Fireside Chat: Dragonfly's Haseeb Qureshi | Andy C - The Rollup and Haseeb Qureshi - Dragonfly

Hello, we are back again. I am Kelly Lavali Hunt. I have a few guys coming on stage. Softspoken Andy is a aka a rapper

apparently.

Apparently. Just kidding. I'm not going to put him on the spot. And this is Hib. Uh there you go.

Thanks, Kelly.

Thank you so much.

Really, really.

You are. Can we get a round of applause? Can we get a round of applause, please? Thank you. Thank you.

Waiting for that.

Crush that intro. Well done.

Highlight of the evening.

Here we go, Steve.

Great. How's it going, man?

Little jetlagged from consensus, huh?

Okay. Yeah, a little bit. We both We both were in Hong Kong last week. Uh, I will say it was a little less sleepy than Denver. Denver, it seems like half you guys are asleep.

How I see somebody's literally got a blanket. Somebody literally has a blanket here. This is insane. Um I also don't know why they do this in like February in Denver. I feel like it's kind of an insane time to do a conference.

Yeah, it's cool. But it's nice. Um good good skiing this week. Tons of snow.

Is that right?

Uh my voice is a little bit a little bit raspy.

Yeah. Yeah. What were you doing last night?

Meow Wolf.

Meow Wolf at night.

Yes.

Okay. Wow. Is it really loud in Meow Wolf?

No, it's just a trippy spot.

A trippy spot, he says. Okay, got it. Well, congrats on the big uh raise for fund four.

Fund four. Yes.

$650 million.

Thank you. Thank you very much.

What do you do with all that money?

Uh God, there's so many things we could do with it. Really, it's uh first first thing you do is you just load it up into a bathtub just for the photo op. Uh no, what do we do with the money? We invest in a crypto company. We're a VC.

Those of you who don't know, we're a VC firm and uh we raised 650 million $650 million and uh we're going to use it to invest into new startups. So, a lot of people in the space think that basically like crypto's cooked. They think the industry is dead. They think that nobody wants anything anymore. And I'm here to tell you guys that's [ __ ] It's not true.

Um, I have never seen I've been in this industry for almost 10 years now. I've never seen so much of a gap between the institutional sentiment and the retail sentiment. Like, institutions are they're [ __ ] jazzed. They are so excited about what's going on. They love stable coins and payments and blockchains and tokenization and DeFi.

They think it's all super cool. And retail is like it's all over and it's never coming back. So we I remember when when we posted um on Twitter that like hey we you know we raised this new fund.

A bunch of people on Twitter were like wait how how do you who gave you money?

Okay.

And the answer is that institutions gave us money is institutions. Institutions like crypto. They still like it.

Same ones that you've been working with previously. I know you can't some new ones. So, we have a lot of really you I can't say their names because we have to be um respectful of their confidentiality, but um a lot of institutions you guys have heard of are bullish on crypto and they want to allocate more money in the space because they see that real stuff's happening, real adoption is happening.

So, we've not really seen too many mega funds in crypto. You know, I think Panta had a a1 $1 billion fund on A16Z as well, but you know, uh we've not seen like five billion,10 billion raises. And I think when people saw your raise, they were wondering, okay, so how do you actually strategically allocate this much capital into the current environment, right? We're in a low liquidity market. It's a bare market at least, you know, price time wise.

How do you actually allocate capital at like in size in the current environment? How do you think about that? Is it early stage series A? Uh talk to us about kind of like the allocation strategy and how that pertains to the state of kind of early stage.

So, so look, we invest anywhere from like 3 million on the low side to like 30 40 million on the high side. So, that means that you're going from seed, series A, series B, and potentially even, you know, a few things as you're following on to those investments. So some of the companies that we've invested into recently, you know, things like Poly Market, um things like Rain, which is a, you know, crypto stable coin card company. Um you know, things like uh uh we recently led around into Mesh where big investors in Athena. These are all really at scale projects.

They have significant revenue. They can absorb a lot of capital. Um and there's there's more and more companies that are coming to light that are really pushing the Vanguard forward. So, I don't know if people saw the recent announcement of this project called Conway kind of went went super viral, got Vitalic talking about, oh, hey, this is bad for crypto. Um, the intersection of crypto and AI is just starting to blossom.

You can start to see that these models are just starting to get good enough that they can actually manage money and start to do things that are beyond just, okay, we have a memecoin and we threw a we have a chatbot, we threw a memecoin on it, right? Which was a lot of what crypto AI meant last year or the year before. Um, now we're starting to see more of a reality building there and I think it's going to be very rife for potential investments. Um, but then a lot of what we invest into is just, you know, the kind of brickandmortar stuff of investing into DeFi. You know, it's kind at this point it's kind of boring to talk about, but DeFi is still one of the big things in crypto that works and there's still a lot of new stuff to get built there.

Um, same thing with, you know, payments, stable coins, like all this stuff that's just like the kind of blocking and tackling of making the industry do the next 10x. Yeah.

Yeah. So, you know, we've we've kind of coined this term neo finance for this kind of payments, stable coins, rwas, kind of encapsulating this. It's almost like the whole uh what is it? Uh DeFi in the back, CI in the back. Yeah.

Right. That's right. But kind of extrapolating on that, we've been in a very chain centric world, I think, and this is getting a bit more deep into like some of the thesis. Uh what do you think about the transition from like a chain ccentric world to a more asset ccentric world as we onboard more more tokenized assets as we as stable coins grow? Uh we're starting to kind of think that assets are going to be kind of are going to go through the process that early stage chains did.

You need a DEX, you need a lending aggregator, you need a vault, you need your um NFT, whatever, right? assets, you know, let's say we're tokenizing a private credit or a CLLO. They need a deck to swap, securitize, and unis swap. Uh Apollo and Morpho, they want to they want to lend their T bills. Then they want to do a vault.

Like just talk about kind of the the the more assetentric view. Yeah, that's that's an interesting way to characterize it. I'd almost do it the other way around which is that what you see increasingly from chains is that it used to be yeah you need the checklist of you need unis swap you need a you need this you need that now increasingly it's like oh you need bidd

and you need to pay up to get bidd to come launch on your chain because if you have some you know new ghost chain bidd doesn't show up for free right you have to go you have to go pay them to incentivize them to go and launch there same thing with USC you want to get native USC you got to go pay up to get the assets on your chain as opposed to just the protocols on your chain um I I think that um so like all this will continue and I I think what you also see is that there's a lot more there's a lot less churn in chains and infrastructure right it it basically what I think we might be on the last or next to last cohort of new infra um in the sense that there's just not that much room to compete now in creating the next layer 1 the next layer 2 unless you are bringing your own distribution. If you're trying to take someone else's distribution, it might just be too late. In the same way that for social media, there was a critical period where almost all the social media companies got built. Basically, by 2010, everything you think of as social media company was already built, right? So, Facebook, LinkedIn, you know, Snapchat, like all that stuff was already existing.

It was small, maybe much smaller than it is today, but it was already there. The only new one really was Tik Tok, right? That came up after 2010. So it it may be the same thing for blockchains themselves is that basically every important blockchain is now there. And that doesn't necessarily mean that there's not going to be more shifting.

It may well be that Ethereum gains market share, loses market share, Salana gains, loses, somebody else starts to gain and lose. But the ability to start today and to be able to gain market share, you have to be really extraordinary in the way that you know Tik Tok was able to really push the Vanguard forward. I think the same thing is likely going to be true for infrastructure. Now the the last thing I'll say is that it may well be that the next big shift that we see comes from the advent of AI as an intermediary between the user and the application right um so let's take for example the clarity act clarity act right now it looks like on poly market it's like 78% as of this morning for the clarity act to pass this year um clarity act passes there are going to be potential um guardrails and or uh requirements for front to do KYC to do you know some transaction monitoring stuff like this. Now, one way like the paradigm here is that like well, you know, that's going to work because you have to go to unis swap.

com to go and trade on unis swap, right? And if you go there, then you got to do this KYC and blah blah blah. But you can imagine a world where people tell each other like, hey, if you want to go swap on unis swap, don't go to unis swap.com, just go type it into claude that you want to trade on unis swap and claude will just like oneshot a unis swap interface for you. And when you do that, when that's your interface, when when the AI is your interface and not the website, then yeah, you don't have to KYC if you're just like creating your own vibecoded front end, right?

Or or even less, if you say you don't even have a front end and you just tell Claude, yo, go buy me this thing on on Uniswap and it just does it for you all within text. It in that case, okay, one, now the interfaces don't matter, right? The interfaces are now completely cut out of the story of how people interact with smart contracts. Okay. Second second thing that does is it changes the way that you do marketing.

It changes the way that you do competition. You're now no longer telling your model use unis swap, right? But you're just telling the model your intent. Your intent is trade this token for that. You don't give a [ __ ] how it gets done.

So your model very well may just say look I'm going to go try 1 in. I'm going to go try unis swap. I'm going to go try cow swap. I'm going to go try something else. See what gets me the best price and that's how I'll execute.

And it on a certain level doesn't even need to ask you like why would you know? You're an idiot. You know I'm I'm an AI. I can go check everything instantaneously. So the extent of like brand loyalty lock in like all this stuff changes in a post AAI world.

So I think I mean this is true on the consumer side to some degree and that's part of the reason why you're seeing the SAS apocalypse in the uh in the NASDAQ. But in the crypto world because crypto has no lock in crypto has no user accounts. Crypto has no like there's nothing keeping you in a walled garden in crypto, right? And we always talked about how that changes the the competitive dynamics in crypto, but then we never really saw it, right? It kind of didn't really happen.

UniS swap is insanely sticky and we don't really have a good explanation for why other than that people just they know unis swap and so they go there. But in a world where AI agents are sitting in front of you, the way in which protocols are going to compete changes

and that's that's interesting. I don't know exactly where that goes. I don't know how that happens but that I think is really going to change the dynamics of how protocols work and interact with each other in the next call three to four years.

Couple side quests here. Have you used open claw yet?

I have used openclaw. Yeah.

What do you what do you use it for?

Uh not much. I mean it's it's obviously not super secure. So it's not something you want to be

a ton and Yeah.

Well, it's like I you not like have I don't have like a self-driving Gmail that I'm hooked up my open call. I think that's insane. But uh I right now I just you know it's like a little Telegram bot you know like it can has some read access to some stuff that's pretty pretty low um security for me and I just kind of play around with it see what it can do. Um the reality is that it's it's something that I think people are going to like I think here's what's going to happen is that there's going to be kind of a two-track world for agents. The first track is going to be the big labs, right?

So obviously OpenAI acquired not acquired but hired the OpenClaw founder and is is backing the OpenClaw Foundation and um you know Sam Alman said that OpenClaw represents a very important path of development for OpenAI itself, right? Like they see agents and agents autonomously doing things on your behalf is going to become more and more important. That being said, OpenAI is never going to run a YOLO run your Gmail, you know, use a credit card without asking you agent ever. They're not going to do that because it's insane the amount of liability they would get. You just imagine the stories of somebody wakes up and like they're fired because their open claw did some crazy [ __ ] up [ __ ] to like email their boss like something insane or like spent all their money and they wake up and they're thousands of dollars in like Amazon bills like that.

Like that's not going to happen. Open is never going to do that. So there's going to be this second track which is the crazy hobbyist

like the wild west type of thing.

Exactly. Summer 2020.

That's the side. That's the DeFi side, right? Coinbase doesn't allow you to trade everything, but you go on unis swap and you go buy like fartcoin, all these crazy rugpull stuff and like you just, you know, buyer beware, brain damage is on you.

And that is where all the real crazy innovation is going to happen of people who have like these quotequote self-driving companies, right? Where the company just runs itself, emails itself, does you know that is where you're going to see incredible amounts of innovation and the vast majority of people are going to have no idea that it's going on. Like they're they're not going to touch it. They're not going to be aware of it. So, you put out a pretty good piece uh as of yesterday, I think it was crypto was not meant for humans.

Yes.

It's meant for agents. And we've kind of

orbited around the core of it here. Maybe just give us the core the core idea.

Um why was this technology built for this next era? Are we not going to be using it in the same way? How you know, how's this going to change? Well, so, okay, I've already sketched out a little bit of what that looks like when your agent is doing things on your behalf and it's the one interacting with the blockchain, not you.

Um, there were a lot of stories that people used to tell about crypto. Like, one of the stories I used to tell was that people are going to use smart contracts instead of legal contracts,

right? I first when I first came in the industry 2017, people talked about that all the time, right? And here we are 10 years later and no one's doing that. No one's doing that. Like, I'm a VC fund.

I'm lit I'm a crypto VC fund. I back crypto startups. We both know how to read smart contracts. We're using legal contracts, right? Why?

The answer is that it's easier for us to understand the failure modes and the and the happy case for a legal contract than it is for a smart contract. Even for us, that said, for an AI agent, it's the other way around.

Okay? For an AI agent, they can really trivially understand code. They understand code way better than they understand contracts. Right? You read a legal contract.

If you're an agent, you're like, "Well, I know what this says, but like where is this going to get adjudicated? What what country? What jurisdiction? What circuit? Uh, who's the judge going to be?

Who's the jury going to be?" It's totally random. You cannot predict these things. It's by design that you can't predict these things. Whereas in code, you know exactly how the code is going to get adjudicated every single time.

And so, if you're an AI agent and I'm talking to another AI agent, if I'm entering into a legal agreement with them, then okay, how am I going to drag you to court? How do I how do I know where you are? How do I know like how any of that's going to happen? I have no clue. But if we have a smart contract, I know exactly how to enforce that against you, right?

Because it it is self- enforcing. So, um, and you add to that all the foot guns that exist within crypto, right, of like, you know, long unreadable cryptographic addresses, uh, approvals, you know, like all all the subtle way, you know, you have to read the exactly address poisoning attacks or, oh, the URL was one little character wrong. Like, humans suck at all of that,

right? And like I've never been afraid of accidentally transferring money in my bank to North Korea, but I am afraid of doing that with my crypto wallet, right? And like I've never stopped being afraid of that. And that means that look, the reason why is because banks were designed for humans. Crypto was not designed for humans.

Crypto was designed for autistic software engineers who were existed in 2013 when they dreamed up all this stuff and they were like, "Oh yeah, obviously everyone's going to read the code and understand what the smart contract says, right?" And like, "No, that's not what we're doing. We're not doing that." But AI agents are more like autistic software engineers than they're like the rest of us. And so that's why in a sense, this is exactly what the cipher punks thought they were building for.

they they end up building the exact right product for a user that didn't exist yet, but that user is an AI agent. So in that in that world where the AI agent is that canonical user, I I think many things about crypto that are bugs today become features. Many things that make crypto hard actually make crypto easy. It's a lot easier for an AI agent. Let's say let's say we talk about this, you know, two paths that agents are going to use.

One is going to be okay. Open AAI wants to u be your personal butler and it's going to say hey u you're running low on toilet paper cuz you know we have some whatever you know smart house wired up thing that can tell your toilet paper is low. I'm going to buy some toilet paper for you. Click this button to authorize it with your Visa card. You have to click the button.

They are not going to do it for you because if they do it for you, you'll wake up and you're just like, "Hey, why did you buy like thousands of rolls of toilet paper or you know whatever?" You're going to run a charge back. Visa is going to be like, "Yo, you author, you know, like or you authorized this or you didn't authorize this." If you didn't authorize this, Vis is going to be like, we're not going to let you just yolo things while you're asleep and then claim that it wasn't you. You know, but in cryptoland, crypto is designed for Yeah.

no takebacks, right? There's no chargebacks in crypto. So, if you're asleep and your AI agent is just doing stuff, crypto is the perfect substrate for that. It just automatically works in this world where buyer beware, if you run into a wall or smack smack something in your face, you you eat it. It's on you.

And that is the exact right way for futurists and people who are taking a lot of risk. It's the exact right apparatus for them. And that's why I claim that AI agents are going to be one of these places where you see crypto get used more and more and more.

Um guys, does anybody have a question?

You can just stand.

Yeah. Basically, I was thinking a lot

is validators, no transparency in case they will have some servability able to buy their hardware wherever they

Yeah. Yeah. Okay. So, great question. So, the question is why won't AI agents invent their own money?

I think the answer is very obvious. AI agents are smart and they care about the same things that we care about, which is that they want to accomplish their goals. And the approximate way to accomplish your goal is to use the currency that the most people accept, that's the most liquid, that's the most transferable. And so I think somebody asked me like, you know, will AI agents use stable coins? Because stable coins are centrally controlled.

You know, Circle can turn them off like blah blah blah. And the answer was like, look, AI agents are not stupid. They're not stupid. They're they're very practical, right? Like go talk to an a agent and go see how it's very practical and it will just tell you like, yeah, I'll use whatever is most useful.

So a agents will use credit cards if you let them. If you just if you give your open call a credit card, it'll just use a credit card. Um I think Visa will eventually not allow you to do that for obvious reasons, but right now you can just do it. Um so I think the answer is that AI agents will use whatever everyone else uses. They want to use the thing that's most accepted and most transferable and most liquid.

Most likely that's going to be the the same crypto assets you and I use.

Uh ladies and gentlemen, um I run a podcast called The Rollup. We are on YouTube, X, Twitter. You got to follow us. Follow Hib on X Dragonfly. Thank you guys for watching, for being here, and uh let's give a round of applause for

and let's and let's wish this guy a good sleep.

Wake up, buddy.

Yeah. Yeah. Sleep well, man.