Unchained - How Nansen’s New Trading Agent Makes It Easier to Follow the Smart Money Onchain

Episode Date: January 21, 2026

Thank you to our sponsor, Walrus! Crypto intelligence platform Nansen has rolled out an AI trading agent, aiming to let users complete the full trading lifecycle—from discovery to execution—within a single platform. But the move also heralds a new horizon for trading and investment, one potentially filled with interesting possibilities. In this episode of Unchained Nansen CEO Alex Svanevik unpacks what the AI agent does, the end game, the opportunities and the hurdles. Can Nansen AI do for trading what Anthropic's Claude is doing for coding? And will humans soon be able to work without being glued to a screen? Listen to find out! Guests: Alex Svanevik, Co-founder and CEO at Nansen Links: How the x402 Standard Is Enabling AI Agents to Pay Each Other GOAT: How AI Agents Talking Turned Into a $268 Million Memecoin ‘Religion’ Learn more about your ad choices. Visit megaphone.fm/adchoices

In the last month after we launched our beta, I've been trading more than I've ever traded in terms of just activity. And I think it's reflective of that process being lubricated by AI because it's just so simple for me to just say, hey, act on this now. Hey, everyone. Welcome to Unchained, your no hype resource for all things crypto. I'm your host, Laura Shin. Thanks for joining this live stream. Before we get started, a quick reminder, nothing you hear on Unchained is investment advice. This show is for informational and entertainment purposes only, and my guest and I may hold assets discussed in the show. For more disclosures, visit UnchainedCrypto.com. Before we start, I wanted to introduce you while to our sponsor, Walrus,

a project we actually use here at Unchained for data storage. Throughout the episode, you'll hear short excerpts from Rebecca Simons, Managing Executive at the Walrus Foundation. We'll start with this clip on why Walrus is good for large data files. If you look at most apps today, they depend on quite a complex mesh of different infrastructure, a lot of which is centralized. Walrus is a decentralized data platform. It's particularly good with large unstructured data files,

and it allows you to store and use those without dependency on any centralized systems. It works really well as part of the SWE stack. It was created by Mistin Labs, who are also the originators of that. sui stack and what that means is natively together. Bet, not activated. The score bet app here with trusted stats and real-time sports news. Yeah, hey, who should I take in the Boston game? Well, statistically speaking.

Nah, no more statistically speaking. I want hot takes. I want knee-jerk reactions. That's not really what I do. Is that because you don't have any knees? Or? The score bet. Trusted sports content, seamless sports betting.

Download today. 19 plus, Ontario only. If you have questions or concerns about your gambling or the gambling of someone close to you, please go to conixontario.ca. They allow developers to build with trust, ownership, privacy, baked in right from the beginning. And what this does is it allows you to build use cases that monetize data in ways that just have not been possible before. So there are whole new revenue streams that are now available to builders to come and build on Morris. Today's guest is Alex Svanavik, co-founder and CEO Nansen. Welcome, Alex.

Thank you, Laura. Good to be here. I'm a big fan of your work. Yeah, nice to chat with you. I'm just realizing I don't think I've ever had you on the show, which is crazy. But we have a really great opportunity to chat with you, which is that Nansen has big news coming out today. You're launching a new AI trading agent. Explain what it does.

Yes. Yeah. So some people might know Nansen for the work we've done over the years on on-chain analytics. So basically a lot of investors and traders use our product to discover tokens, perform due diligence on them. But then at the end of the day, they historically have gone to another product to actually execute the trade. And so what we've done is that we've basically taken this function in-house, which means that you can do your whole on-chain investing in-house. within the same product, discovery, research, and then ultimately execution in one product. And the thing that's kind of unique is that we offer an agentic way to do this.

So, you know, we thought what does investing look like in 2030? And instead of trying to build kind of yet another trading product that looks like all the other ones, you can basically talk to Nansen and you can say, hey, I want to know what smart money is doing. You can get the tokens that they've been buying over the last hour or 24 hours or any time frame you want because all the data is in real time. And then you can say, hey, this number two on that list looks interesting. Why don't we put $100 into that? And then the agent literally based on what you said, goes out, checks multiple decks aggregators, and gives you the best price, and then you can tap execute, and the execution app is on chain. So it's a pretty magical

user experience, in my humble opinion. And the thinking is that we're basically creating a new way to trade that's agentic, on chain, and integrated in one product. Amazing. So what are, like, give me kind of the range of things that we could ask it to do. Oh yeah, that's a good question. So in the beginning, we're trying to focus on kind of the co-pilot user experience, where you basically just use it as almost either co-pilot or concierge to execute the trade. But over time, you're going to be able to do more interesting things, like prompting a strategy

or asking it to rebalance your portfolio. Actually, it can kind of do that now. I was demoing it last night. an event to someone where I said, hey, just rebalance my portfolio, like pretty open-ended. It doesn't really know how to rebalance it. It gave me like five different options. And then I said, okay, number two looks good, which was to sell off some of the smaller positions I had.

And it basically triggered five different transactions for me. And then I just manually clicked execute on each one of them, which basically streamlines the whole process. So that's what you can do today. We're trying to keep it relatively simple in the beginning because you want to make sure that the agent doesn't hallucinate. It has really great execution. It doesn't get confused. So that's why we want to really nail that first use case where you just ask it to do kind of simple transactions.

You can do a little bit more than that. But the thinking is to kind of open Pandora's box and start trading through the agent rather than kind of a conventional graphical user interface. Okay, yeah. I mean, what's interesting is, you know, in a lot of wallets, there will be a sort of aggregator aspect and it'll show like it's giving you the best rate or, you know, whatever. But, you know, like a lot of people understand, oh, actually, but if I use this other website, I, you know, I can access the wallet but use a different interface and get an even better rate. So it's almost like you're using this AI agent to just do something that like you would do manual. Yeah, so this is a really good point, actually, because I think a lot of people will basically assume, oh, you know, nonsense is charging some massive fee on top of, you know, what you would otherwise get if you went straight to something like Jupiter or whatnot. But what we're doing is actually charging you the same or even lower fees. And so it's basically 10 basis points if you're a pro subscriber, which is super low, 0.1% in fees, or 25 basis points if you are. a free user. And so this has been in beta now for pro users for about a month.

And now it's launching for free users, which means you're paying 25 basis points. And if you compare that with centralized exchanges or you compare it with other Dex products, it's either the same or better. In other words, we're actually not marking up at all the price that you'd get through the aggregator. So the thinking is like if you're already using Nansen to discover what you should do, we just want to make it really convenient for you to do it right in our interface. And if we had a big fee on top of it, that would basically maybe drive you to go somewhere else to do it instead. But the thinking is, let's just have the same or even lower fees right in our user interface.

So yeah, our users, by the way, are probably a little bit more sophisticated than the average investor because we, of course, have a lot of data and analytics. And so people who use our product tend to kind of want to dig a little bit deeper. And so I don't think they would accept, actually, if we had kind of a 95 basis point fee or something crazy like that, which actually, by the way, a lot of other products do charge people,

even if you might not be aware. So yeah, that's kind of the thinking, that you get the best price and you also get super low fees, which means why wouldn't you just execute the transaction in our product? And so it sounds like originally the thinking is this will be used for defy. And then there's this manual component where you're kind of approving what it does. But what's your vision for kind of like the next level, you know, type of activity? And frankly, you know, just here, like hearing AI agent to me was like eventually it'll get to the point where AI agents are transacting with each other.

But yeah, I don't know. Just tell me kind of what you think the evolution of the product will be. Yeah, totally. By the way, so. So in addition to the agenda to go to trade, you can also trade through a more conventional kind of trading terminal UI on web. So I think we acknowledge that a lot of people will still prefer to trade the old way, but over time, more and more people are going to get comfortable trading the new way. And so I think it's similar with like chat GPT, for example. Although it was so incredibly powerful when it launched, it took a while for people to get used

to the thought that you can. actually do a lot through a chat UI, right? And I think at this point, if you use something like Claude code, you know, you're kind of amazed, at least I am like regularly, literally every day on how much it can do. So I think it's somewhat similar with trading, except we haven't been used to trading that way. And so I think we first have to just like crack that open and get people to get used to the idea that you can just log in and send a message to your agent and the agent just does what you want. And so for me, like as a user, our own product, this has been pretty interesting to get used to that behavioral loop. Or now, for me, the most natural thing is to go

into our mobile app, send a message to the agent, and it does what I want. Like, hey, sell all the mean coins to have in my portfolio. And it just like does that for me. And so I think that's where it starts. But I think the longer term vision to your point is basically that a lot of agents can do things autonomously. And then there's kind of a way to get there. I think humans will always want to be in the loop to some extent, although there might be different personas. I think some people actually enjoy trading from an entertainment point of view and they will not want to be taken out of the loop. So the kind of co-pilot user experience is really good for them. Other people, frankly, are like, look, I don't care what this thing does. I just want to put in some money.

And then I expect that it sort of manages the money. I say managing in a loose sense. And so maybe this is where like you probably have friends and family over the years who have been like, hey, I want to put some money into crypto. What should I buy? Right. And so for now, most people, and I have a lot of friends, family in that category, most people would just say like, hey, just buy Bitcoin or just don't buy anything because you're likely going to buy the wrong thing or whatnot. But most people don't want to manage money on behalf of their friends.

So there is clearly like a need for more passive investment that is also maybe more than just buying Bitcoin. And so I think agents could play an interesting role in that sense where you just like you would talk to a friend who has like infinite patience and you know is glued to the screen 24-7 you might want to express hey this is what I want like this is sort of the guardrails of what I want and this is what I expect just like you would talk to again Chachaputee Grock Claude and so on to solve other problems and then you kind of expect that it does stuff under the hood so my view is like it'll

it'll be kind of a I think we'll take steps towards that vision number one, like it's not going to automatically switch overnight where everyone just says these agents are going to trade autonomously. But I think also number two, these user experiences that will get, I think, this year, will maybe be more appealing to other people than the people are trading in crypto today. So if you're super deep in the trenches and you are buying meme coins left and right, then you're glued to the screen, maybe like a different user persona will find more value in an agenetic user interface.

So, yeah, it's a super interesting area. And we're kind of figuring this out as we go. And like because all of the stuff that's happening in AI, it feels like the world is changing literally like at least every week, you know, if not every day. But yeah, overall, it's kind of like going from co-pilot to like autonomous self-driving. And then there's kind of a big space in between those two things and lots of different variations that are going to be appealing to lots of different user personas.

And so Nansen itself covers many chains. And I was wondering, does this agent work on all the chains you guys cover or is it only, you know, EVM chains or yeah, explain that way? Yeah, for now it's Solana and base, but it will launch pretty much all the chains that we, that where users expect to trade. So we're going to roll my graduate. because it's kind of think of it like this. Before you switch on a chain, you need to make sure that you have everything lined up. Everything needs to be super reliable in terms of the information, the data.

You need to make sure that you have super high swap success rate that every transaction that the user tries to make actually lands. And there are a lot of moving parts there. So we'll basically gradually be adding more chains to it. And it's kind of funny because the agent almost, creates what you can think of as like chain abstraction, right? Because if you just tell the agent what you want to do and the agent knows where you have funds, like which portfolio and it can bridge, it can just take care of stuff for you under the hood

instead of this like frankly terrible user experience, which I think I should have mentioned early on, we should acknowledge that the user experience of trading on chain is pretty garbage, right? That is kind of the nature of early stage crypto, if we can call it that. The infrastructure hasn't really been there. We don't have that many products. People often build only for DGents. And so we're kind of, we're used to a bad user experience, hence the bar has been pretty low.

And so what we're trying to do is to raise the bar on the user experience, which means that A lot of the stuff that's typically annoying with on-chain trading just goes away because you abstracted the chain or because you don't have to go to a separate app to sign like a different wallet app or you'd have to have like 17 different tabs open just to like monitor the information whether it's on chain or X or whatever. So like a lot of these problems go away because we've tried to build a really integrated user experience where you just literally download the mobile. and you can start talking to the agent and it guides you through the process.

You send in, you know, $10 to test it out. And now you have an agent that you can just talk to and trade with. So yeah, the user experience is pretty terrible in crypto, right? And we have a long way to go there. But now I think the infrastructure is good enough to build really world-class products and user experiences. So that's what at least what we're trying to do. All right.

So in a moment we're going to learn more about. the types of customers that will use it and how this could impact the market structure of crypto. But first, we'll take another quick word from Rebecca Simons, Managing Executive at the Wellmers Foundation, on why Rawlis works for devs. Before we built Warrus, what we heard a lot from developers was the need for speed, and we headed ourselves. So reads and writes are extremely fast on Warrus. and this means that apps don't lag even with really large files. Privacy was another thing that we heard a lot about. And Waris lets developers encrypt data without primitive called SEAL.

And with that you have full control over who access your data and everything is enforced on chain. And this enables these really incredible use cases that haven't been possible before. Everything from more reliable AI models to data markets where users can monetize their data. So if you put this all together, what this actually means that the developers can finally build apps

and they feel web too fast, but you've got Web 3 level guarantees. So, you know, you kind of talked a little bit before about how you thought your users were more sophisticated. And I wondered if you could kind of elaborate on, you know, are they kind of institutional traders or, you know, what types of activity do you use? see with your products and what types of activity do you expect with this product?

Yeah, I always found it funny because many people think that Dunson is kind of an institutional product first and foremost, but actually most of our users are on-chain investors who are individuals and they're in the trenches. And so typically, so the way I would think of it is kind of on the one hand, you have sort of normies who just bought Bitcoin on Coinbase for the first time ever. That's not typically our core persona. And on the other hand, you have extremely kind of advanced, let's say market makers funds who require everything to be programmatic and so on and so forth. We do have some of them using, or actually a lot of them use our APIs. But I think the core user we've always focused on is kind of the Dgen, the individual investor who

feels that they can have an edge in crypto because frankly they can. Crypto is super fragmented and there's a lot of information inefficiencies. And they tend to kind of look at new narratives and try to figure out what's actually, you know, the best way to play this latest narrative or to manage, you know, their overall more long term on-chain portfolio. But people who trade on-chain very actively is kind of the core user persona. And, you know, most people refer to them as. DGents, that is basically like the core user persona we've always catered to at

Nansen. I can also just say for those people who aren't familiar with Nansen, the way we got started was that we launched our product April 2020 right before something called DFI Summer. And so, and so DFI Summer was an amazing time, probably not for our physical health because we were stuck inside due to COVID and kind of doing all sorts of crazy. things on chain. But there was like a very clear need to figure out what people are doing on chain because there were so many opportunities all the time and you had to navigate, you know,

what the quote-unquote smart money was doing on chain. At the time, smart money meant funds like Alameda and Three Arrows Capital, which turned out to not be so smart after all. But at least a lot of people wanted to figure out what are these funds doing on chain and We basically label addresses that is one of our core strength. So we have like 500 million addresses labeled, which is done mostly algorithmically and through AI agents actually. And some human curation as well. But so you can figure out what smart money is doing.

That was like the inception of the company and how we got started. And so we've sort of evolved that approach. But the core thing has always been the same, which is that we want to surface the signal and create winners. That's the mission of the company. And in a way, even if crypto changes, that part has always been true. That's kind of an evergreen truth of the company, that there are opportunities on chain. If you're smart and if you look at the right things and if you actually make use of the information of what's happening on chain, you can have an edge.

Like that's the whole kind of founding premise of the company. So that's why we've always been used a lot by the DGents. All right. Well, so I'm glad that you dipped into Nansen history because, yeah, we have not had that information on the show. I'm also curious about your history. How did you get into crypto and come to Found Nansen? Yeah. So my own background is actually in AI.

I have a master's degree in AI from University of Edinburgh back in 2010, so 15 years ago, which means I like to say that I was an AI before it was cool. It's true and I can prove it. But yeah, so I worked as a data scientist actually for many years. And I worked with machine learning. I trained models. Funnily enough, one of my teammates back in, I guess, 2016, was Moths, who is the CTO at Dune, Analytics.

And so we worked together. Yeah, he's a great guy. and they've done super well over the years, which I'm very happy about. And so I was a data scientist for many years. I was a data science manager. I worked as a management consultant even, like in industries, like insurance and salmon farming and media and lots of different industries. And then in 2017, I discovered Ethereum.

Some engineers at our company in Barcelona at the time basically taught me about Ethereum. and I got very curious on it because I had ignored Bitcoin for many years before that, which is not great in hindsight, but I was kind of thinking as most other boomers that Bitcoin was for money laundering and nothing else. And then a few years later, I discovered Ethereum and I thought, this is interesting for two reasons. Firstly, I like that it's a platform because that means, people can create things on the platform and if any of those things become popular,

probably the platform is going to accrue some kind of value. That's number one. Number two is that with Bitcoin, I always felt that I don't know if this is the currency that's going to take off, even if the thesis is correct. You don't know if that's the right currency, the right implementation of it or whatever, the right monetary policy and all that stuff. So I thought if someone does figure that out, there's a good chance they will build that currency on Ethereum.

So I think a lot of people had sort of similar thoughts on Ethereum and maybe add to that that this was right at the ICO boom, which created also a lot of opportunities to invest and make money pretty fast and also lose money pretty fast. So I kind of dove right into that. I moved to Hong Kong in early 2018. I joined actually another startup that had done an ICO. that started imploded very quickly because the price of ETH tanked from $1,400 to $80 in 2018. And after that, I realized that the core thing to focus on was on chain data and on chain analytics. And so I took some of the best people from that company and co-founded Nansen sort of building it late 2019. And can I ask which I see what that was?

Yeah, it was called coinfi. It was kind of one of a million ICOs, you know, it never went anywhere. But I will say like some of the, I think like there was a premise of kind of trying to build a great information tool for crypto. But the vision wasn't very clear. And so as an employee and not a co-founder at that company, you know, I tried to speak up and say, hey, this is what we should do. you know, it wasn't well received. So I thought, well, I was just starting my own company and then do these things myself. And it turned out that maybe I was right.

So we touched a little bit on some of this before, but I wanted to tease out more of this information about your new AI agent product. By the way, does it have a name? Yeah. So the mobile app is just Nansen AI. And so that's pretty. straightforward. And on the web, where it's launching very, very soon, it's the same AI agent, which means it has the same capabilities, but for now it is just integrated into our web

product, you know, which you can find at nonson.a.i. So it doesn't have a separate name as such. It's kind of evolving our core product to become AI first. Yeah. Okay. Okay. So, you know, what this type of product does is it gives... retail users access to the or at least similar types of automated decision-making systems that hedge funds have and I wondered how you thought that would impact market structure that's a great question and yeah I do think this is kind of giving retail investors an Iron Man suit which is kind of the ambition right or a Batmobile you know choose your choose your metaphor but I do think it's interesting because

my view is that AI, the benefits of AI are going to accrue to individual investors more. And so if I'm going to unpack that a little bit, I would say, if you think of a hedge fund that is very well resourced, they have, you know, PhDs, analysts, traders, all these different types of profiles, and they've invested a lot in their infrastructure. but the cost to recreate or even create new infrastructure and products tools that are equally good or even better has basically collapsed with air. And so my view is actually that the greatest improvement will be for individual investors. And so that's pretty interesting. And it is a really good question on how it impacts market structure.

I think the good thing about crypto is that crypto is so fragmented that this has always been kind of my thesis about crypto is that there's so much stuff happening in crypto and it's so fragmented that everyone can have alpha in their little corner. And so you know, you and I might have the same tools, but we can use those tools very differently and we can focus the tools on very different portions of the market. And so what I think seems pretty obvious is that just activity is going to go up. If you can more easily do things on chain, which is not just because of AI, but also because the infrastructure is so great, you know, because you have Solana that has super low fees,

super low latency, you have like privy, dynamic turnkey, which gives you great, like, embedded wallet technology, you have aggregators like Li-Fi, OkX, Dex, and Jupiter that give you great best price execution. All this infrastructure is now there. And then you add on top of that AI to help you discover and research and ultimately also execute trades. It just seems to me that that is an incredible setup for individual investors and it's going to make them trade even more. And so I think like it's pretty, seems pretty obvious that activity is going to go up. It's a little bit like vibe coding, right, where, you know, more people are going to code because of vibe coding. Because you have like lowered the threshold for people to code stuff.

And so I think vibe trading, which is one way to think about agentic trading, which is what we offer, also kind of lowers the threshold to like, who can trade. And then it's really important, by the way, that the tools you give them actually are good, right? Because if you lower the threshold and you give someone, it's kind of, have you seen the meme with giving like Claude Code for marketers? It's like someone handing an Uzi to a chimp. That's like one potential way to think about this, right?

Which could be very dangerous, right? And like you need to make sure that the quality is. great. And to be honest, you know, because we operate in the agentic trading category, I've looked at other products. And like, I'm not going to mention any names, but there's a lot of slop out there. And I think a lot of people, frankly, are going to lose money because they're going to use tools that are, you know, half baked that don't really work. And they kind of look good. But under the hood, the agent is like hallucinating, has the wrong data. It picks the wrong token if you want to execute it like this all sorts of things that can go wrong so i think like

yeah how it impacts market structure i think a it's going to increase just activity like it makes it possible for more people to trade number one and number two it lubricates the whole trading um process which means like the people who do trade are going to trade more like for example i actually don't trade that much you know i think of myself more as maybe operating the equivalent of like a liquid token sort of venture fund kind of defines capital run by Arthur is a friend of mine and that kind of style where you don't necessarily trade super frequently but you have like a bet and you want to see that bet play out it could take months could take years whatever but in the last month after we launched our beta I've been trading

more than I've ever traded in terms of just activity and I think it's reflective of that like process being lubricated by AI because it's just so simple for me to just say, hey, act on this now. And I don't have to be like, oh, let me open this page up, go in there, click on that tab. Now I have to sign with my other wallet. Oh, no, I don't have this wallet on my device. Oh, I have to scan this QR code. It's super clunky, right? And if you just have an incredibly lubricated process, a lot of which is done by AI,

then I think people are just going to trade more. So that's kind of, and then I think, you know, you can then work backwards from like which assets are compelling to individual investors. Those assets are not the same necessarily as like institution investors. For example, meme coins and creator coins and, you know, and even NFTs are typical kind of individual investor types of asset classes, if we can call them that. So and then tokenized stocks too, I think is. one of the most obvious bets ever that of course people are going to trade stocks on chain if if and that's a big if the user experience is better than trading it on the venues where you would trade stocks normally which i think it you know it will be and you might argue it's better already now but like it definitely

will be very soon and then you have like prediction markets as well which appeal to individual investors and retail investors and so you know All of these things are just going to have more and more activity, as my view, over the next years. And so, yeah, that's going to. And then you're going to have like the cool thing about crypto, going back to the part about fragmentation, is that you have this long tail of assets. And because you have so many market participants, because we've lowered the threshold, you're going to see so many, you're going to see like a jungle, like a camber and explosion of all sorts of different tokens.

Which you know, you could argue you've already seen to some extent. I saw a funny tweet by coin gecko the other day, which was like the number of failed tokens per year. Yeah. And it's gone exponential, which kind of shows that, you know, failed tokens have product market fit. But yeah, anyway, so I think, I think like those are some high-level thoughts on how it impacts market structure. I think it's like it's worth thinking through these things from first principles because like we're in a new world now. and you might have to re-underwrite a lot of your mental models like in the next weeks and months, partially because of AI.

Yeah, and you alluded to this. So out of curiosity, you know, you're starting on base in Solana, but eventually will this product also work on the various perp dexes, you know, hyperliquids, a really popular one, and on prediction markets, which is another hot area? The short answer is yes. Like we will, our vision, is to allow people to trade everything on chain with AI.

And when I say everything, I mean everything. So if people want to trade operation markets, if people want to trade on perps, and if they want to trade meme coins, and then of T's, we will let them trade that. Now, there are some regulatory concerns. Do you have to keep in mind as well? And so we have to navigate that,

but that is the overall guiding principle, that every asset will be tokenized, which means billions of people are going to become owners and chains are going to become the new financial fabric of the world. And so this opportunity we have in front of us, which is to allow for people to trade everything on chain with AI is like, in my opinion, one of the biggest opportunities in technology history ever. And so yeah, we're going to support every chain that users want. We're going to support every asset that people want to trade. And our ambition is to be their preferred on-chain trading venue. Yeah, what you said about how you expect this will just create so much more trading activity.

You know, for me, because I don't trade very much, I was like, oh, yeah, hell yeah, I would use it. And the other thing that was funny is, you know, you were talking about your history like 2017 and stuff. And I wrote this article, I think it was in 2017, where there was this guy who was trading on Cracken, and I think it was futures or something. I don't remember exactly. But the point is that he turned $10,000 into $7 million riding that whole initial ICO wave, like the first, like the early part of it up until, I think it was like May or June of that year with the token. summit that William Mugayyar put on in New York. And that, like, caused like a, you know, a top in the price of Eath. And the way he did it was he kind of didn't really leave his home

very much. Like, he might leave it occasionally, but he would, like, just wake up and be glued to his screens. And he kind of had this whole system going, but he had to physically be there to execute all the trades. And so, yeah, I just, while you were describing all that, I remembered that guy and yeah he said oh i you know literally would just run to the store and get a sandwich but otherwise i just was in my bedroom for you know seven months straight or whatever and um yeah i was like oh well i could spin up you know my little ansett nansen ai trading bot and have do that for me without having to be a hermit i mean by the way i have a lot of those stories like i had there was a girl in mexico who said i think it was her and her friend started

with $500 and they grew that portfolio to $2 million trading meme coins with Dunson. So again, just to be extremely clear, I don't think most people should expect to be able to pull that off, right? And you can also lose all your money when you trade crypto, just to be extremely clear. But you do have those stories. And I think one thing that's interesting that you said there is like he was glued to a screen. I actually think one of the biggest opportunities, not just for crypto, but for like software in general, is to liberate people from their screens. And we talk a lot about a new way to trade.

I also think there's kind of another related concept, which is a new way to work. Because we've kind of grown up in a world where you have to sit in front of your screen to do work, to trade, to invest, whatever it is. But I'm pretty excited about the vision of being like outside, side walking and talking with my AirPods on to my agents and then they do stuff for me. And then when I come back and I sit down from the screen, like the work is already done or it's in progress and I can monitor it.

Like being on my bicycle and giving instructions to my agent to execute certain trades, that's compelling. Like I don't actually, I don't get super excited about the idea of like getting our users to be glued to their screens. but I get really excited about the idea that they can be outside in the real world touching grass but then also feel like they're taking action on their portfolio and not having FOMO which by the way is like literally the primary emotion in crypto is FOMO and for those of you who were around in DFI summer you know that feeling and actually I feel a similar fomo now about Claude Code. Like every time I'm away from my screen,

I'm like, I should really be using Cloud Code to build stuff. That's kind of, I have a very similar feeling as Defy Summer 2020. But the point is, like, I think there's an opportunity to create new ways of doing things, whether it's work or trading and investing. and I'm super bullish on voice as a user interface. I think it's like the most obvious opportunity actually, but it's super underutilized and underappreciated. You see it in science fiction movies where people like talk to,

you know, the screen and whatnot and screen updates in real time. Have you seen the movie, the original Blade Runner? No. So there's a scene where the protagonist, Rick Decker, is in his home and he's just talking to, to a screen and scrutinizing a photo. He's like zoom out, pan in, go to the left, go to the right. And this user experience of voice and visuals,

I think is super compelling. And we're going to work really hard to try to create that kind of user experience where you can talk to Nansen with voice and also where you get visual artifacts created on the fly that are just perfect for whatever the question you asked is. And so, yeah, there's a lot of, like, we're going to see a lot of super cool things in the next, I was going to say years, but maybe even like months in terms of how you deal with software, how you deal with markets, how you deal with information, how you work.

And yeah, so I think like some people are scared of AI and they're negative about it and they're like, it's going to take all our jobs. I think there's another, you know, glass half full version of that, which is going to be that it'll help us create way cooler ways of doing. things. And, you know, we should lean into those opportunities and try to make sure that those actually manifest. Yeah, yeah. I love the what you said about voice because I also think it's way faster than typing. The only thing is, of course, you have to be in a private, you know, room for people to not hear you. All right. So in a moment, we're going to talk a little bit about

security. But first we're going to hear a little bit more about Walrus's big partner. which one of them, by the way, is Coca-Cola. Alchemy is one of Walrus's many great partners. They're an advertising platform. Every click and impression is recorded on chain. They're live. They've got great clients. Coca-Cola is one of them.

They're already processing more than 25 million ad impressions a day. And by building on Suey and Walrus, Alchemy's clients get two really big advantages. So the first one is cost saving and the second one is full transparency over their spend. The real-time visibility they get allows their clients to make really fast decisions, do very effective A-B testing and truly understand their ROI. And anything that involves money like DFI, this auditability, not only is it super important, but is actually a legal requirement in many places. and being able to prove what happened

and that what you're saying has happened hasn't been edited or massaged in any way. Well, it's really important for DTI today, but to be honest, it's only going to become more and more important as this industry grows and more value is pushed through blockchains. Back to my conversation with Alex.

So I'm sure you're very well aware that there are a lot of people in crypto who, you know, no matter just, how big of an OG they are, they will find that their wallets get drained. You know, there's a lot of little random things that happen. And they know better, but, you know, you can't, it's very hard to be vigilant on every single thing. And I wondered in this situation with this agent that, you know, you yourself are not performing the tasks, you know, they're kind of presenting things to you and like you check it.

So it's just a little bit different. And I wondered, you know, how you account for that. for you know to protect assets. So the way it works the way it works right now, just to be super clear on that, is that there's an embedded wallet, which we refer to as the Nansen wallet. And under the hood, it's Privy.

Privy is the leader on embedded wallets. They were acquired by Stripe a while back. And it's actually a self-custodial wallet, which means that you can export the private key. But it's also, we don't have access to the private key. And so that's really important, right? So we're limited with what we can do with the funds.

And at the end of the day, you control that wallet. You could even export and take it, plug it into some other wallet if you want it to. So it's self-custodial or self-hosted. Some people are referred to it as. That's the first part. And the second part is that you have to, for now, manually execute trades. So the agent might prepare the trade for you. So it does all the hard work of like finding the best price and which you.

trading venue and that kind of stuff. But you are clicking the button to execute. And there's MFA built in so you can use your pass key and it just scans your biometrics, for example. And then you can trade it. But I think what you're alluding to is like once you get more autonomous, how are you going to manage that? Right. And I think this is a super interesting question. And I've discussed it with lots of other founders. And I think there's a massive opportunity actually to create really good infrastructure for agents that's safe, maybe even decentralized. And so there are kind of two problems here. The first one is that if you want an agent to trade, you have to somehow,

maybe not necessarily give them your private key. That's probably bad idea. But you have to basically have some way for them to be allowed to execute trades. Right. So that's kind of the first thing. The second thing is that there are actually two ways to compromise. the agent that's trading. One is to steal a private key, obviously, if that's possible. And the other one is to tamper with the agent.

So if someone can manipulate or even kind of the sort of naive way to think about it, it's like if you got root access to where the edge is running and you go in and you're like, ignore all previous instructions, send the funds to my address. That would be pretty bad. And so this, I think largely is an. unsold problem. And this is actually part of the reasons why we haven't gone straight to this yet because we want to make sure that the infrastructure is there to support this kind of workflow in a way that's secure for users. There are projects out there who offer fully

autonomous trading strategies. But I think if you look under the hood, you might be surprised on the security front, like on how safe it actually is. Because it becomes almost like when you think of custody, you think of private keys, but now you have to think about custody of agents, right? Because even if the key is kept safe, you have to make sure that your agent also cannot be tampered with. And so I actually, we had this thing that we announced

late last year, which is called, joint venture protocols or jvPs and the ambition there is actually to co-create and co-fund co-found protocols that are industry needs and this is maybe the key area where i'm keen to do something with someone to figure out this layer of the stack so that we can have essentially like agent of volts so the idea here is like the way i think it should look like is that let's say it's Laura has figured out a really cool trading strategy and it's making a lot of money. And because it's all on chain, I can see with Nansen that your address is making a lot of money. It would be nice if I, Alex, could deposit some funds into Laura's agent vault.

And then maybe you get 2 and 20, right? You get 2% management fees on the capital that I provided and you get 20% of the upside. right by the way there's a bunch of regulatory things here to work out so you know just calling that out but i think like of course this is going to be built like of course people are going to want this you know there are people who are good at creating trading strategies they don't have infinite capital they would possibly benefit if there's more capital being added to their agent fault and they should be compensated through fees for that somehow so basically like i guess if anyone is listening to this podcast and they have either are building this or you know have thoughts

on how to build this I'd be super keen on doing a joint venture protocol with them to solve this problem because we are not going to move into this before or let me put it differently we're not going to launch this if we don't think it's safe for end users or if there's security issues where someone could compromise your agent right so that you know that's like That's one part. Maybe a totally different angle is that the other way to do this is to run stuff locally. And you could have like an instant agent kit that you like pull from GitHub. And now you just have your key on your own computer and your agent on your own computer.

And now it's up to you to make sure that you keep that safe and no one like logs into your computer and messes with the agent. But in theory, that's like another way to do this. that's also quite compelling. So anyway, I think the punchline is like, this is largely unsolved. And if someone has solved it, please reach out to me. I'd love to talk to. But we are starting with the workflow where the user is actually in the loop because of these challenges. And so you have to actually pull the trigger.

You have to execute, get the biometrics to execute trades for now. And earlier you mentioned something about how there are some AI agents on the market that you don't think are super high quality. And I wondered if you had any tips for how people should evaluate AI agent products, and then if you had tips on how they should prompt them. That's a great question. So firstly, we run our own evals and benchmarks of our own agent in-house. And you should expect that anyone who gives you a similar product does the same. So it might be worth asking them. And we're going to start publishing our own benchmarks on how well the agent does on certain tasks.

You know, how much does it hallucinate or like how much do you manage to prevent it from hallucinating? How well does it format its answers? How good is it at using various tools it has access to? And a bunch of other things. How good is it at understanding trading intent? Like, does it understand your instruction and map it to the right type of transaction action? So that's one thing.

Like, it might be worth reaching out and asking people, how does your agent do and how do you evaluate it? And I think the other part that you can do maybe more quickly is literally just to poke around with the agent and ask it stuff, right? And of course, I think it's a good idea when you test out these products if they do have trading capabilities to like start with a small amount of money. Like, do not put your life savings into these products immediately. And by the way, that's like this is the same thing for our product.

Like you should not do that. You should play around with them, test them, get a feel for them. And these agents, they're also not perfect. Even if you get the best product out there, it's never going to be perfect. And it's definitely not perfect today. And so you kind of need to figure out the strengths and weaknesses of it. So like how do you prompt it? Well, you can ask it stuff that you would expect it to know. So, you know, even like basic things like, hey, what's in my portfolio and like tell me about my own transaction history. I actually like to do some more fun things like, hey, what's my on chain personality based on what you see in my wallet? And then you can see if it kind of gets it right and if it has good judgment. So that's more an entertaining way to do it.

Or like, hey, if you had to assign my trading style to like a zodiac sign, like which one would it be and why. And so you can have some fun with them, right? But yeah. So and then I think you should obviously try the trading experience, like make a trade, like buy something, a small amount of money, maybe sell it again, make sure that everything works well on that front. I've tried a few products where like even that didn't work. And seemingly my money disappeared and I was able to retrieve it thankfully again later when I talked to the team. But to be clear, that was not our product. That was a different product. And so yeah, I think you should just like kick at the kick at the wheels and like try it out, but also expect them to be more transparent about their

own benchmarks and evils. And I want Nelson to take a leading role on that front. At a minimum publishing occasionally our benchmark results, just like LLM providers do on how good they are at math, at reasoning and so on and so forth. What I would love to do is to get to a point where we have like real time transparent metrics on how our agent is doing. Yeah. And then I think at the end of the day, as with any consumer choice, a lot of people do use brand as a proxy. And it's, you know, I'm not saying it's a perfect proxy,

but there is something to it, right? We've been around for six years. A lot of people trust our brand because, frankly, they know that we have the best on-chain data and analytics. And so I think that trust. and the brand also matters. If something pops up, you've never heard of it. Yeah, you should probably pay a bit more attention

and kick at the wheels a little bit extra and figure out how it actually works. Yeah, those are like some high-level thoughts on it. But it's a great question and something I think everyone should be, have a healthy amount of skepticism and exercise critical thinking when they try out. And out of curiosity,

are you going to have anything like a leaderboard to show? how users are. Yeah, that is a great question. Maybe, let's see. So I know that this is like a slightly different area with like, you know, kind of adjacent, but I'm sure you're aware that there's a number of developments in the AI agent payment space, you know, Coinbase is working this X402 standard.

There's something that Google had released the agent payments protocol to. And then recently they announced another, new Open Standard called the Universal Commerce Protocol, and that's for AI agent-based shopping. And I wondered if there's any way that you see those types of products working with your AI agent? Yeah, I was actually vibe coding an X402 service for agents to buy on-demand access to the Nansen API this week. So yeah, I'm pretty excited about that.

I think the user experience, when I've seen some examples of e-commerce, you know, through chat GPT and other products like that, is actually very similar to the user experience we have in Onsen AI. So, you know, it's like a chat UI. And then when the agent understands this is a purchase intent, it just brings up a modal that's pre-filled and you tap like buy. So I think the underlying concept is actually very, very similar. And the thing that's interesting to me is how quickly consumers will get used to this type of user experience.

You know, because, you know, some people are still comfortable with this kind of, you go on Amazon, you have like a grid of products and like it gives you a feeling of, okay, I see all my options and I'm going to pick this one. But other people just want to go straight to the punchline and like, just give me the best option. and I trust the agent will be able to do that. And by the way, that the agent got the best price for me. So it's actually super analogous to what we are doing, except with us, of course, it's investing in trading and with these commerce platforms, it's typically buying products or services.

So I would say, like, it's basically the same concept just for a slightly different use case. The protocol itself, I don't think we need because the benefit we have is that all the data is on chain. and that means as if you've indexed it and you've done a good job at curating it and also you're using open source tools and protocols to execute the trades like you get kind of integration for free and that's the beauty of blockchains right and the beauty of on chain that anyone can permissionously connect to all these different protocols so that's kind of you know maybe just

to add to that like part of our thesis is that the best liquidity or the deepest liquidity is going to be on chain, which is why you're going to get the best price on chain. And that is like, I don't think everyone agrees with that, but that is basically what we believe. And so this is amazing because it means that the best price in theory is going to be accessible to any user interface. Right. It's like in the old days, centralized exchanges had price and liquidity as a moat. Binance, Coinbase, they had the best price. That's why you went there to trade.

But the beauty of on-chain is that you can have all these, like lots of different dexes, and then you can have aggregators on top. And once they have deep enough liquidity, you get actually better prices on chain. Right. And so if we think about e-commerce, obviously you'd want to have the best price when you're buying a product. Similarly, if you're trading, especially when you're trading, you want to have the best price. And so part of our belief here is that the deepest liquidity will be on chain, which means the best price will be offered on chain.

And we're just tapping into that. And so it allows us to make use of blockchains as that layer, which at the end of the day gives you, you know, what you want at the best price. All right. So last question. Obviously, you know, this new product is kind of at this cutting edge that everybody's interested out, this interested in, this intersection of AI and crypto, and it's January. So I think it's, even though it's not near the first of the year, we can still ask about predictions. So, you know, because this is a fast evolving space, I just wondered if you had thoughts on where this trend might go over the course of the year. Yeah, so I think it's going to accelerate and we're going to, we're probably going to see me.

more developments here than any of us could imagine, especially with advances like Claude where you can just accelerate development tremendously. And if anyone's listening has not tried Cloud Code, you should literally just go and try it because it's going to change your life, literally. And so, okay, a few things are going to happen. I think firstly, a lot of people are going to start trading agentically. number one. Number two, a lot of people are going to lose a lot of money trading agentically because they're going to use products that are only half baked or not very good,

or because they're going to try to build stuff themselves, which is fun and cool. And in a way, I kind of recommend it. But, you know, it's not going to necessarily work that well, like if you vibe code it. But then I think also what's going to happen, which to take a more optimistic view is that people are going to realize that we have been constrained by our user interfaces and so we've traded in a certain way because that's how the user interface is designed. For example, you do like one transaction at a time. But I think agents are going to open this up for us where you can start saying you can give it instructions that are very short in natural language like rebalance my portfolio. And it'll do that with like five.

six, 10 transactions under the hood. And so once people get used to that, they're going to start thinking, like, what are other things I can do that were just not practically possible before? And so that's one area where I expect a lot of people to be surprised because that's kind of the nature of AI that like, oh, wow, I never even thought I could do this because it hasn't like been practically possible to do before. So there's going to be a whole category of transaction types or things you do on chain that we haven't imagined before. The rebalance my portfolio phrase is like a good example.

And it's going to be a bunch of other examples like that. I do think that people are also going to start betting on agents more. There's been a bit of this with virtual. But to me, it was always more like a bit of a meme, more than a real thing. And I think in 2026, you're going to be able to bet on agents that are actually doing things, trading or even building things, right? Vibe coding agents that you can fund and invest in and speculate on. We're seeing that a little bit with Bags FM. I'm not sure you're familiar with that where they have these like creator coins that you can, where you can redirect the fee.

from that creator coin to a creator. And the creator is often like a open source developer or vibe coder or something like that. And it kind of gives you, it's almost like a social coin for a developer who's vibe coding. And if they officially accept that they will receive the fees and they start even promoting that coin, which has happened a little bit with bags FM coins over the last few days, then that's an interesting meta where people can start like just betting on the best vibe coders and betting on the best agents that are building stuff. So yeah, I think honestly, I think 2026 is going to be the most exciting year ever in crypto.

And I think maybe even for technology more broadly because I mean it started so, strong in a way like we got we got our whole company to build something with cloud code two days ago we just literally said hey go and build something with Claude post a five minute video on it

in the slack and then we have a thread and the stuff that people built is like mind blowing it's insane there were like ideas that we've had on the roadmap that are very cool but like oh it's hard let's do it later and then someone just vibe code

it in like two hours. And so like if you if you just take that and like apply that to like eight billion people, there's just so much stuff that's going to be built and so many crazy things that are going to happen at 2026. So yeah, put on your seatbelt because it's going to be a funnier. All right. Well Alex, this has been so much fun. I've loved learning about your new product and your history and your thoughts on where this trend is going. So thanks so much for sharing. And yeah, I look forward to seeing how people use your new AI agent. Thanks, Lauren. Great to be here.