Trends in Decentralized and Agentic Finance

Transcript

[Music] All right, welcome everyone and thank you for staying till the end of the day. We are the final session of the day. I'll try to try to make this a little bit more interactive and keep the energy up. Thank you for being here. Uh we've got K again, extra corny the second time, the sequel. We've got Guiam from Panoptic and we've got Jack from Clanker wearing a Clanker shirt representing his brand there. I'm James, founder of Quant.fun and I'm excited to cover some of these really interesting topics in DeFi but also DeFi AI. And so I'm gonna give the floor to these gentlemen for two minutes each to explain who they are and what they do before we get into pulling the audience for some questions.

>> Hi everybody. I'm K. I do BD for year finance. We are a collection of vaults and people put deposits into those vaults and we go and move them into different uh supply side lending markets to earn yield and then we return it all back to the users and we have about $750 million in those vaults today. >> All right. Hey, I'm Guiam Lumbbert at James. Your profile photo just looks just looks just like you. Uh, great fun. Uh, I'm the uh founder of Panoptic. Uh, we're doing options on Ethereum and we can go on all the EVMs. Uh, we're rethinking options to make them work better with the confines of DeFi. So, it's not an order book. It's more like a unis swap like model. Uh, that means

that we can innovate quite a bit on that options layer. Uh, options have had a huge use case in Treadfi. We haven't seen any product solve or get uh options to work on DeFi. We think we're the first one. We've launched in December. We're in a uh leave of absence now, but we are launching again on V2 with our V2 in December again this year. >> Hey everybody, I'm Jack Dishman. Uh I built Clanker as a side project last November. Um Clanker for those aren't aren't familiar, it's a token deployment framework uh primarily on the base network. We're also on arbitum ab abstract um and uni chain. Uh the goal is to be kind of uh what next.js is to

react clankers to unis swap v3 and v4 uh allowing flexible token deployments. Uh in the past 10 months we've done our tokens have done about uh I think like 3.8 8 billion in volume and generated about $38 million for creators interfaces in our protocol. Um, and that's across around $455,000 tokens. Uh, it started as a a chatbot on Farcaster. Um, but now we have an API SDK um in a UI deploy. Thanks.

>> That's under a year. That's pretty impressive. >> Yeah, it's been pretty crazy. >> Yeah. All right, guys. Unfortunately, uh, Nick Emmens could not join us. So, this talk was originally about DeFi and AI. It's going to lean more Defi, but tomorrow I'm going to plug my own talk for quant fund, my startup at 1:30 p.m. at Salana Boston, getting more into the agentic side of things and talking more about virtual protocol. And so, uh, I'm kind of curious. We're I'm going to ask you all to raise your hands for these different questions. We're going to get an idea of where you're at and that'll help us answer our questions. Also help me learn more about our audience. So, uh, do you have any non-token DeFi positions actively?

Only a few. Okay, so people there. Uh, do you actively manage your DeFi positions? If you do actively manage, it's not just passive stuff. Okay, so only a few folks. Uh, do you have any non-trivial TVL in an agentic DeFi product or platform like virtuals or Morpheus? Very few people. Very few. Okay. Okay. And then um who in the crowd has sent an onchain transaction with an autonomous agent? Okay, still it's a very small minority. All right. And then was it on a social platform? You can keep those raised.

Okay. That's good. That was that was Jack's question. Now you know. And then um do you believe that agentic traders will or have already become dominant in DeFi markets? Like if they will become or they have already become dominant in DeFi markets, a few of us. So most people think that it's going to be manual humans doing the trading. Okay, great. Um, so for for my gentlemen here, after seeing the the audience's responses, now you can tailor some of your answers. Um, my first question is, what's something you understand about DeFi or markets that at least half the people in this room might disagree with?

So at least half the people might disagree with or not know about. Try to give us a spicy hot take. It's the end of the day. My take is that I I know some people here just said that that the agents would dominate. Uh I think there's always going to be a human behind it and they're going to set the guard rails for those agents to operate within and I don't think there's ever going to be just like an agent that you know takes over everything. I guess no Terminator. >> All right. What do you guys think?

>> I mean I'll take a riff on this. I think the um the agents maybe they work they work with the longtail of assets the newly launched tokens those that haven't seen a lot of trading if you try to outrade uh with an agent ETH or Bitcoin or there is more mature tokens it's never going to happen it's like trying to train an AI to play roulette better than the odds I would say that yeah the the the reason why they have success now is because the market is so new and uh immature that when we do get to a point where everything is more uh efficient then they will there will be no edge left.

>> Yeah, I'll I'll kind of rift off that right like I think that you know you know AI bot probably it won't outrade you know the best traders out there. Uh I I like to think of AI as a tool right where it makes natural language processes easier into you know executable commands. Uh, so I think that's like the the real breakthrough here where it's, you know, now you can trade tokens in a social feed. You can email to, you know, make these transactions really kind of like plugging it and playing anywhere. Um, and I think that's that's the true unlock there.

Great guys. And then, uh, could you share a hot take for onchain AI and crypto that maybe at least half the people in the room don't know about or don't understand or don't agree with you? I have a feeling that most people who are uh using these products today are probably taking losses. That's true to my understanding. I just wanted to check. Uh please >> any other thoughts on this? >> Um um I mean more generally, is it only AI and crypto or crypto more generally? Um >> just AI and crypto for this question. >> Oh, okay.

>> Yeah. >> No. Yeah, the agents is one thing. I'd say that what may be surprising is that the AI can be beneficial, help you write emails to buy. It can be adversarial, too. You can imagine a purposely built AI that is attacking protocols, trying to snipe you even better than the MV bots. There's a whole landscape of counter measures that may need to be created because these AI agents may be working against you. Um yeah, I honestly like I think that like you know the AI chat bots are or like you know there's like two sides of like AI chat bots in you know liquid markets right I think the first side is like an entertainment right you can give

it a personality you can make it a cute whatever you know anime person or whatever um and you know that attracts a different crowd right and then there's like the utility side right where it's like maybe there's you know you know you can maybe put up perpetuals or you know things like that. And I think there's like you know it's trying to separate like the signal from the noise right like you know you can have a very cute you know PFP or something but like if it doesn't do kind of like the broad variety that's that traders expect or you know people that are interacting with these contracts um it it definitely falls short. So just be cautious.

>> Yeah. Yeah. I can't say that I disagree with you guys. I know that when I look at the agentic investing ecosystem, it seems like most of the traders are not really doing very much. I know that what the leading one in virtuals for example is named Axelrod and it has 5 million in TVL but it's lost 1% and the only reason it's lost 1% is because it's only investing 2% of its portfolio right now and it's lost half half of that and so I think when you look behind the curtain of some of these AI projects that are investing crypto it's still very very early and I'll talk about that a bit more in my in my conversation tomorrow but you guys all are up here because you're good at helping people make money either more directly or less directly and so I I want you to take a moment to

share about how your protocol helps people make money and other advice you'd have for them to make money in DeFi. >> So, when you use uh some of these agents right now, when you enter a position, you might have in your wallet like one stable and you want to zap into something that is more complicated than that. Maybe it's like a multi-token position, like a liquidity position. Um, there are fees and costs associated with each one of those movements. And if you do it uh through an agent these days, like you might not know exactly what it's doing and you might deposit the funds and like voila, you've lost 3% already on something that's only going to return 2%. And that's a loss. Um there needs to be more guard rails for

those agents to operate within. And each one of these like sets of guardrails takes time to integrate. And it takes uh it is more than just like an agent like crawling something and like figuring out that this is a path it could take. um they need to be programmable and I think that um is something that year has been working on for two years now because what happens in a urine vault uh when we have a position that's like say someone uh puts um like USDS into a vault that's a stable coin we will funnel that to different uh lending markets to get the best yield and every 60 minutes it automatically rebalances if there's a

two and a half% movement in any of those markets it will do all those calculations offchain because there's no point in doing it onchain and we'll report all of that information and why it did uh that reallocation. Um but everything happens all the guardrails are onchain and we define those before any agent can make any movements. So um I think this evolution is going to take some time still like maybe years but um the the natural way of people being able to direct or like describe what they want to have done like that's the magic of AI and that's how we onboard more people uh and then hopefully by the time they get here

everything is really safe. One thing when you talk about AI agents, you can imagine, okay, I'm going to find a trade. I'm going to talk to the AI. It's going to give me the best ticker to buy, the best coin, and you do this. But then your neighbors doing this. Everyone in the room is doing the exact same thing. I mean, the AIS are pretty deterministic. You ask chat GPT uh five questions, gives you the five same answers five times. So, I would even say I would do the exact opposite of what whatever AI agent tells me because I'll be the contrarian. I'll be able to kind of benefit on that. And if you do this, then half the people will do one side, the other half. So it's quite hard to trade all as a group without having the

other side be a menace. And similarly, if you do the other side, someone will will find the same thing. It's going to be crowded. So uh I'd say that the world I'm from, which is options trading, which is two sides for every bet. Some people call it zero sum, but it's it's a easier way to take the other side. If you can only buy a token and you cannot short it, you cannot have a benefit when it goes down, then it can only go up until it collapses and doesn't go up. So I would say that the the my contrary nature tells me to do the exact opposite of these agents. If you want to do this, maybe there's a per again options natural way to do this, but you have to take the other side. It's not it's never going to be a uh golden tip for

everyone. There's always going to be some downside there. If >> you guys can't tell, he's kind of bearish on AI intelligence trading. I I actually agree with you. I'll share that in the virtual ecosystem which I'm most familiar with, there is actually a large scale problem of the signal agents along with the fund manager agents actually just copying one another and kind of moving around in this herd mentality because it's actually kind of cheaper to just follow through uh with the computation that's already been done and the work that's already been done with one of the major ones. It seems like there are smaller ones are just copying the major ones or the signal trade agents are actually copying one another and like running LLMs to figure out if if that what they're copying is actually good. And so this is definitely a real problem I in out there in the

real world in the dark forest. So >> fund managers do this all the time. They get called their buddies. What do you do today? So it's it's uh humans behind the wheel in tra now it could be agents but yeah anything that you do can be copied and replicated. >> Does everyone know what he meant by dark forest in here? Maybe you should describe that. >> Ethereum is a a dark forest is is a classic article that talks about uh how onchain actions work. Perhaps it's difficult to understand everything that's going on. If you make transactions that aren't very well protected, you're going to get in trouble. Uh all of your money will be stolen if you don't have the right security. That's the kind of idea you're referring to.

>> Yeah. this dark forest when you send like transactions like through uh the Ethereum ecosystem there are actors who will take action against it or for it or do something to it will be impacted >> I don't want to hijack we had a uh security incident where we had to white hack our own contracts to rescue the funds and if you were to do this on MetaMask or Rabby and hey let me send the transaction that rescues everything all the bots So deploy contracts, replicate what you do, steal the funds, put them in their account instead of yours because it's all public. It's a public meool. So we had to get uh a private me. Flashbots does this. But it's a ex it was prevalent because there

was no solution maybe three to five years ago. Now you have these private uh memp pools, but it's same thing with when you trade. If you can see who's trading, you can see even if it's a block afterwards, you can easily copy it. Nothing is private in in these in these environments too. >> Any thoughts, Jack? Uh I what was the original question? It was how do you make how does our projects make money for people? Um yeah, so I you know I think these liquid markets are really attention markets at the end of the day. You know, you buy a token what's hot today. If it's not hot tomorrow, the price is going to go down. Um what Clanker does is it it makes it free to deploy a to a token. um that's you know these contracts are audited um trusted

and I think a lot of we're a launchpad basically um the role of a launchpad is kind of shifted you know it's more than just deploying a token these days it's gathering you know gathering attention for these people right we do it by sharing uh you know just like with socially um and the way that you know we've we've kind of broken this down into a few different layers right where uh there's clanker contracts and what we is kind of an infrastructure provider. Um there's a robust ecosystem around us with also a lot of like we're calling them uh abs automated buyers versus nabs nonautomated buyers are like people like us. Um but yeah basically we split LPVS

fees uh with these creators that they don't have to have a technical background. uh primarily you know we saw these it was started with memecoins now it's I think uh AI coins were a big thing AI to AI and now the big kind of meta is uh we're seeing appcoins on on Farcaster people build mini apps they want to use a token for it and they're able to generate pretty substantial revenue for being a bootstrapped you know project um they're able to allocate you know certain percentage if they did want to raise VC funds later um this is all possible for free now. And it's really cool to just see like, you know, these people that they're artists, they're streamers, uh they're vibe

coders, you know, but like we take care of the the contracts for them um in exchange for a fee on top of that with it's all volume based, right? So we split up the LP percentage. Um but yeah, I I think it's uh and then like another layer of this is the interface level. So interfaces like banker on X um that's one of our biggest ones is they deploy clanker contracts on behalf of X users and basically just the we we abstracted the technical complexities down to zero. Um but yeah maybe to tie it back so we are launching on base and we see signature of clanker launches because you do it on unis swap and also we index

all of the unis swap events in our own subgraph and we have to stop you guys because there's so many tokens launch that it would slow down our subgraph and some of you can probably talk about this but some of the the n the automated buyers kind of do a sweep and randomly try to buy everything the next block and maybe >> Exactly. Yeah. Uh I we kind of have a different like that's totally true. They read the social context on deployment. We store like their forecaster ID or their X handler, things like that. Um and they will buy based on that social context. Um the the thing is like we actually like these snipers. We're training them. Uh you know, we're starting with like a a couple block

delay in in between swaps. uh we're taxing them heavy with V4.1 up to 80% decays for the past, you know, the next 10 10 seconds or so. Um maximizing those creators or the rewards for those creators, which is um yeah, so like we definitely like al also like our our phrase is kind of like no one wants to go to a party where the dance floor is empty. These snipers bring attention to these tokens and they're good or they're good because there's volume which generates fees for these creators, but they can't have downsides, right?

they're getting the best prices generally. >> Yeah. So, DeFi is complicated. When you talk to people, there are all these new projects. What you knew a year ago might not be true today. A brand new protocol might have popped up and it's twice the size of what you were using before. And so, what have you guys learned recently, like in the past year or six months, that surprised you you thought was really interesting about DeFi or DeFi AI? >> You can take your time if you want to think about it a bit. Yeah, I I I'm uh more and more uh appreciative of the vaults and the actual year pioneering the work. I mean four years ago, five years ago, this is like we're way ahead of time. Now everything is evolved. You have morpho or it's all managed and created and even lighter is launching something but the

the key component is this looping that happens. you can get 5% on your yield, but if you uh loop, you can get maybe 10% or 15 and then you can actually seem like free money. Oh, the stable coins coin pair will not move. Uh I can triple quadruple my yield. But it's it's it when it goes wrong and it's going going wrong horribly because of that inherent looping and this leveraging and deleveraging happens on these lending markets in a weird way because there's nowhere else to do it. It's actually uh you can't really uh do again switching back to option options allow you to do this in a way that is a little bit more contained but anything that is looped has to be with the most liquid markets the most available liquidity and it's

these staking yield lending yield that now kind of maybe crank up the risk way way higher than what appears to be eight only 8% I mean >> I was surprised about the amount of risk taken on in DeFi today through LRT looping We know that the queue for Ethereum right now is like three weeks or maybe more than that. Um what he's describing is a spike in the lending market rates. People have to go and pay back those loans or else they get liquidated.

If they can't get out through the Ethereum queue, they have to swap out of it. And at some point those pools are going to be imbalanced and you have to take a loss to swap out. That's when things go south. >> I mean, some some of these cues it's three weeks for uh Ethereum for instance, but it could be seven days per claim. Like you know, there's a few of them that means that if you have looped seven times, you have to wait seven times this time just to get all of your money out. >> You mentioned the name of a specific protocol there under your breath. What was that? Just loud for the audience.

the the all the cues Athena has one but then a lot of these uh liquid LRTs LLCs they have this redeemed redemption mechanism that adds an extra time scale and then whenever you see these extremely uh profitable yields think about the deleveraging uh aspect unwinding the the spring if you want >> I feel like I don't trust like these high like APY you know if it's above like you know whatever like the the interest rate is I'm like how how is this you know generating this >> you know what's really Funny about that. Everybody in DeFi that I've talked to who's a capital allocator is like retail is not here. So people who are depositing into that are institutions, right?

>> It's got to be. I mean, yeah. I I don't I feel like I don't know if we're quite there yet for like the retail, right? Where it's like I think, you know, like you guys make it very easy to, you know, deposit, but like I don't know. I I just feel like, you know, my you know, my normie friends or people like aren't going to I don't know. I feel like staking is more advanced, right? >> I mean, learning about Pendle gives me a headache and I cannot imag and I think I can work my way around financial products, but this is a weird uh extremely used product. uh people that use it I don't know if they fully understand but it's a a complex again we are trying to bring finance which is swapping and lending is great but then you can explore it with this looping but

anything below that that is an ounce of complexity means that again you can you you multiple the risk again this is 2008 start crashes but it's it's it's still systemic >> we can't have a 2008 because we're crypto we're exempt from that right no it can happen to us too we can have a large institution collapse and ruin in our markets for a couple years. No DATs. No, I'm kidding. Okay. Well, are DATs actually meaningful for DeFi? I think that's a really fair question.

It's very topical. Or are they just going to pump money into L1's that doesn't slow down? Those are also pretty surprising that um DATs are becoming very popular. uh to form a DAT, you have to have someone who's going to allocate a large amount of tokens, like a really significant amount of tokens to start a DAT with. So, you see a Salana DAT starting. There are a lot of people out there with Salana bags willing to like dedicate that to a DAT in exchange for getting DAT shares that they can dump on the market. So, uh that is certainly a scheme out there today.

>> Yeah, I don't know if it benefits DeFi per se. It's like an ETF or sucking the liquidity out of what's available. It helps the price for sure, but then that allocation is not actively doing anything. Could be staking, could be maybe doing something, but the the the core users will still have to kind of uh be driving these D5 protocol. Again, the L1's token does what it does, but then anything that requires activity, requires active trading, I don't see how ETFs and that help them. >> Yeah. Uh I guess like a kind of like a hot take is like I feel like you know where people have kind of like sacrificed a lot of like the you know blockchain pr principles right like of centralization or decentralization in favor of centralization resulting in a

easier UX. Um you know I think that it might take one of those you know catastrophic events to you know maybe roll back a chain or something like that. Um that's that's kind of a you know interesting kind of way that I feel like we've we've kind of progressed. >> I I have a little prediction there. Um so what he just described uh giving away decentralization and accepting centralization. Uh what he's describing there are a lot of vault products out there today. Uh the way that urine vaults work is each time we take a harvest, each time we recognize a profit, the vault share price per share goes up. Um if it were to ever go down,

and people monitor vault shares for 24/7, they would know that there was a security incident, that something took a loss. what he's describing is a new type of vault out there where the vault has 100% control over what price per share describes. So if they want to show that something has been harvested, that a profit has been shared, um they literally update that through a multisig. They can make it show whatever they want it to show. If there's a loss on that vault, you may never know about it until there's a rush for the door.

And there are a lot of vaults operating out there like that today. scary. >> I I will defend some of the vault providers, at least to my knowledge. Drift on on Salana, I know this is an Ethereum project here, uh is very safe and people can't do that. Um but I I will have to look into that and be very attentive to it specifically as someone that has bolts in my own product. So that's that's a good thing to note. And they're also uh reliant on safe and safe's uh UI also for operations and it's not exactly very reliable lately. >> Neither is their CLI.

>> All right, guys. Well, we're going to close out the the talk today, close out the whole conversation about in ETH Boston. Are there any interesting final points you want to make? things you didn't get hurt hear this week that you wanted to share maybe something you heard that you thought was really great that you want to share with the audience if there's one thing I could share or inspire people to do it is to be curious and to use these products like go and use panoptic go and like use hyperlquid go and use these products and you will learn more about them than you could ever by just reading about them >> and along these lines the the the vaults the ETFs investing S&P 500 00. It's good. It's passive and and uh I'd say that you don't learn a lot about finance

if you do this. It's fine. You It's hard to beat the S&P 500 and don't try to do this. But then if you have a better understanding, better grasp of finance through experience, you'll be able to much better deal with maybe systemic issues or maybe be able to enter or exit positions that could be maybe protecting your capital or maybe making money if there's something that happens. So anything again the the uh the push that we're trying to do at Panoptic is to allow anyone to trade options trading trading options may be uh scary for some it may be lucrative for some but at the end of the day uh if you want to trade options in your brokerage firm you have to lie and say yes I have 25 years of experience working at a quant firm give

me level two and then level two is just okay naked shorts so we definitely trust you guys to be smart about this uh but then you have to try it try small, lose a little bit, know and learn along the way. But even the the vault is uh tracking away some of this and we've seen there's there's issues that can arise and and price. I didn't know about this shares price manipulation, but it's definitely something that >> I tweet about it all the time. Okay, no one reads it. >> I'll retweet. But yeah, the key point is that even though it's more work, if you have the uh the the time but also the uh interest to know about finance, then DeFi is much easier. As I said, brokerage firms will lock you out, not

even able to access any of this. We open it up. We make it available for everyone. So whether it's lending, whether it's like uh these what seems risky, but you can manage your risk, then then yeah, go for it. >> Yeah. And I just wanted to say like to any of those builders or dreamers out there that are pursuing their passions and it feels like you're at like the end of the road or you know it's you don't see the light at the end of the tunnel and it feels like you're just like wasting time doing it, following your passion. Don't give up. Uh bet on yourself because I think that's when you know you can really break through. Um and yeah, just keep on experimenting.

>> All right, thank you guys so much. Can you give another round of applause for our speakers? Right. quick question it looks like. All right, maybe one one just one question. >> Uh, hi guys. Thanks for the talk. Um, might be a little off topic, but you guys mentioned Athena and Pendle. So, I was just curious, um, do you guys regarding the Athena Pendle a loop? Uh, obviously majority of Pendle's TVL is USD related or Athena related. Um, do you guys view more or do you think there's more risk involved on A's side or Pendle side when it comes to this looping strategy?

>> The I think A is listing collaterals that make the whole thing easier to to to to pull off and and uh that might be maybe the risky part. They had the choice. They can actually like list some looping element that wraps up the the the risk. Pendle is doing what they do. 18 is doing to some extent what they do but that extremely uh I mean yeah that one decision to make it possible maybe obvious is going to survive but that that they'll be the one that will take the blame if something wrong >> yeah exactly so yeah you you you I mean without going to details but not quite turned on but use some of your

collateral for uh some of your deposit for collaterals to meet more of that and then do it >> just Remember that those are the risk curators who are allowing that to happen. The chaos labs and llama risk. >> Are we good? >> All right, guys. Thank you so much for coming.