can create market microructure that encourages participation encourages competition. So

that if one participant cannot provide tight spreads or has to widen out a 100x then that means that somebody else can come in and widen out only 50x and if

that is still very juicy then somebody else can come in and widen out 10x. And

so we are definitely very pro competition. We don't want any one actor

competition. We don't want any one actor to, you know, be seen as the only canonical one. And we think that it's

canonical one. And we think that it's necessary to have multiple participants because competition kind of drives the space forward. It's definitely what's

space forward. It's definitely what's kept us motivated and um and innovating.

Yeah, I think I think you commented on maybe what the right conclusion is there is that you know the expected behavior might be that spreads widen out during those periods, but it should not be

expected that you know all propm stop quoting during high volatility periods.

That's that's more of like a growing pains type of situation that we're experiencing.

>> I consider it like a skill issue. Like

in our case, our system wasn't ready. Um

now it's much better suited for that.

Um, but that doesn't mean it'll never go down ever again. And what happened was we opened up a huge opportunity for some of our competitors who did very very well during that period. So we were

unhappy that we couldn't participate.

Um, and if you're unable to be profitable trading in those conditions, then yeah, like there's an opportunity for somebody

else to come in and and offer the service that you're doing, but able to to monetize it. I think ultimately the incentives need to be there though and

it's um unreasonable to ask anybody to to quote unprofitably just because >> Yeah, I I think that makes a lot of

sense. Um it's going to be hard to ask

sense. Um it's going to be hard to ask someone to lose money on purpose without incentivizing them in some other way like like you're saying. Um, maybe to get outside of Salana a little bit, I

guess I'm just curious here if you can talk to like what are some of the key differences from your POV in trading on like soul versus like ETH L1 today or

even like the L2s like base and could or would or why wouldn't perhaps humidify and other propm expand to a chain like base?

>> Yep.

Um, so Eth one currently still has 12 second block times. Uh, and Solana is now down to effectively 50 millisecond sub slots within these 400 millisecond

slots. Um, that changes what you need to

slots. Um, that changes what you need to do. Well, uh, on Ethereum, being able to

do. Well, uh, on Ethereum, being able to price assets is very, very important, but latency is a little bit less important. You only have to transact

important. You only have to transact once every 12 seconds. So it's very important what you think the right price is and how much liquidity you have at that very moment is on Salana it's definitely more high frequency in nature

like every 50 milliseconds there is some sort of transaction that can happen and then the shred goes out and being able to consume data and respond to it very quickly and land your transactions very

quickly is still kind of more of an art than a science maybe the temporal guys would disagree they they seem to have figured it out for us it was very challenging to understand how exactly to land your transactions do send this

thing through Jetto, do you send it through, you know, vanilla, uh, TPU? Um,

whereas on Ethereum, the the market micro structure is a little bit more well understood. You kind of the supply

well understood. You kind of the supply chain is more well documented. Um,

people are more confident in these actors because they've been around for a long time and they have like very predictable behaviors. Um, okay. So,

predictable behaviors. Um, okay. So,

that's Ethereum L1. Um, Ethereum L2. I

think that base is probably the the closest to being able to offer a similar micro structure. Um, maybe another alpha

micro structure. Um, maybe another alpha leak. This one not about us. Uh, one of

leak. This one not about us. Uh, one of our Salana competitors, Tacera, uh, seems to have launched a propm on base.

So, I won't, uh, I won't dox the address, but for, uh, for those sleuths, feel free to go and take a look. So

yeah, I think they've already proven that this can be done. Uh the micro the market microsructure again is very very different. You have a single sequencer

different. You have a single sequencer there.

Um and people kind of have an idea of where it is. So the place that you need to get all your data to is like better understood on Salana is kind of a rotating leader uh around the world. You

kind of have to have good connectivity everywhere.

one secret in the industry that people maybe didn't really understand well um call it

six months ago was that using dark fiber in order to move data around is a very important advantage.

Now double zero is kind of you know blown the spot for everybody here.

Everyone now understands that having better networking, having bigger pipes, uh, is critical to running a very performant chain. But I do remember

performant chain. But I do remember accidentally leaking that, you know, dark fiber is a very common tool for sophisticated HFT uh, firms to to one of our direct competitors at the time.

Actually, I don't even know if humidify was running at the time, but I I definitely shared that insight with some of our uh, now competitors and um, and I was told that within a couple months they had implemented it and seen like a

nice improvement. So these are the sort

nice improvement. So these are the sort of um microsstructure differences that that do matter. I don't think propm can only uh

matter. I don't think propm can only uh exist on Salana, but I think currently they're by far best suited for Salana.

And Salana is catering to you know the conditions required to operate effective propm so that it can it can serve its users to make Salana the home of

trading. So

trading. So uh the way I see it, it's not just about seeing what the current substrate is and accepting it. It's also in informing you

accepting it. It's also in informing you know the broader client teams, the broader community, apps, routers, educating people on what is needed in

order to create the desired outcome and uh and yeah like Salana's kind of really double down on trading. So this is like why it's very fun for me to to engage

with them about that.

>> Right. So it it almost just makes the most sense today like you're saying to kind of be laser focused on Salana because in a kind of similar way Salana

seems to be laser focused at least on kind of leaning into this uh sort of opportunity.

>> Yeah. If um if the client teams aren't able to make the required optimizations then it's difficult to unlock you know the the microsstructure required and um

yeah so far like I I kind of study a lot of different ecosystems. I think everyone has its strengths and weaknesses from a trading perspective. I

think currently Salana is like clearly the uh the top chain, but there's other chains trying to compete as well. So,

we'll keep looking at those. Humidify

though it home will always be on Salana.

>> Okay, cool. Glad to uh solidify that point. Um you did mention you've

point. Um you did mention you've mentioned temporal a couple times. I

wanted to dig into that a little bit.

Um, you know, I know there's some kind of connection there between humidify and temporal and I guess there's some other sort of like efforts or projects under that umbrella like Momi Harmonic for

example. Like do you have a connection

example. Like do you have a connection to those other efforts or is there cross collaboration among those teams? Like I

would imagine there's benefits to like having a really good understanding and software tools around you know block building and TX landing and trading um that would kind of like you know have

synergies there. What's how does that

synergies there. What's how does that look for you and and the humidify team?

>> Yeah, so the humidify team is myself, one other person that I work closely with and uh some of the engineers from the temporal team. Um it's kind of like

a a joint venture of sorts. Um but we only focus on humidify. So, uh, we use Nomi because it's the best product to

land transactions for Humifi. Um, but I actually don't have any too much knowledge about what Temporal's plans are outside of Humifi or what they're

doing with Harmonic. I kind of get the same news as everybody else. Um, I guess Ben and and Connor, if you guys want to uh talk to me about uh making me an adviser or something, I'm I'm open to

it. But right now like uh our focus is

it. But right now like uh our focus is just on humidify and using nomi makes a lot of sense. I don't believe there are any plans to vertically integrate to um

to do anything specific with harmonic but it's definitely like a very interesting area that we're we're looking at carefully. Um

what we do do is we try to study like JTO which is the current um primary way of landing transactions and seeing like what the best way to do that is and

proposing improvements and um proposing um different ways to to make the microsstructure even more efficient. Um

yeah, you're going to have to talk to the temporal team directly about whether there's some grand plan. I'm not aware of one.

>> Cool. uh we can maybe we can dig into another relationship um on onchain in that you know a lot of volume and and maybe this is kind of a path dependency

thing with Salana but you know a lot of the volume for propm comes from aggregators like Jupiter like Titan like Dflow for example >> um and it seems the dominance of those

aggregators maybe has helped in like propelling the market share growth um does I guess does that reliance on them

present any challenges or like if if Jupiter and DFlow change their algorithms for example and kind of where they're routing or how they're routing does do you have to react to that or are

you guys really just you just focus on on what you do in terms of pricing and quoting?

>> Yeah, that's a really good question. Um

I think that itself has kind of evolved over time. I think one thing that

over time. I think one thing that allowed propmmes to be successful initially, which is no longer the case, is that there was really one super

dominant aggregator uh Jupiter. And if

you could connect your propm to their flow, that would be enough uh opportunity to to justify the work

spent. Um and so now there's, you know,

spent. Um and so now there's, you know, this very competitive landscape. I think

Jupiter is still the top aggregator, but Dflow, Titan, OKX, they're all competing for more um more users, more market share, and I think it's great. I think

it pushes the entire space forward. Um

for us, initially, we kind of had to conform to the requirements set by Jupiter. And because we're still fairly

Jupiter. And because we're still fairly early, I think the other aggregators kind of just set the same requirements to to support as many propms as quickly

as possible so that they can also get access to the the best liquidity and the best pricing.

Um, but I kind of expect this to change.

I expect, you know, humidify to put in features that we require the aggregators to support. I expect the aggregators to

to support. I expect the aggregators to continue launching new features that will need to support and kind of just work on this together. Um, we don't have a specific bias or preference for any of

the aggregators. Ultimately, we're just

the aggregators. Ultimately, we're just our customers are all the users on Salana, but if one aggregator is shipping new features better and faster than the others, then they're going to

start winning and we're definitely going to have be happy to support that.

Maybe with that in mind like about serving kind of all Salana traders are or your users is there is there ever a world where a humanifi like moves up the

stack to a consumerf facing app or does that just not make sense like you just you stay on the back end and and you keep doing what you're doing and scale that.

>> Yeah. So maybe I can explain our place in the ecosystem for people that aren't fully aware.

So yeah, our our logic is composed of, you know, this off-chain oracle we mentioned, this onchain smart contract, but yet we don't really have a front end. Um, what we do is we have a smart

end. Um, what we do is we have a smart contract which these routers can access and they ask us for, you know, what is the most amount of soul that I can buy

for a,000 USDC and we give them the best price. the other AMMs all you know

price. the other AMMs all you know effectively return a price and then the router picks the best venues and composes the best you know swap and then gives that to the user so the user can overall leverage all the liquidity on

chain. Um what the routers do is a

chain. Um what the routers do is a little bit different than what we do. Uh

we do have some experience in coming up with routing algorithms. Um but it's not really an area that we think we're you know best in the world at. It's

something that we think the routers are doing a pretty decent job of. We've

enjoyed seeing continued innovation here. I think DFlow recently announced

here. I think DFlow recently announced something like, you know, JIT routing where they check what the price is in real time rather than relying on what the price was at the time of the quote.

We quickly saw Jupiter also implement something very very similar in addition to a sleuth of other a slew of other features. So this is what we like to

features. So this is what we like to see. The other part that um makes it not

see. The other part that um makes it not obvious for us to pursue is there's a lot of go to market marketing that goes into this and yeah here I think like the

aggregators are just killing it. I think

like the Jupiter team is is uh is probably the best in the business at building a community of users building a product suite and then bring more and

more users to Salana. So I see our role as providing the tightest spreads, giving the be best trading outcomes, but we definitely need partners at the

aggregator level and above and beyond.

Um because the the the ambition for humidify is not to be like the all-in-one app where we do everything.

We kind of see ourselves as the technical trading focused service provider that allows a lot of this stuff to happen. But we definitely

need our our partners to to accomplish that mission.

>> Gotcha. And I guess speaking of partners, um mentioned it at sort of the top of the hour, but you are or Humidify is doing its wet token launch um via

Jupiter's new DTF or decentralized token formation platform. I guess like I said

formation platform. I guess like I said before, if you want more details about the token specifically, go check out that JUP stream. But uh I am curious like specifically on this choice like

there's a lot of ways to go about launching a token. I mean we just saw Monad do like the first ICO sale on Coinbase for example. Um why why DTF?

Why work with Jupiter? Like what was the reasoning behind that?

>> Yeah so I think it was pretty obvious to us that it was the right choice. Um

there's a lot of alignment in the principles of DTF and the product offering itself. We care a lot about

offering itself. We care a lot about transparency. Even though we've been

transparency. Even though we've been fairly secretive like the the reputation behind the people uh that work on humidify is extremely important to us.

So we we want to do things the right way. Um and having transparency about

way. Um and having transparency about how tokens are disseminated, where the token wallets are, what's going to happen, like this is something that we care about. Um so that was a really good

care about. Um so that was a really good fit. Obviously like we just talked about

fit. Obviously like we just talked about Jupiter is one of the top aggregators in the space and has been there since the beginning. You know we've kind of

beginning. You know we've kind of evolved some of this market microsructure with them and they they they are very well informed on what the

what the right um path forward is. But I

would say that's like pretty orthogonal to the GTF platform. This is like kind of a separate offering. Um the users there I think are folks that are kind of our audience. Um, currently it's still

our audience. Um, currently it's still more Salana focused, but we are trying to bring more attention to um, bring more attention from other ecosystems as

well. I think we're still early on that

well. I think we're still early on that uh, maybe too early to do something a little bit more crossplatform.

Um, and finally, yeah, we're excited to be like the first uh, project on DTF. We

wanted to kind of kick things off in a very successful way for them. So, um,

yeah, I guess we'll we'll see, uh, we'll see how it goes. I think the launch is tomorrow morning.

>> Yeah. So, we'll see. I guess we'll see that go live on Jup. And I think the I guess timing of the actual PGE is maybe

this Friday or the 5th. Um, so yeah, to your point, I guess we'll see we'll see how things go and we'll see how that plays out. Um, I don't want to take up

plays out. Um, I don't want to take up too much of your time here and we're getting near the end of our time, but maybe as one kind of final question, I'd love to get your kind of maybe broad

thoughts on Salana generally like developments around the chain itself, the infrastructure. I think you see a

the infrastructure. I think you see a lot of excitement as we have for the past couple of years around, you know, fire dancer at some point. You know,

let's get to a million TPS. What are you most excited about and maybe the humidify team about in terms of like the improvement of the underlying chain?

Like what's what's most important to you? What are you looking forward to

you? What are you looking forward to improving kind of over the next, you know, year, couple of years? Whatever is

important.

Yeah, I'm I'm definitely going to be biased here and pick something that we can play a role in. Um, which is kind of

becoming the primary home of liquidity on day one. There's

definitely like a huge amount of tech that is that we touched on today and we can kind of spend hours talking about that supports this. But

I think the game we're playing right now is kind of who can improve utility for existing Salana users the most. But in

the long run, we want to help users trade on decentralized platforms. Um, and there's just like a huge amount of work that that goes into that. I know

MCP is maybe a year out, maybe six months depending on who you ask. Uh

there's a bunch of interesting other companies working. You know, I don't

companies working. You know, I don't really have any personal opinions on which designs are most likely to win out, but I just like the discourse and I like the level of

competition that's kind of um forced everyone to to push even harder. And

yeah, what I would like to see from Salana is just a continued emphasis on doing a few things well or maybe just one thing well and uh becoming the best at it rather than trying to do

everything like in a mediocre manner.

So, a little bit of a copout answer. Um

I'll try to get back to you on some of the uh technical uh things that I think are the most exciting. Maybe just to follow up on that kind of day one venue

point you made is like what's the what's the biggest need there or like if it's technical, if it's, you know, integrations or additional support or

just continued build out of of things that exist today? Like what's the the next step change there to make that um even more inviting or more meaningful than it is today? Yeah, I think what we

saw with the Monad launch last week was it requires a bunch of people pushing in the same direction. Um, you need to be able to bring the asset on chain. Uh, I

think in this case, Wormhole was able to do that with their Project Sunrise initiative. Uh, once it's there, you

initiative. Uh, once it's there, you need multiple exchanges to kind of uh, list the asset and trade the asset. you

need those venues to be able to do it without relying on other external price oracles or relying on a CFI listing. Uh

that's kind of touching on the stuff we talked about with ore. Um and you finally you need users like uh if nobody wants to trade there's not this entire flywheel isn't going to work. So there's

definitely a bit of a chicken and egg problem. But what I think is necessary

problem. But what I think is necessary is you kind of have to play for keeps here. You can't just uh halfass it. Um,

here. You can't just uh halfass it. Um,

if you want to become the home of trading, you need to be able to support all these things from day one. And then

once people start trading on your platform right at ICO or TGE, they're probably not going to want to bridge their assets to another chain to uh to to leverage the benefits there. They're

probably just going to want to stay there. So, I kind of think you have to

there. So, I kind of think you have to do everything well. I think, you know, you don't want to go to to Coinbase, participate in the pre-sale, uh, receive your assets, and then

realize that humidify is probably going to be the best venue. Bridge your assets to humidify, you trade on humidify. Oh,

now you have to bring your MON back to to Monad to stake. So then now you transfer them. Like this sort of um

transfer them. Like this sort of um fragmentation I think is something that we're going to see uh be addressed by a number of other um parties. But I think

at least when it comes to trading the asset, investing in the asset, providing the best outcomes for people trading $10 of tokens and also $10 million of

tokens. Um, all this kind of needs to be

tokens. Um, all this kind of needs to be done in concert.

Awesome. Great answer. Um, we can we can close it out there, Kevin. It was great to have you on. Um, I'm really glad you could kind of pull back the curtain a

little bit and share some, you know, maybe they're somewhat known secrets among kind of those who know. Um, but at least for maybe some of our more the

layman out there that just uh like to trade coins on chain. Um, it was great.

>> Hopefully some stuff that nobody >> learn more hopefully some things nobody knew about that that would also be great too. Um I know you you it sounds like

too. Um I know you you it sounds like you did drop some alpha on uh some specific things um as I think back. So

uh anyway, great to have you on. Uh I

don't want to take up any more of your time, but uh good luck with uh the token launch and good luck just generally with Humidify and uh I guess keep on uh you

know with with the competition on Salana now. It just seems like I don't know

now. It just seems like I don't know month over month that it seems to be kind of ramping up the level of competition between some of these teams. You know, you mentioned Juplow Titan are going at it and I think in the propm

space you guys are definitely going at it. So, uh good luck and uh thanks again

it. So, uh good luck and uh thanks again for coming on.

>> Thanks. It was it was fun.

>> Awesome. All right. Thank you guys so much for tuning in. Uh you will hear from me and Lightseed again uh at the end of this week. We will have our first

roundup episode with some well-known friends and faces of the show. Uh Alique

One, Dan Smith will be on the episode, but a few other uh familiar faces from Block Works Research will be joining as well for our first roundup back. Um, and

then we'll we'll move to kind of a traditional, you know, probably one-on-one roundup style where where I'll bring on some of those, you know, one of those guys uh each week to kind

of chat about all things happening on Salana. But, uh, thanks for tuning in.

Salana. But, uh, thanks for tuning in.

Glad we could, uh, chat with Kevin about all things humidifying and propm. And I

will see you next time. Peace.

[Music] [Music] Heat. Hey, Heat.

Heat. Hey, Heat.

[Music] I feel you.

[Music]