Loading summary
Transcript39 lines
- [00:00]
A
Foreign.
- [00:08]
B
CEO of on chain data and analytics platform Nansen, Alex Spanovic. Alex, welcome.
- [00:14]
A
Thank you. Great to be here.
- [00:15]
B
Nice to meet you in person.
- [00:16]
A
Likewise.
- [00:17]
B
That's what I love about these events. We get to meet all of our friends from the Internet, irl.
- [00:22]
A
We exist.
- [00:22]
B
Yes.
- [00:23]
A
We're not robots.
- [00:23]
B
We're real people.
- [00:24]
C
Proof of human.
- [00:25]
A
Maybe I'm a robot, but yeah.
- [00:27]
B
All right. I know you're speaking on a panel. It's called A New Way to Trade, Optimizing AI and data. Give us a little bit of a preview. What do you want folks to take away from that talk?
- [00:37]
A
Yeah, I think generally the one thing that I'm focused on right now and the whole team is focused on is agentic trading. And so we think that people are actually going to trade in a different way going forward. I think it starts 2026. I'm sure a lot of others are going to pursue the same opportunity. But I'm sure when you use software today, you mostly, or to a large extent you use AI like interfacing with products. And I think the same thing is going to happen with investing in trading. And so instead of having like lots of dashboards and charts and tables get really overwhelming, you just talk to an agent that you can work with end to end on discovery, due diligence and ultimately to actually execute trades. And so for us, the new thing that's coming very soon to Nansen is that you can indeed execute the trades on chain through our product. And so I think a lot of people will think differently about the product. We're not just an analytics product, we're actually an on chain product.
- [01:37]
B
More broadly, will the trading platform have an agentic AI component where I can speak to my AI companion?
- [01:45]
A
And it's amazing. So are you guys familiar with vibe coding? Right?
- [01:51]
C
Like you use how do we avoid vibe coding nonsense.
- [01:55]
A
Creating vibe trading. And imagine that you can just talk to the agent and figure out what are people doing on chain. What is smart money buying? You know, this token here, what's the deal with these addresses? Why is this address selling? Why is this buying? And you can kind of go through all that with the agent and then within the same user experience, you can actually pull the trigger. And so that's kind of how you should think about it conceptually. It's a new way to trade. It's not really based on kind of UIs, lots of clicks and all that stuff. You just talk to the agent and so yeah.
- [02:31]
C
All right, so first of all, let's recognize we've got some Alpha here, Nansen coming into the trading ecosystem, going to have a lot of competitors going on, but also a real top tier analytics engine. We've used Nansen for a long time of trying to understand what's happening on the blockchain. Then you get to trade all that information. Sounds like I could speak to it and say, hey, I'm really interested in what these types of people are doing with these types of wallets are doing, how these tokens are going to perform over the next. And it'll be in essence, build me a portfolio.
- [02:59]
A
Yes.
- [03:00]
C
Based on that.
- [03:02]
How do you think the, the normal world, who likes to think they're smarter than the robots, how do you think they're going to take to AI trading? For me, it's always been a natural that of course the machines will be better than we are, but there's an ego element.
- [03:18]
A
That's right. I think again to draw the analogy to coding, right, you can think about what cursor and vault and lovable and these products have done. I think number one, they've made existing engineers more productive, the people who've embraced coding with AI, but they've also opened up coding to people who aren't engineers. And so I think similarly, agentic trading is going to make existing traders better at trading, hopefully investing, but it's also going to make investing trading accessible to 100 times more people. Because if you've ever tried to onboard a normie to on chain trading, like they might be able to do Binance, Coinbase, centralized exchange trading, but if you get them on chain with their own wallet and like a Dex, it's actually extremely confusing and the user experience is very fragmented. Right. You go on a dex, that's confusing. I connect my wallet. If you're on mobile, you have to go and sign it in a different app. So we've thought, you know, what does trading look like in 2030 and how do we build that user experience today? And so the difference with Nunson is that number one, it's fully on chain, number two, it's agentic. Right. And number three, it's actually a self custodial wallet. So it's not like a centralized exchange. We don't have the private key, we're not a custodian, we're a technology service provider and you can export the funds if you want to. And so I think we're going to be part of a cohort of new or next generation trading products and I think a lot of them are going to come to market in 2026 every trading product should be pursuing agentic trading. And I think we have an edge because we have such great data. We have the best on chain data in the world and so we can give that directly to our users in this product.
- [05:02]
B
I got to ask you, I think everyone watching I've experienced experience this when I use AI, sometimes the AI hallucinates.
- [05:08]
A
Oh, great question.
- [05:08]
B
How do you think about these AI agents potentially hallucinating, giving incorrect information that then informs someone's investment thesis?
- [05:17]
A
Totally. So you have to quality assure this stuff. You have to run evals. Right. Evaluations of the agent and how it does. We run our own evals on a bunch of different dimensions, hallucinations. Does it understand the trade intent? Does it from the prompt you give, does it actually bring relevant data? Does it format the output well? So we run evals on all of these things for all of the messages we get. We have our own kind of ground truth data set that we run evals against. And so again, if you think what is 2026 going to look like? You're going to have a bunch of agentic trading products and frankly, most of them are going to suck. If we're being honest. People are going to lose money and they're going to be frustrated because there's hallucination going on. There's like poor quality user experiences. And so our bet is that because we have the best data and we have a great AI team internally and we have a trusted brand, people are going to want to use our product. That's basically the thesis.
- [06:14]
C
So I like to think about this in the model of like this is Moneyball for finance.
- [06:21]
A
Right.
- [06:21]
C
You now suddenly have an unbelievable analyst who can look and say all these different data points can come in. They can inform how do we create the best team, our team being our port. Yeah, I get worried when I think of AI, same way I worry about prediction markets. How can those be gamed? Right. If someone knows where you're pulling your data from?
- [06:41]
A
Totally.
- [06:41]
C
How does someone suddenly start to push some data there that they know? I'm doing a long play here to then be able to game the system. How are you preventing those who might have enough to deploy or enough to sort of train your AI without you even knowing? It's being trained to make decisions that in the long run someone's then going to pull a, you know, pull a trade against?
- [07:02]
A
I think the short answer, it comes back to these evals. You have to run religiously evals and quality assurance in house. Yeah. We've also been investing a lot in security, in house. I think when you. Anyone who does anything with AI has to acknowledge that it introduces a new threat vector in terms of attacks. And so you need to be very careful with that. Not just on vibe trading, but also vibe coding. If someone manages to get Claude code or cursor to inject some bad code into your code base, that could be devastating. So this is generally a problem for all AI usage. So we run evals, we've invested a lot more in security, we're getting new certifications. So, yeah, there's a bunch of different things we're doing.
- [07:49]
B
Alex, before we let you go, we are in the uae, talk to us a little bit about why this market is important. We know you pull a lot of great data from here, so maybe even lean on some of that data. What are you seeing when it comes to the Middle East?
- [08:02]
A
Yeah, we have a lot of users here. We've actually had some team members relocate to this region and so we're super bullish on the region overall. I engage with some of the regulators here and they're incredibly forward leaning and intelligent. It's striking, actually. I think maybe some of the best regulars in the world are based here. Yeah. So it's a really exciting part of the world and I'd love to do more business here, frankly.
- [08:28]
B
Alex, thanks so much for joining us at the desk and giving us that alpha. I think our audience is probably really loving it right now. Thank you.
- [08:35]
A
Thanks for having me.
- [08:35]
C
Alex, thanks so much.
- [08:36]
A
Thank you.