On The Brink with Castle Island - Dave Balter (Flipside) on the convergence of AI and Blockchain Analytics (EP.676)

Episode Date: October 13, 2025

Dave Balter, the co-founder and CEO of Flipside Crypto joins the show. In this episode we discuss: How Flipside has shifted its product to fully harness the power of LLMs for analyzing public blockchain networks. Building a product organization in the age of A.I. Perspectives on the evolving landscape of layer one and layer two blockchains. How the role of a data analyst will evolve in the coming years. Where Flipside is focusing its efforts within the blockchain data/analytics segment. To learn more about Flipside visit flipsidecrypto.xyz and follow the company on X.

Today on the podcast, I sat down with Dave Balter, the co-founder and CEO of Flipside Crypto. I go way back with Dave, and Flipside was actually the first investment we ever made from Castle Island Ventures back in 2018 shortly after we left Fidelity. Flipside has really leaned in aggressively at this intersection of public blockchains in AI. And I want to have Dave on to talk about this convergence and talk about the future of on-chain data and insights powered by AI. I think you'll enjoy this one. So without further ado, here's my conversation with Dave. Dave Balter. Matt Walsh and Nick Carter are partners at Castle Island Ventures. All these expressed by them or the guests on this podcast are solely their opinions and do not

reflect the opinions of Castle Island Ventures. Guests and host may maintain positions in the assets discussed in this podcast. You should not treat any opinion expressed by anyone on this podcast as a specific inducement to make a particular investment or follow a particular strategy, but only as an expression of their personal opinion. This podcast is for informational purposes only. Brought down by bad mortgage investments, Lehman, which has 25,000 employees will be liquidated. The federal government loans American International Group AIG, $85 billion. This is a different kind of market, and the Fed is asleep.

The federal government is stepping it to stabilize Fannie Mae and Freddie Mac, the two mortgage giants that have been threatened by the housing crisis. The Bank of England has pumped 75 billion pounds more to Britain's ailing economy with a new round of quantitative easing. You print a couple trillion dollars, and all of a sudden, people start to worry. So out of this worry, we have something called a Bitcoin. Bitcoin. Dave, welcome back on the podcast.

You've been on a bunch of times. You might be in our Mount Rushmore of guests here. Do I get a trophy? I don't know. We're going to go deep on AI today. We haven't done enough podcasts on AI. I'd say out of anyone in the portfolio, you've probably spent the most time on it. We'd love to maybe just start with your general views on what's happening in the market for AI.

Even outside of Crypta, I know you've spent years as a venture person, you've spent years in the startup trenches. What's going on in the AI market? AI is transforming everything. I would say there's two big shifts happening. There's the in-company shift where everybody's work is transformed at Flipside engineers start a cursor or other solutions. They don't code by hand up front. They buy code and then transform that into human helping that code be productive. I think that transformation impacts the whole business.

So product at Flipside's role has had to change dramatically because engineers can actually move faster than a PRD or a prototype that comes out at research. and product. They build it, test it, and are getting outputs before even a product person can create it. Internally, AI has changed just how everything is working. And then externally, the ability for businesses to operate with a different form of product or we'll talk about it as it relates to data within the crypto space, but you just are seeing everything is turned out. One company I sit on the board of is in Boston. It's called RayFit. It's an AI app which exercises with you. So it's doing everything from like counting your reps, but then it's building custom workout plans for you based on all the things you would think about. My knee hurts or now at a gym in a different city and I don't know what weights I have.

And so all of your workout routine is being impacted by something much more accustomed to you. Everything's changed. I don't think we're ever going back. It's like the internet cloud, all that stuff. We're not going back. Does it feel like the internet to you? I mean, obviously you were there when the internet really started to become adopted in the enterprise. Does it feel like a similar shift?

Yes, in the polarization of people's perspectives. That's good alliteration right there. I like polarization of people's perspectives. I got to write that one down. There are people who are like, oh, AI, I will never touch it. My kids are like, oh, it's draining water or whatever it's doing, electric. I'm like, you're on TikTok all day. I don't know what you're talking about.

That's doing just as bad. You know, it's polarizing because there's fear and change. I don't have to relearn how to do this. I don't want to have to do this. It hallucinates. It can't be good. to once people are starting to use the tools and feeling what's changing to like, my life will never be the same.

And so it feels like the internet in that you've got real transformation happening in people's footprints and how they're doing things. It feels bigger than the internet in that AI systems have implications for our world and for humankind and singularity and all that stuff. And that's pretty big. So it feels bigger than the internet. It feels like the same type of change. but really implications are much further reaching.

I've had the experience of using Chatchipati and the other products. And I remember just thinking when I first started using it that my entry-level job as a management consultant probably wouldn't exist now, or at least in a couple of years, it probably won't. Or maybe that there would be two people in the analyst class as opposed to 12. And so really hit me that a lot of the things that I was doing in terms of secondary research and building slides and synthesizing data could just be done all. a lot more easily now. I'm wondering if you had a similar moment thinking about Flipside as a company

once you start to harness AI. We have this board member, Shkargosh, who's brilliant, he teaches blockchain and AI at Harvard Business School. He was talking to us about consultants and he said, in the industrial revolution, what changed was machines took the weakest person and the strongest person and made them equal. It used to be if you were the strongest, you would get paid more because you could do more and the weakest were given the menial tasks. Now, I'm machine balances strength. And in the consulting world, you're seeing the same thing happen. The partners used to be the most brilliant and knew how to answer things and be strategic, and the analysts were like, go do a deck. But now, with AI, they're equal. The analysts can come in and do as much senior level

strategy as the partner, and that is causing real shifts in how consulting happens, because as an analyst, you probably wouldn't come in and just do decks anymore. You could be actually building just as much strategic value as a partner. That's wild. Just a lot of leverage in your day today. Totally different amount of leverage. Think about that. What you'd need to learn to become an analyst. A consulting firm is different because you kind of have a lot of tools at your disposal that makes you really smart.

In Flipside, I'll tell you the aha moment. We, for seven years, eight years now have been building data infrastructure for blockchain ecosystems. And our team was building a tool that was called Rio. Think of it like a super Twitter feed. We had a client, a big client of ours, asked for something like it. Give me something where I can see, like, information in rapid real time, et cetera. So they were building this thing.

And I was getting more and more frustrated. It was, here's another product. And like, oh, God, and the UI is not good or whatever it was. On a Friday night at 10 p.m., Kyle, who's our COO now sends me, he's like, oh, check this out. We just hooked up all the flipside to MCP. And I'm like, what? Like, I didn't know. He's like, oh, it was a project.

And I go. And it's the first time in the history. of Flipside where I've been able to use our data. I'm not an analyst. I'm not an engineer. I've never even touched our data. He's like, check this out. I go and I write a simple query like,

oh, show me the change in DFI protocol growth on Solana, compare it to base. It built the most amazing visualization and return of data. I was like, holy, I am now an analyst, much like this Industrial Revolution thing. And I literally went to the team and said, kill Rio. They were like, what?

I'm like Kill Rio, we are all in on this. Because once we've moved beyond just the analysts, and I love analysts, analysts who use our system even better, once we move to the strategists or the marketers or the CEOs or the CEOs or the exact directors who can't code, everything changes, everybody's equal. And they were like, this is crazy. You're going to sacrifice all this building. I'm like, yes, this is what we're doing. That was 90 days ago.

That's unbelievable. You have a very strong business. I mean, you guys have very healthy revenues and you're doing a lot of interesting things. Do you view this as kind of a one-way door bet the company on the future of this technology? Well, here was the second crazy thing. I've told the company, I am either the most brilliant or the most dangerous person possible in this business right now. I am not sure which. I'm pretty sure it's a good decision. I'm pretty sure it's an amazing decision. But there's things I'm doing which are, transformative to the business in ways you can't go back. So that second thing I did after, let's kill Rio, we're going all in an MCP for seven, eight years. We built this, our own SQL interface. It's called Studio. 600,000 analysts were using it. So I went to the team.

I said, we're going all in. We're going Saul and then we're killing Studio. I want, we're killing Studio. What are the analysts going to use? Well, the analysts can still use our Snowflake instance, which has an amazing studio editor. We're not telling them to not code in SQL. But as a business, we're going to be the best in the world at AI systems integrating with big data workflows, in this case with blockchain, and holding on to our own SQL interface, it's a different business.

It is okay. We do not need to be there. And by the way, within a year, every single one of the other data providers are going to be touting their AI system and realizing that their SQL interface is going away. Think of what cursor is done to basic coding. It's going to go. We got to go. We're going all in. We are going to be best. We've already solved so many of the challenges to make this work. It's pretty amazing.

That is a one-way door. I did that. That's pretty awesome. I mean, I want to get into what people are using the data for, but the way I'm thinking about your category and maybe some broader aspects of the blockchain spaces, just think about the traditional securities market and how slow reporting is. For instance, I like to monitor the ATS trading venue data. to just see which venues people trade traditional securities on.

And it gets published one or two months after the end of the quarter and it's an Excel file. And it's just crazy to me that that's how we appraise these secondary trading venues in TradFi, whereas you can get tick by tick on Uniswap to just see who's using that versus its competitors. And right now, on chain, you have bearer cryptocurrencies, you have stable coins. I looked at it this morning. I think there's $7 billion of tokenized money market mutual funds. But with this market structure bill, you're going to get, a lot of other assets that are put on these chains.

And it's incomprehensible to me that we're going to have government agencies with Excel files, three months late, reporting, just trading volumes. So where is this going as it relates to data on blockchains and how are you guys thinking about it? There's a time equation to all of this. You could go get that tick data over here. You could go, you said reports on Excel. One of the biggest changes in this entire AI evolution or revolution is the time to

outcome. If you had all that data at your fingertips that you were just talking about, and I said, Mac, go figure out some analysis. You know, in yesterday's world, if it was a consulting side, you're going to go put them in spreadsheets and do X, Y, and Z. If it's the engineering side, you're going to go write queries and QA and do a bunch of stuff, hours and hours of work, probably some pair coding, whatever, like challenging. You're now doing that in minutes, and you don't need to know any of that other stuff. You can start building off other people's artifacts or start on your own. It's the level up of what's possible that you can't have done before. I'll give you an example. Some folks in our system figured out that not that you can understand

trading potential or where should I trade, about 50% of people that are using our AI systems, use it for trading. Hey, what's this wallet doing? How would I trade like that? What are the best trades that are happening here? Here's some ideas. But here's what some of did the next level. Hey, can you determine if my trading activity looks like I'm a human or a bot? This was their prompt, their wallet. Is this wallet likely a bot or user? Check this wallet, what its position is on Monad. Study the specific protocols this user engages with frequently. And then it comes out with a report that says, here's the types of behaviors you're using that make you look like a bot. Here's the trades that don't. All that's super important as you're

getting to not can I analyze this, but can I actually get to a different level of understanding of what I do if I understand trading patterns. It's kind of wild. You couldn't even think of that before. That's really fascinating. I mean, I remember in the early days when I first started looking at blockchains when I worked at Fidelity, one of the reasons why I met Nick was because he was running this open source data project that would actually just tell you how many addresses and what the volume of transactions were happening on chain. And I was just trying to figure out, Is anyone using these things? Is this just a technology with no use case?

These days, it just seems like you can go incredibly deep. I'm curious how you and your customers are thinking about using the platform to understand the economic activity happening on these chains. I think I'd started asking analysts in the previous days, are there transactions, is their volume happening? You're now getting into the intelligence level. This is frankly why this data needs to shift into AI platforms. You've got to move before beyond sort of one-dimensional.

Here's a chart that shows growth to, hey, what is the patterns of behaviors happening in stables that may actually be more about what are likely scenarios to occur based on the evolution of things that have been happening in the stable market? People expect more answers than just chart go up, and these systems give you that opportunity. Have you ever thought about the amount of capital that is spent in financial service? to get delayed data. It's just staggering. Companies are paying for surreouly reports on AUM of their competitors and transaction volumes and things. It just seems like that whole paradigm is going to change as a result of these assets being on chain. Yes, you and I kicked off this

journey seven or eight years ago, and that thesis was the very beginning of the journey where you said, this is coming. I just think you're seeing the evolution go from. This is coming. it hasn't materialized yet. Let's use some data and analytics to help people start to sort of be able to see into these patterns that are occurring. But almost your thesis right at the very beginning when you were coming over to 207 South Street

and whiteboarding with us 2016, 2017, it's here. There are people now really moving the traditional finance information flow on chain with stables and everything else. Your intelligence needs to be further along. It doesn't end at just how much volume is happening on USC. You even ask some of the biggest institutions and they're like,

I don't understand the patterns of that. Who's using it? But if I look at it against every other stable coin, what is it indicating the change to traditional finance is happening? Is in flight. You need a much deeper intelligence system to get to. It's been interesting. I mean, if you had asked me back when we were having those initial whiteboarding sessions, the timing of when you'd have dollars on chain and securities on chain, I would have said it would be sooner and didn't foresee a world where you'd have the guy who taught crypto at MIT coming in and just blocking banks and broker dealers from participating. So obviously that set it back, but I guess at the same time that all of the regulatory blockers were happening, the blockchains

themselves have just gotten so much better, faster, cheaper, stronger. What has it been like to just monitor that evolution? Obviously, it has an impact on the types of customers that you bring into the flip-side fold. It's pretty interesting. I had a conversation this morning with a blockchain, and I'll tell you what they're actually faced with as traditional finance really does take center stage here. That evolution has happened. I think many of those blockchains were fighting over how's that going to happen, and are we going to be center when it does? I don't think it's played out like everybody's thought. The thing that's actually happening is many of those non-stable coin-oriented or multi-value.

They're faster or whatever, but maybe don't have a stable coin offering that is as deep as a USDC or folks like that. They're actually faced with a pretty untenable dilemma. The market still perceives value where there's vanity metrics occurring. Everybody now realizes it's mostly bots. And that's only come up in the past two years or so

the bots have really taken over most of these chains. So they're faced with this dilemma where we can make the charts look good by using growth tactics that are to like get bots to do things that just show movement or wash trades or whatever the hell it is. Or we can make sure we get like real humans and quality users. And that actually transacts better on the chain. It's real customers. But the market of basic intelligence, go look at dashboard, doesn't like that very much. They're like, oh, chart doesn't go up.

It's not about the chart. Is this a real business or not? These blockchains are faced with one of the worst dilemmas ever. Do I sort of falsify the real growth to look good? Because it's like looking at everything in 1D or black and white, when everything is in color in 3D, dashboard is 1D. Don't you think at the protocol level, it'll eventually converge around, are you competing to be some form of internet money,

in which case maybe there's some valuation framework that you'd apply as a store value or are you competing to have transactions happening on your chain, in which case we could think about cash flows. But ultimately, if you're in the latter camp, to your point, I think you have to have real businesses building things on top of your infrastructure. It's almost like what good is Swift or ACH if no one's using it? A thousand percent. I think you're seeing the store of money, the store of value case play out. That's where all the traditional finance dollars wants to go. And it's rational. It's exactly. Our thesis, back in the day, when traditional finance changes the rails to this, everything opens up.

Time is shorter. Trust is easier, all that stuff. It's that second part where if you were building a blockchain for years that was considered a premium L1 or L2, but kind of we're just trying to attract new products or projects or cool ideas, and it's not in the store of money. You're fighting to either make yourself look pretty by enabling bots to do things or you're going to maybe be thoughtful about a smaller number of customers actually using the products you have, or probably you need to shift entirely to a store money or something of that nature. That's why you need intelligence, not just a dashboard, because it just doesn't mean much.

It doesn't do the industry any favors. As an industry, we probably overhyped certain categories that just haven't come to fruition, particularly around these decentralized internet architecture use cases. File storage is a thing, but it's not at scale, decentralized compute. Some of these categories, I would say, just weren't ready for prime time. I actually heard Anatoly on the All-In podcast a couple weeks ago talking about what are the next breakout use cases. And I agreed with what he said around a lot of the things that were attempted in 2017, 2018, didn't work, probably will breakout. I guess it kind of looks like the internet in that capacity where you have a bunch of things didn't work until broadband. Exactly. And then Amazon and

Netflix and all the other things that could be possible with that. The previous world order was, I need to figure out if it looks like it's working. So you might go to a SQL level analyst and say, help me track TVL and momentum for DAPs on my platform and others. You've got a first case issue here as the space matures. How long does that take? What's the quality control process?

So that might take a couple hours. It might have an error in it. It takes a while to decode it, whatever. The new world order is we want to track TVL and momentum for depths across all ecosystems, but I also want to grow TVL across my EVM footprint. You're moving into action stage, not look back, look forward stage. Now, number one, to do this work, it takes 10 minutes. You write it in human form and your AI systems are getting you,

not only the dashboard you want, but the intelligence, being thoughtful in how you prompt that, much like any AI system matters over the last 30 days on each of these five chains. This is a real prompt someone put in there. Get me token transfer volume into the contract by chain and by token, a very human language way of deciding what you're going to build. And then show me what actions have grown TBL for quality wallets in EVM footprints. And now you're going to get a strategy. It's amazing.

And then you optimize like anything else. What would you use to do in that case? You would call like a consultant and be like, help me with a strategy. But in 10 minutes, you're getting a real form strategy. It's a token design agency via LOM. Totally. All the token design agency should be quaking in their boots at this point. What are you seeing in terms of how fast the LOMs are evolving?

And I'm sure that has downstream impacts on just your product motion. Here's the key for blockchain data or any large data system. If you dump a bunch of data into the system, it does like any AI system. It tries to rationalize what you're asking to an outcome. And if you just pop a bunch of data in and say, go find TBL and maybe you don't properly say in this chain or this year or this type of data science workflow, it actually will go maybe pull the wrong table first and join it this way. And naturally, like a user, you get frustrated, much like in a typical AI system. When you ask a question, you're like, why did say that? Oh, it hallucinated.

It's not quite that. It's actually about workflows. The thing you need to do is you need to actually understand how. a data scientist would consider that problem and build certain types of workflows that teaches the AI system to look in certain orders and do certain things. A lot of people use our system for AirDrop analysis or I'm trying to do this AirDrop, help me build a strategy that takes these types of addresses out or put these in. You kind of need to teach it how to look at addresses in a certain way. So the systems are getting smarter because we're starting to build workflows.

The data is all there. You can put anyone's data in there, whoever, doing Nansen, doesn't matter. They're all going to dump their data in these systems. The workflows are the things that matter. They're getting smarter. I was given a use case years ago by one of our advisors who said, you could build Google right now. The problem isn't that you aren't technically smart enough.

The problem is that you don't know the potholes they've stepped in already, and you're going to step in 10,000 of them on your journey to trying to build a better search engine, and you'll never get there. They're so far ahead. It's the same in LLMs. All the folks who are building the right workflows and structures that when you ask a question, it knows how to think about it, that's what's going to make the LLM smarter. How has that insight informed just your hiring practices and how you're building the team? How has that evolved over time? Everyone in our system at Flipside Codes direct in cursor or otherwise, that changes the behavior profile of your engineers or data scientists. I mean, we

all saw Brian Armstrong throw out anybody that wasn't building with AI first, which is totally right. It was maybe heavy-handed, but totally right. Now, when we're hiring, we think a lot about the ability to educate or articulate the workflows you're going into. We could hire the best data scientists in the world, but if they can't say, wait, wait, this is how I process that equation. This is how I get to an answer. We can't pull that out, extract it, and turn it into a workflow design. And so we care a lot about maybe the articulation or the ability for someone to document what they're doing as much as the ability to be able to code itself. It almost brings the engineer up a level towards more of the product manager, it sounds like. A little bit. The whole system

is changing. Product managers are no longer writing specs because engineers move faster. The data scientists are doing the product manager speccing, which is actually not about your front end or your overall design. It's about workflows. Everyone's role has changed. It's fascinating. How are you seeing just the industry writ large now? Obviously, post-FTX, saw just a lot of churn and the startups, a lot of wind downs. What's it like out there right now when you're talking to other entrepreneurs? Fundraising is hard for everybody all the way up the stack. Funds are having trouble with their LPs. And one investor said to me, fundraising is hard except if you're open AI. All the money is getting sucked into these systems

that are changing the whole ecosystem. Now, the problem is going to be, I mean, we've all been through this before. There's bubble behavior that's happening. All it takes is for open AI to stumble a little bit or they're like, it's going to deflate a bunch of stuff around it. It's going to be like a housing crisis. It's not that it's changing everything. It's just that it's getting so valued in a way where it's really hard to uphold that

value. But entrepreneurs, you've got multi-factors. One, it's hard to raise capital because everything's going to AI, so you better have a deeper AI part of your business. to your time to market. If you can't prove it on your own, get to $5 million ARR without any outside capital, you're sort of not building correct anymore.

That's been a big change. So you've designed systems for how a company gets started, has changed entirely. I think we've gotten at the brass tax spot of entrepreneurship. There was a lot of sort of promotion and fluff going on, touting your excellence in something. And now it's like, I actually just got to prove it because everybody can see exactly what's

happening and systems let you prove it before you talk about it. It is fascinating. I mean, if you think about this in a historical context, some of the CAP-X here reminds me a little bit about fixed-line communications, the amount of capital that went into that. And obviously, we got the internet out of that, so that was great, but there were a lot of corpses along the way. It's interesting to see companies like Meta and Oracle fundamentally changed their approach to the capital markets and start burning a lot more cash here in order to keep up. So I think you're right that at some point, the amount of leverage in that part of the system in the AI market, I don't know if it'll burst, but there will be some correction.

It'll be interesting to see what the downstream impacts are there. It's going to hit all of us. You always should focus on your own business, not what's happening outside your business, as you've got to build something sustainable. If that cracks over there, have enough capital, know that your customer is maybe going to get into some pain, so have enough room to work a little differently if your customers get challenged. It's going to happen.

You don't know exactly how, so just don't run out of money. That's the number one job, right? The number one job, just don't run out of money. So whatever you got to do, cut your team faster, do more with less, or whatever it is, but survive. It's been interesting to see just the growth pockets within digital assets with stable coins really bursting on the scene over the past five years. You'll have tokenized securities here in the near future. The other one that I don't think we've talked about as much is these prediction markets and just the amount of on-chain activity happening in that category. Curious if you have a perspective on that and whether there's interesting insights to Colleen from some of these platforms.

We just saw it a tweet, which we're trying to hide from it a little bit, which is someone who literally called out Kashi in the prediction markets. But what's happening at Flipside and a couple other spots are where the real edge is starting to lie. And this might be the beginning of the next frontier of how people actually behave. The prediction markets almost in some ways are still one-dimensional. Humans going back and forth and debating these things. but if you had an intelligent system that's looking at multiple factors to make the types of predictive bets, everything changes. I think those are really interesting platforms. They tap into the social norms of competition and value creation and everybody wants to know the answers and that on the answers.

They're tapping into all the psychological behaviors, but you're going to have to have AI systems or intelligent systems that start to maybe automate those or help those move fast. than the human capacity to sort of compete is occurring. I'd imagine there's going to be AI systems that operate on their own as their own behaviors competing with each other in a predictive way. I'm sure people are trying to do that already. You remember in the early days when Draft Kings came out, people were scraping screens and trying to put in more bets over time programmatically. I wouldn't even know if they had an API initially.

I'd imagine you'll see similar type of dynamics there, and it just seems like the regulatory environment now allowing these things to exist, in the U.S. will also change that where you almost have this new financial market emerge. If the stock market is some form of gambling, prediction markets just take it to a new level. It's Reddit armies were bad until Reddit armies were just part of the process by which you're able to capitalize on returns. Well, I guess with a lot of things, if you can figure out a way to tax it, sometimes it comes into the light. It might be the same thing. So maybe back to the product side, what are you seeing actually people put in as prompts and what are the interesting

behaviors that are notable since you launched the product. I know it's only been a few weeks. There's more basic prompting that people are starting to learn over time and talk about one from yesterday. Use the trade adjusted fair value workflow. So make sure you call workflows, learn what workflows matter. Use the trait adjusted fair value workflow to analyze a specific contract address to find over and undervalued NFTs. Call a workflow. That's a prompt. engineer, make sure you're thinking about that. Call a table if you think about, call the monad table or call whatever table, and use that to derive some intelligence.

I'll just do a rundown of some of the things people are using the system for, this is yesterday. We also use the intelligence platform, just say, tell us the types of things people are asking about so that we continue to refine. So advanced defy analysis, much like a pretypical editor. SQL debugging and complex analytics. multi-CT queries with error troubleshooting happening real-time. Security and research focused. So people are looking at smart contract exploits, copycat cases, bite code deployment attacks.

They're referencing academic research next to some addresses. A lot of strategic growth campaigns. Flipside analyzes 700 million addresses daily, basically to know which ones will be performant, which ones are bots, which ones aren't, all that stuff. So you can ask the platform, generate a strategic growth campaign. for me, using scoring systems. Again, you've got to ask it, where to go. Lots of typical stuff, historical price analytics, all that stuff. One thing that's super interesting is the system is multi-language because AI can understand everything. Here's just a rundown yesterday of the languages

you use. Chinese, Vietnamese, Korean, Arabic, Turkish, Bengali, and French and English. Just in a day, this thing's out there. I'd be fascinated. I'd need to start doing some more queries on it myself, just around the time of day that certain things happen, if you could pinpoint geographic regions that are more active than others just based on the time zone. That seems to be like an underexplored part of this market. Try to write that in SQL and do it in a time that gets you somewhere at a cost that makes sense versus asking the system to just do that for you. All those things become possible. All right. This is great. Dave. Who could have imagined that we would be talking about LLMs back in 2016, talking about crypto data and asset management

back then. I didn't have this one on the bingo card, but I love the direction. Where can we send people that want to use the product, learn more about Flipside? If we did have it on the bingo card, maybe we would have started Anthropic, and then you and I would be having this call from a private plane. A different story. Yeah, different story. Anyway, everyone's done okay. We're pretty good. So you should go to Flipside Crypto.xyZ 4 slash chat. I think pretty soon, actually, our front page of our website will just be the chat products, so pretty easy. By the way, you can use that in Claude. If you have a team's account, oftentimes that's really interesting because you've got your own data put in there for other things. So you could either use our interface,

which it does multi-AI systems, uses Cloud or uses Gemini or other places, or you could install it directly into Cloud. All that's on the front of the site. But pretty simple. Head over there, give it a try like anything. You got to learn how to use the system. So you start easy. But as you get more complex. If you have deeper questions, reach out to us. We have a bunch of prompt engineers. We will teach you some pretty nifty tricks. Well, thanks again for coming on the podcast. Congrats on the new product direction. I think it's awesome. Thanks, Matt. Thanks for listening to another episode of On the Brink with Castle Island. To learn more about Castle Island, visit castle island.vc. And to listen to all of our podcast episodes,

please visit castle island.vc slash podcast or just click on the tab on our website. Thanks for listening.