Latest / The Index Podcast / Aptos CEO Avery Ching: Open Intelligence, Data Ownership & Why AI Needs Crypto
Transcript
- 0:12SPEAKER_01: Hey everybody, welcome to the index.
- 0:15SPEAKER_01: I'm your host, Alex Cahaya, and today I'm joined by Avery Ching, co-founder and CEO of Aptos Labs.
- 0:21SPEAKER_01: Avery was the lead engineer behind Meta's Libra project, and today he's building Aptos, which is a high performance infrastructure layer for the next generation of the internet.
- 0:32SPEAKER_01: And it's designed for speed, reliability, and institutional adoption and the rise of AI-powered applications.
- 0:38SPEAKER_01: Avery, I'm super excited to have you on today.
- 0:39SPEAKER_01: Thanks for being here.
- 0:41SPEAKER_01: Thanks for having me on.
- 0:42SPEAKER_01: So I I always start the show the same way, and I'm curious just to hear your story and kind of what gets you out of bed in the morning to work on Aptos.
- 0:50SPEAKER_01: You know, given your background, there's like a million different things you could have done with your life.
- 0:53SPEAKER_01: And you know, I've obviously followed Aptos for a very long time.
- 0:56SPEAKER_01: And yeah, would love to just hear more about that and what gets you out of bed to do this hard thing, building a layer one.
- 1:02SPEAKER_01: And uh thanks for being here again.
- 1:04SPEAKER_00: Well, thanks for the question.
- 1:05SPEAKER_00: I I think these days it's more about what can keep me asleep.
- 1:09SPEAKER_00: Is there's so much exciting things happening in the world uh that you don't want to sleep a lot.
- 1:13SPEAKER_00: And I've actually been really thought thinking a lot about the challenges of AI uh as it's been progressing through this society in the last six, twelve months at a rapid rate.
- 1:23SPEAKER_00: Um, some of the things that we've been thinking about here are more about the challenges of open uh open open access to models, uh open uh the thoughtfulness about how people use our data and the role that decentralized technology can play in that.
- 1:36SPEAKER_00: And I know that there's been a big struggle over time over to see what that could evolve into, but it's pretty clear that these are the fundamental challenges uh that are the most important in probably the next three years for our society.
- 1:47SPEAKER_00: And so that's that gets me really excited.
- 1:50SPEAKER_01: I, you know, I share the same passion, especially around open access to open models.
- 1:55SPEAKER_01: I've been talking about it was when I used to have a sponsor called Bagel Network.
- 1:59SPEAKER_01: They're a really cool company, and they're um in our space, the intersection of AI and crypto and blockchain technology.
- 2:04SPEAKER_01: And uh together with them, we coined this term called open intelligence.
- 2:08SPEAKER_01: I mean, I'm sure other people have come up with it, but I feel like we were one of the first.
- 2:11SPEAKER_01: And the idea is just open source super intelligence.
- 2:14SPEAKER_01: And I I really believe that intelligence is a human right and that everyone should have access to it.
- 2:21SPEAKER_01: And I worry that uh, you know, we've been seeing it with Fable 5, Bit Getting Gated, Mythos, GPT, I think 5.6.
- 2:30SPEAKER_01: Um, these are real problems that that if you can't have access to this and a few only a few people do.
- 2:35SPEAKER_01: So I'm curious what your thoughts are on that.
- 2:38SPEAKER_01: And then a secondary question is like, how does decentralization play into it?
- 2:42SPEAKER_01: Because I I've seen a couple of takes, but I'm wondering what your take is and how Aptos can play into that.
- 2:48SPEAKER_01: I think it's super important.
- 2:49SPEAKER_00: Yeah, I think the the examples you touched on are very, very relevant and I think just the tip of the iceberg, honestly.
- 2:55SPEAKER_00: I think the regulators and governments are trying to figure out like how how this technology, which is so fundamental to change in society, can be controlled or used um within reason and and securely.
- 3:07SPEAKER_00: But but the the thing is that um, you know, you have to start the data sources.
- 3:11SPEAKER_00: What we are doing today is we are feeding LLMs and uh inference with all of our minds and our data.
- 3:18SPEAKER_00: And that that leads to a situation where like we had the internet before, where the internet world's information was organized and searchable, accessible.
- 3:25SPEAKER_00: That's that's great.
- 3:26SPEAKER_00: That actually is just a benefit, a huge benefit for everyone around the world.
- 3:29SPEAKER_00: But now you take it just a further where you're taking that information that's being used to organize now, it acts on behalf of people.
- 3:34SPEAKER_00: In fact, kind of taking people out of the equation as intermediaries, that means that collectively, we as a society, we as people start to lose that fundamental right of like action ourselves.
- 3:44SPEAKER_00: And that is a huge challenge going forward.
- 3:46SPEAKER_00: I think the models are just a good example, uh, right, but it stems actually all the way from the data portion of the equation.
- 3:52SPEAKER_00: And so this is an area where we've been really thinking about a product called Shelby we've been building uh for over a year now.
- 3:59SPEAKER_00: Uh, it was built with the idea that um decentralized kind of storage needed a high performance, uh very powerful solution that could compete with today's uh cloud providers and hyperscalers in terms of performance, reliability, and even be more reliable because it's actually got um servers span throughout different kinds of infrastructures that are more uh tolerant to that.
- 4:19SPEAKER_00: But yet with the performance and costs, uh, which is a win, like egress fees are a challenge for anyone who uses cloud infrastructure.
- 4:27SPEAKER_00: You have to stay within cloud or pay massive fees to get out.
- 4:30SPEAKER_00: Um, and it was it was a good starting point, uh, but that system has now evolved to be the centerpiece of how we think about the data side.
- 4:38SPEAKER_00: And um also, like you said, uh I think great term, open intelligence.
- 4:42SPEAKER_00: Intelligence comes from two things it comes from data and our ability to reason on that data, right?
- 4:47SPEAKER_00: So you have LLMs that are providing the reasoning capabilities, and you have the data which we are providing openly to the largest frontier models in the world.
- 4:54SPEAKER_00: And then we have even others in in other places of the world distilling those models, right?
- 4:59SPEAKER_00: And extracting data out of those models and using them for their own kind of model training, right?
- 5:02SPEAKER_00: So that is a very interesting confluence of factors, which you know, you can see a world in three years very dystopian, right?
- 5:09SPEAKER_00: Where there are a number of very powerful models being built.
- 5:12SPEAKER_00: Access to those models is available only to a few, and the average person is going to be stuck at a you know, a very small productivity factor, ability to kind of generate value for society compared to what those people with large access to those models are in the future.
- 5:27SPEAKER_00: Yet their data has been used to actually produce a lot of that kind of collective brain power and reasoning capability.
- 5:33SPEAKER_00: This is this is unacceptable, right?
- 5:35SPEAKER_00: This is not an acceptable outcome in the next three years or even two years, right?
- 5:38SPEAKER_00: Or whatever the time frame is.
- 5:39SPEAKER_00: It's accelerating so quickly, it's hard to predict.
- 5:41SPEAKER_00: Uh, and I think this is the only, this is this is an area that we in the kind of space that care about this problem need to need to work very quickly to solve.
- 5:49SPEAKER_00: Uh, that's where something like Shelby comes in, and we are starting to build uh a lot of infrastructure around Shelby to support kind of this um this effort for people to really have control over how their data is being used, um generate revenue from their data, also uh support when it is used for training, for fine-tuning, and also enabling people to be much more productive in their own lives with uh the data they provide.
- 6:10SPEAKER_01: So there's so much we can talk about here.
- 6:11SPEAKER_01: I think the high-level thing though, you know, is control and ownership.
- 6:15SPEAKER_01: That's just from a narrative standpoint.
- 6:17SPEAKER_01: It's you're seeing starting to see it on Twitter a bunch where people are talking about using more open source models, for example, because then they can control their intelligence stack.
- 6:29SPEAKER_01: I mean, even like Brian uh Armstrong at Coinbase had that big post the other day that went pretty viral.
- 6:34SPEAKER_01: It got like 3.4 million um views, and he was talking about how they're using AI at the organization and some of the things that they're trying to do to create efficiencies and you know not spend as much on the top-tier models and things like that.
- 6:48SPEAKER_01: And but what it comes down to is they've built a bunch of software in-house to control their intelligence stack.
- 6:53SPEAKER_01: And they don't just rely on Claude, they don't just rely on open AI, they can and they control everything about it.
- 6:59SPEAKER_01: And that's like, again, you know, you've looked at some of the stuff we've built.
- 7:02SPEAKER_01: It's in the same vein.
- 7:02SPEAKER_01: We're not building at the data layer, we're building more at that opera the like operating system layer for agents.
- 7:08SPEAKER_01: Um, and and the reason we do it all open source and keep it open is because it gives the user and the company control and ownership of their intelligence stack.
- 7:16SPEAKER_01: And I think the thing of the point you're making about data, what's funny about that is it's like one of the few things that everybody agrees on.
- 7:23SPEAKER_01: Like I literally just saw you know a bunch of tweets from Trump, regardless of what you think about him, talking about ownership in these companies that are making these models, like wanting to take a stake in the companies because they've used so much data of all of us.
- 7:38SPEAKER_01: And then Bernie Sanders over here on the other complete other side of the spectrum, you know, talking about a sovereign wealth fund to control like 50% of these AI comp top you know world model companies like OpenAI and Anthropic.
- 7:51SPEAKER_01: Now, I don't necessarily agree with either of them and like how they're gonna go about it, but when you have everybody agreeing that this is a bad thing that's happening, you know, you know you're kind of on to something directionally.
- 8:01SPEAKER_01: I I want to dig into Shelby and better understand actually how it works, like in how people contribute data to it.
- 8:11SPEAKER_01: Um, because right now all my data is just on the internet, like in your right, GPT Claude.
- 8:16SPEAKER_01: They just have read every blog article I wrote.
- 8:18SPEAKER_01: They know my whole LinkedIn profile, like every post I've ever made, probably.
- 8:22SPEAKER_01: So, you know, is it really just for enterprise or is it for me personally as an individual user?
- 8:27SPEAKER_01: Or like how do you view the problem of like all this data has already been absorbed?
- 8:31SPEAKER_01: I guess I'm I'm creating more data.
- 8:33SPEAKER_01: Um, and I I have so many questions.
- 8:35SPEAKER_01: This is like what the these are the like the things flowing through my head when I think about what what you're you're talking about.
- 8:41SPEAKER_00: 100%.
- 8:42SPEAKER_00: I I think it's uh the the challenge here is that AI is now one of those like niche technologies, right?
- 8:47SPEAKER_00: It affects every part of the industry.
- 8:48SPEAKER_00: So all the things you bring up are extremely relevant.
- 8:51SPEAKER_00: For individuals, every day you and I make decisions.
- 8:54SPEAKER_00: You and I expose kind of our our weights in in different circumstances, and those weights evolve and change over time.
- 8:59SPEAKER_00: And that's something that, you know, luckily, I don't think has been totally captured by AI, right?
- 9:03SPEAKER_00: Some of the public stuff has, for sure, on anything we post publicly on LinkedIn, Facebook, uh, etc.
- 9:09SPEAKER_00: But the maybe the decision of how, you know, I made a decision, you know, what to eat for dinner tonight, why?
- 9:14SPEAKER_00: Why did I make the decision?
- 9:15SPEAKER_00: Or, you know, why did I decide my wife and I decided to do on date night, or you know, other things like that.
- 9:18SPEAKER_00: So those are things that are still quite private.
- 9:21SPEAKER_00: But I would say that and enterprises are are right now trading off productivity gains for kind of this um kind of corporate ownership of data.
- 9:30SPEAKER_00: So the same challenges there, just in different forms.
- 9:34SPEAKER_00: And of course, for an AI lab, uh, for for someone generating models, they they're they want to get access to all that data.
- 9:40SPEAKER_00: So I do think there'll be definitely regulate, like regulation will kind of force them to say, like, you need to kind of get consent and ownership and and permission from users for enterprises to use that data.
- 9:50SPEAKER_00: There already are already, if you use like the API services, they're not supposed to use that data for training, as you know.
- 9:55SPEAKER_00: But you know, who knows what happens behind the scenes?
- 9:56SPEAKER_00: Hopefully they don't.
- 9:57SPEAKER_00: Um maybe there's ways around that.
- 9:58SPEAKER_00: I don't know.
- 9:59SPEAKER_00: But uh, but I'd say those are starts.
- 10:01SPEAKER_00: But the the larger point of the picture is that we can actually start to enforce this technologically within the product, where the product itself just can't uh have direct access to data.
- 10:11SPEAKER_00: Maybe it trains on data that, okay, like very simple situation, right?
- 10:14SPEAKER_00: And not not too not too crazy.
- 10:16SPEAKER_00: Our data goes into a storage infrastructure, it cannot be kind of read without our permission.
- 10:21SPEAKER_00: If it is read, it's gonna be read through a secure binary running on a trusted execution environment that is encrypting the data from the source to the kind of training or inference and then never decrypted in public.
- 10:35SPEAKER_00: That's how like a secure kind of infrastructure.
- 10:37SPEAKER_00: That provides a nice, a nice place in which your data is never gonna be exposed externally.
- 10:42SPEAKER_00: And maybe you make it quantum proof so that uh if you're really paranoid, so that you know this is well protected for well beyond your lifetime.
- 10:48SPEAKER_00: That kind of thing I think is is a very simple solution uh to this kind of problem of kind of allowing the data to flow freely uh throughout the system.
- 10:56SPEAKER_00: And now if you have a system like that where both consumer, personal, and enterprise data is all being secured, uh, maybe with different safeguards, maybe with different levels of replication, different kind of performance tiers or whatever, this and that, I think that's a pretty solid outcome for the way that things should be compared to what they are today.
- 11:14SPEAKER_00: And and so Shelby provides that.
- 11:17SPEAKER_00: Shelby's gonna, in its full form, will provide a lot of that infrastructure, yeah.
- 11:20SPEAKER_01: Very cool.
- 11:21SPEAKER_01: And so what it what does it look like today?
- 11:22SPEAKER_01: Like how do I, you know, as a as a user or customer or somebody wants to adopt it, what does that look like?
- 11:28SPEAKER_00: Today, Shelby is extremely performant and uh in a testnet scenario.
- 11:32SPEAKER_00: So if you just you know get testnet access, you can take a look at it.
- 11:35SPEAKER_00: Mainnet will be coming very soon.
- 11:37SPEAKER_00: Or not Mainnet, but rather like a kind of a a more public instance will come soon, uh, maybe call it a beta of some sort.
- 11:44SPEAKER_00: And I think that will be starting to have onboard its first enterprise customers uh to putting data on-chain and very highly performant, showing that you can actually get cheaper costs to hyperscalers, especially due to the egress uh uh reduction of costs and global access to the data around the world in a kind of a global namespace.
- 12:00SPEAKER_00: But the security and the uh AI specific features will be coming, I think, very shortly.
- 12:05SPEAKER_00: And we'll start to see like the ability for this kind of full encryption cycle and protection cycle of that data from ingestion to usage uh to inference be, I think, um quickly implemented and supported without the system.
- 12:21SPEAKER_01: And how how does like Aptos, the network, connect to this?
- 12:24SPEAKER_01: Is it incentivized?
- 12:26SPEAKER_01: Is there like something you have to pay?
- 12:27SPEAKER_01: How does how does it work?
- 12:29SPEAKER_00: So the way it works is is essentially there's there's fees for for storage and there's fees for kind of uh reading data, like the same kind of similar structure you would think of from a hyperscaler.
- 12:39SPEAKER_00: And some of those fees will be used for transaction fees in the Aptos side and and and use APT.
- 12:44SPEAKER_00: And then the other set of those fees uh are gonna be used to you know power the Shelby token.
- 12:48SPEAKER_00: But I think the main thing here is that um the cost to a user will be will be low and also more important, proably safe and secure.
- 12:58SPEAKER_00: And have build a generate confidence that your data is your data.
- 13:02SPEAKER_00: So there's two pieces of that.
- 13:03SPEAKER_00: One is gonna be around like, is your data kind of private?
- 13:06SPEAKER_00: Is it secured?
- 13:07SPEAKER_00: Of course, like those things are important into encryption, the levels of security of that encryption to be kind of forward uh forward secrecy proof.
- 13:14SPEAKER_00: But there's also the other point of like, is my data always available?
- 13:17SPEAKER_00: Right.
- 13:18SPEAKER_00: So in a situation in which a government can kind of shut down a provider of storage or like heavily control access to it, I think you know, your data might be secured in some sense, but it might not be available.
- 13:30SPEAKER_00: So that I think is a pretty bad situation.
- 13:32SPEAKER_00: You can imagine, you know, it's like our collective intelligence that floats uh in our brains and across different storage platforms.
- 13:39SPEAKER_00: Now, like if it's not available, it becomes like you're having amnesia, for example, uh in the car because you forgot everything that happened to you in the past.
- 13:46SPEAKER_00: That I think also is a fundamental right to have access to your data all the time, as you said.
- 13:50SPEAKER_00: Fundamental right to open intelligence is a fundamental right also to have access to my intelligence when I when I deem uh appropriate.
- 13:56SPEAKER_00: And this is something that's important for all humanity.
- 13:58SPEAKER_00: And I think this is where something like a system like Shelby can be an important part of that stack, powering that infrastructure and supporting that particular primitive that no other one can.
- 14:07SPEAKER_01: And what are some of like your initial use cases you guys are looking at?
- 14:10SPEAKER_01: Sounds like you might have some enterprises kind of in the pipeline to be early adopters.
- 14:15SPEAKER_01: Can you talk about any of those examples?
- 14:17SPEAKER_01: Sure.
- 14:18SPEAKER_00: So there are a lot of there are a lot of parties that are very interested in this space for different reasons.
- 14:22SPEAKER_00: Some are interested in the data collection side of things.
- 14:24SPEAKER_00: And so uh you've probably seen companies like Scaleai that um source data from experts that also need to store that data, provide labeling on top of that data as also payment and maybe even royalties potentially in multiple revenue stream splits.
- 14:36SPEAKER_00: I think that's one set of customers that we've seen is very important uh for the product because their CapEx costs tend to be quite high when they use hyperscalers, and Shelby's a great alternative and it can also provide a global solution.
- 14:47SPEAKER_00: So if, for example, you've got cameras watching your hands, not in one place in the world, but many, and you have labelers of that data, human labelers uh operating all over the world, you don't have to worry about these massive eagers fees that are kind of coming from your request to a single storage solution.
- 15:01SPEAKER_00: I think Shelby can handle that much better with cheaper costs.
- 15:04SPEAKER_00: And we have a number of partnerships in place with uh with entities that are very interested in that particular use case, uh, which is a it's a big and growing business.
- 15:10SPEAKER_00: Um all that is kind of the under underlies, I think, a lot of the collective intelligence of frontier models today.
- 15:16SPEAKER_00: Then there is uh other use cases around people who are looking for the decentralized aspect of things.
- 15:22SPEAKER_00: Uh so even you can imagine today's prediction markets are pretty straightforward in terms of uh operating on price or smaller feeds of data that can fit on chain, but more advanced analytics of data uh that you want to see keep secure.
- 15:35SPEAKER_00: If you want to, for instance, take the World Cup, uh, which hopefully uh you've been watching the US do fairly well in uh this year and also many others.
- 15:42SPEAKER_00: But um, if we want to bet on other statistics or um take wages on other statistics besides just the core score, maybe derived metrics, I think Shelby would be a great uh great use case for kind of storing larger sets of data and then providing more advanced kind of uh prediction markets for different types of consumers.
- 15:59SPEAKER_00: Uh, we also see the use case of Shelby being important for um different kinds of systems around even ads.
- 16:05SPEAKER_00: Uh ads, transparency in ads and fraud in ads is a big challenge in the space.
- 16:09SPEAKER_00: And there are several companies have approached us in terms of providing more transparency to add impressions, ad clicks that weren't possible in the past.
- 16:16SPEAKER_00: So I would say we're we're we're excited to see like a number of uh large corporations and and even smaller entities kind of launching a Shelby as soon as it's available and live uh in production.
- 16:26SPEAKER_00: That's something we're we're we're very focused on pushing forward in the next uh next one to two months.
- 16:31SPEAKER_01: Well, one thing I've noticed from the you know being in crypto since 2016 is marketplace type activity usually tends to do pretty well.
- 16:40SPEAKER_01: And it sounds like you've got some of that in the makings, right?
- 16:44SPEAKER_01: Like you want some people to come and build services that can monetize data that need these guarantees that Shelby has.
- 16:51SPEAKER_01: And then other people might use it because they want to store their data there securely.
- 16:55SPEAKER_01: Like I'm an enterprise, I'm running, you know, let's say we install our agentic operating system, Corbett's for somebody, and they want their data to live somewhere securely that's cheaper, doesn't have the data ingress fees and all that stuff, and and we could layer it on top um for them.
- 17:12SPEAKER_01: And that would be like a great use case.
- 17:14SPEAKER_00: Yeah.
- 17:15SPEAKER_01: I I'm curious to hear more about the decentralization aspect.
- 17:21SPEAKER_01: Like who's running the network?
- 17:23SPEAKER_01: Is it a bunch of individuals or how is it architected?
- 17:26SPEAKER_00: So the the idea is that the storage providers should be should be across a large number of entities, but they should be enterprise grade.
- 17:35SPEAKER_00: They, you know, I think it's very hard to many times people try to build storage systems across like people's random storage, excess storage capacity, you know, on a desktop or laptop.
- 17:44SPEAKER_00: And I think that just unfortunately does not provide very good SLAs and it's hard to build a business on top of that.
- 17:50SPEAKER_00: It's very hard to build a business on top of that.
- 17:51SPEAKER_00: I think most of those systems have been relegated to niche products and niche use cases.
- 17:56SPEAKER_00: And then they're useful for some things, like City at home is a great example of that, right?
- 17:59SPEAKER_00: And then a couple others, but but they're not gonna be used for kind of these enterprise grade AI applications and use cases for production.
- 18:06SPEAKER_00: So it's gonna be providers of storage that can that can demonstrate high quality uptime, they can demonstrate the ability to run enterprise grade.
- 18:15SPEAKER_00: And I and you know, you may ask, is that is that possible?
- 18:18SPEAKER_00: Is that something that's feasible?
- 18:19SPEAKER_00: Um, and I think the answer we've seen already in the Aptos blockchain is yes.
- 18:22SPEAKER_00: 100%.
- 18:23SPEAKER_00: Aptos blockchain infrastructure is extremely reliable, extremely you know, production great.
- 18:27SPEAKER_00: Uh, we we we've announced our our P99 is like really is like sub one second now, like P99, right?
- 18:35SPEAKER_00: You know, sub half a second, but the P99 is like under us under a second now.
- 18:39SPEAKER_00: Like, and that's that's insane, right?
- 18:41SPEAKER_00: That that's that's a really fast, reliable infrastructure that I think many people thought crypto and blockchain could not achieve with decentralization, but it's possible, right?
- 18:50SPEAKER_00: And you can achieve the same thing as we build on top of that infrastructure, now even more reliably, even on top of that, because the storage infrastructure doesn't need to rely on blockchain more than kind of a coordination layer, right?
- 19:01SPEAKER_00: And it's a coordination layer, I think there's even some forgiveness and fault tolerance that exists for that.
- 19:05SPEAKER_00: But building on top of a core layer like Aptos allows us to have very, very high confidence that something like Shelby can be, you know, enterprise grade and support the the most difficult use cases in the world.
- 19:15SPEAKER_01: So does it inherit Aptos' security?
- 19:19SPEAKER_01: Like is that how it works?
- 19:20SPEAKER_00: It inherits, yeah, it definitely inherits your Aptos security, but it'll often also have to build things on top of it because the data problem is different than the blockchain problem.
- 19:28SPEAKER_00: The the existed the thing with blockchains, they can never support large amounts of data on top of them.
- 19:32SPEAKER_00: They just it's too expensive, right?
- 19:34SPEAKER_00: You you imagine like it's passed through too.
- 19:36SPEAKER_01: Like you're just validating a transaction and that's bit that's basically it.
- 19:40SPEAKER_01: That's right.
- 19:41SPEAKER_01: So that's what they're good for.
- 19:42SPEAKER_01: Yeah, yeah, exactly.
- 19:42SPEAKER_00: Value transaction is really good for asset exchange.
- 19:44SPEAKER_00: That's pretty much the only thing it's good for.
- 19:46SPEAKER_00: Um, and that's something I think the Aptos is is really suited for.
- 19:49SPEAKER_00: Is the um we didn't talk about that as well, but that's I think the the core piece of you know what we believe that blockchains are good for in terms of asset exchange.
- 19:56SPEAKER_00: And then in terms of coordination there, I think a blockchain can be an amazing kind of replacement for kind of global coordination systems that have been built in the past.
- 20:04SPEAKER_00: And that's what we were seeing with Shelby and uh both in both in the kind of production use cases, but even our experiments to see that you know Aptos works extra, I mean Shelby works extremely well built in the system, very, very reliable.
- 20:14SPEAKER_00: And I think for a system that needs to be up all the time, needs to be available all the time, and provide access to your collective intelligence all the time, it's it's got to be fitting those capabilities.
- 20:23SPEAKER_01: It's really interesting the conversation around enterprise grade infrastructure in a decentralized world.
- 20:30SPEAKER_01: Because, and I'm I'm a nerd about this.
- 20:32SPEAKER_01: Like when I you know, my ever since I joined crypto, I've been fascinated at like where the where the data centers and the servers live and the people running them, and how do you like bootstrap these decentralized networks?
- 20:43SPEAKER_01: And I would I would argue just thinking about it where you're talking, the only network that really has had the ability, I say had, because I do think it's not in reality this way anymore, to have non-professional actors contributing to it is Ethereum.
- 20:59SPEAKER_01: I don't even think you can say Bitcoin anymore because Bitcoin's gotten so big that really it is just a bunch of big mining companies.
- 21:06SPEAKER_01: That it's not there are not that many individual people who contribute hash hash power.
- 21:11SPEAKER_01: They just can't.
- 21:11SPEAKER_01: It's not profitable.
- 21:13SPEAKER_01: And I think the same is kind of true in Ethereum now, too.
- 21:15SPEAKER_01: Although I think anybody can just run a client still and connect to it.
- 21:19SPEAKER_01: Um, but they but again, they still need a stake and they and they're not gonna get it.
- 21:22SPEAKER_01: That's right.
- 21:23SPEAKER_01: And Solana ended up being that way too.
- 21:24SPEAKER_01: Like it's primarily professional operators.
- 21:27SPEAKER_01: And some of the operators that got in early are small, like one or two main shops, but they are pros, right?
- 21:33SPEAKER_01: Like they are running enterprise grade bare metal in a data center and they know what they're doing, right?
- 21:39SPEAKER_01: Otherwise, they wouldn't have the stake and they would lose the stake.
- 21:41SPEAKER_01: And the small operators that have maintained their stake in that ecosystem, it's because they're so damn good at doing their job.
- 21:48SPEAKER_01: But there just aren't that it's hundreds, right?
- 21:50SPEAKER_01: It's like not tens of thousands.
- 21:52SPEAKER_01: And you know, the node count on Ethereum is not actually tens of thousands of individual operators.
- 21:57SPEAKER_01: It's I don't know, it's it's not it's not that many.
- 22:00SPEAKER_01: Know how many it really is, but it's not tens of thousands, you know, of people.
- 22:03SPEAKER_01: So I largely you've seen this consolidation in who runs these networks.
- 22:08SPEAKER_01: And that's okay.
- 22:09SPEAKER_01: Like I don't think it's a bad thing.
- 22:10SPEAKER_01: I think some people would think it's a bad thing, but I don't.
- 22:13SPEAKER_01: I think it's just the way it is.
- 22:14SPEAKER_01: It's just where product market fit came from.
- 22:16SPEAKER_01: I don't know if you have any thoughts about that.
- 22:17SPEAKER_00: But ideologically, right, it'd be great if more people could do that.
- 22:21SPEAKER_00: But practically, like productivity is uh again, if you think about the productivity trade-off we're making for AI, right?
- 22:26SPEAKER_00: We're giving up our data, we're giving up our all our inside thoughts to get productivity wins.
- 22:30SPEAKER_00: And I think that's no different than any other industry.
- 22:32SPEAKER_00: So if if crypto infrastructure can't run on this like ideal mindset of everyone can run a node, I think it's just gonna consolidate the way you said.
- 22:40SPEAKER_00: And um, and that's just the way the economics and and capitalism works.
- 22:44SPEAKER_00: And you know, as long as we can maintain the core principles, right?
- 22:47SPEAKER_00: The core principle is that no one entity can take down this infrastructure.
- 22:51SPEAKER_00: I think that's the most important.
- 22:52SPEAKER_00: Right.
- 22:52SPEAKER_00: Right.
- 22:53SPEAKER_00: Uh and if we can maintain that, that that's the that's the thing that's gonna help this infrastructure win and um and support things that just aren't possible in traditional, like kind of enterprise software, then I think that's that's a huge that's a huge thing.
- 23:04SPEAKER_00: It's not small.
- 23:06SPEAKER_01: Yeah, I agree.
- 23:07SPEAKER_01: I'm curious what your take is on the market without talking about token price.
- 23:12SPEAKER_01: I don't really care about token price.
- 23:14SPEAKER_01: I more care about like just the general trend with the market.
- 23:17SPEAKER_01: Um, and this is something that's been on my mind personally, like both as an individual but also as a founder of an agentic open source software company, right?
- 23:25SPEAKER_01: Like crypto seems to have like decoral, decoupled from the stock the stock market in some ways.
- 23:33SPEAKER_01: And at the same time, it I'm having like PTSD from multiple bear market, bull market cycles in crypto, and I'm looking at what's happening in AI and I'm like, are we about to hit 2000?
- 23:47SPEAKER_01: You know, like the year 2000 in this space.
- 23:50SPEAKER_01: And I don't know, how are you wrapping your head around that?
- 23:52SPEAKER_01: Because I think you're really well positioned with the product, and uh, you know, you guys are well capitalized, like you will survive long enough to get to product market fit, and this is I think you're on the right track.
- 23:59SPEAKER_01: But just how are you thinking about that just at a high level?
- 24:02SPEAKER_00: Yeah, I mean, I think we kind of made this prediction like more than a year ago when we we kind of talked about our focus on two things, which is decibel and Shelby, and said these are the two core areas of of use case for blockchain technology.
- 24:15SPEAKER_00: One on this on the decibel side, we think markets, exchange, exchange of assets is the most important and building a very powerful uh on-chain exchange.
- 24:23SPEAKER_00: We want to prove it could be done.
- 24:24SPEAKER_00: It's it's done, it's in production today.
- 24:26SPEAKER_00: 25 millisecond block times on par with uh some of the fastest exchanges with extremely tight P99 latencies, operating on 50 plus markets.
- 24:35SPEAKER_00: Uh, you can see our dashboards are all everything runs in public, fully on-chain, you know, fully decentralized.
- 24:40SPEAKER_00: Everyone takes shortcuts in that space, whether it's closing, closing source and code, or it's kind of putting some centralized aspects in terms of the matching engine or other parts of the system, uh, because it's it's a very hard problem to solve.
- 24:52SPEAKER_00: We solved it, we did it, it's live, it's working really well.
- 24:55SPEAKER_01: Where's the dashboard?
- 24:56SPEAKER_01: Where can I go check that out?
- 24:57SPEAKER_00: Uh yeah, you can I think it's uh you can go to decibel trade.trade.
- 25:01SPEAKER_00: Um that's the product.
- 25:02SPEAKER_00: Um and it's if you're not, well, if you're in the US, it's gonna be, of course, not available to you, but outside the US it's um it's it's pretty available.
- 25:10SPEAKER_00: So yeah, live, running, and uh, and and and doing well and adding new markets every every every week.
- 25:16SPEAKER_01: Very cool.
- 25:16SPEAKER_00: Um so that that just proved that you know that can be done.
- 25:19SPEAKER_00: And also it's it's it's in proof in the pudding.
- 25:20SPEAKER_00: We've done over four billion in perpetuous volume in four months and continue to grow very rapidly in terms of assets and also onboarding and traffic.
- 25:28SPEAKER_00: I think there's a huge opportunity in that space when you think about AI, right?
- 25:31SPEAKER_00: People talk about agentic payments.
- 25:33SPEAKER_00: Agentic payments is another another form of trading, of course, but agentic trading, I think, is gonna be uh even even bigger, to be honest.
- 25:39SPEAKER_00: And that's an area that I think we'll we'll see innovation with things like decibel because any agent can easily integrate into it, can start trading without uh without any kind of hurdles, and uh access to global markets in basically you know the time it takes for cloud to kind of whip up an integration, which is not long at all.
- 25:56SPEAKER_01: Probably what what do you think is the state like you know, I've been watching the agentic commerce space for a while.
- 26:01SPEAKER_01: Um, you know, we pivoted my whole company to agentic commerce back in May.
- 26:04SPEAKER_01: And then now we've settled on this agentic operating system.
- 26:07SPEAKER_01: And part of the reason why we why we've landed here where we are is because it's the thing that we built this agentic operating system for ourselves, and then other people wanted it, and we started getting customers in traction.
- 26:19SPEAKER_01: Whereas like the agentic commerce space, we did about, I don't know, maybe 15 million transactions on mostly on Solana with with our product Fairmeter.
- 26:26SPEAKER_01: Um, but the space just seems super early.
- 26:28SPEAKER_01: And you know, as a startup for us, like AWS, Solana Foundation, all these people like launching competitive products to what we built, like they can keep shipping and experimenting in perpetuity as a startup.
- 26:38SPEAKER_01: We have to find product market fit and revenue.
- 26:40SPEAKER_01: And so I just like I thought it's still pretty early for agentic commerce.
- 26:46SPEAKER_01: I think that I agree with you on the trading side.
- 26:49SPEAKER_01: And I'm curious what your sort of perspective is on like the state of the union for agentic on-chain activity and like subcategory of trading.
- 26:57SPEAKER_01: Like, where are we on the map of that journey?
- 27:00SPEAKER_01: I think super early, right?
- 27:02SPEAKER_00: I I think that everyone is looking at these these areas, like everyone's adding MCP servers to their product, right?
- 27:08SPEAKER_00: That's a very healthy uh starting point.
- 27:10SPEAKER_00: But the the challenge here is I think this is where the uh decentralized exchange community can work a lot quicker than centralized ones, uh, because those will be limited to, I think, also their own products.
- 27:22SPEAKER_00: Uh, and I think what we want to see is kind of almost the um the full access to all products and decentralized land.
- 27:30SPEAKER_00: Uh the ability you can do things there are just amazing, right?
- 27:32SPEAKER_00: If you're if it's available to you, you can onboard with a USDC on decibel and trend and start trading immediately.
- 27:38SPEAKER_00: Like that's the very simple process.
- 27:39SPEAKER_00: And and now you've got access to so many different markets.
- 27:42SPEAKER_00: And with AI, you're gonna be able to make trades that are much more complex than you could in the past.
- 27:47SPEAKER_00: I would argue like perps in many ways was enabling technology from uh traditional markets.
- 27:52SPEAKER_00: You you before you had options and you know you can always buy spot.
- 27:55SPEAKER_00: Um, a lot of people stay away from options generally, right?
- 27:58SPEAKER_00: Most of the population who trades stocks doesn't trade options.
- 28:00SPEAKER_00: Um, but perps made that much more accessible to trade directionally with leverage and to a large, large, large, large, large much larger community.
- 28:06SPEAKER_00: In some sense, prediction markets are even more enabling to people, right?
- 28:09SPEAKER_00: It's a simpler way to trade things with trade.
- 28:11SPEAKER_00: Is US gonna win?
- 28:12SPEAKER_00: Is US gonna lose?
- 28:13SPEAKER_00: And here's kind of your odds for that.
- 28:15SPEAKER_00: Like people have been doing that in sportsbooks for a while, and I think that's also like kind of what you see in prediction markets.
- 28:20SPEAKER_00: It works well.
- 28:20SPEAKER_00: It makes it simpler for people to access those markets.
- 28:23SPEAKER_00: But I would say AI makes it even simpler, right?
- 28:25SPEAKER_00: I might have a simpler idea, right?
- 28:27SPEAKER_00: Uh we could discuss like your long USA, for example.
- 28:29SPEAKER_00: You've got the great sign in the background.
- 28:31SPEAKER_00: You know, put produce me up uh a set of trades that make sense for me, monitor those positions, and then you know, make sure that uh you know it it balances across my other portfolio assets uh in not just uh decentralized space, but across like my even my traditional uh you know retirement packages, my my bank account, all these other things that I have in the in the works.
- 28:50SPEAKER_00: Um so that those kind of powerful technologies are are still nascent.
- 28:54SPEAKER_00: We haven't seen good ones yet, but you know, in Aptos, we're we're working very hard to bring those products to light.
- 28:59SPEAKER_00: And I think there's a lot to ship soon.
- 29:01SPEAKER_01: Yeah, I think it'll be super interesting.
- 29:03SPEAKER_01: And I agree with you.
- 29:04SPEAKER_01: I've been reading, uh, you know, I've got this whole like historical background because I'm in New England and you know, I bought this stuff at a thrift shop actually down the street, which is hilarious.
- 29:11SPEAKER_01: But I'm reading this book about the Mayflower right now.
- 29:14SPEAKER_01: Really great book, super interesting.
- 29:15SPEAKER_01: I it's like going way deeper into that whole story than I ever have, but I feel like we literally just left England.
- 29:20SPEAKER_01: Like we're like, we're like a hundred miles offshore, and there's a lot of rough water in between too, as far as like what happens next.
- 29:27SPEAKER_01: Um, but I do think you know what you guys are doing is super interesting.
- 29:31SPEAKER_01: I think uh another company that I spoke to uh last year, I guess it was like, yeah, maybe seven months ago, is Nansen.
- 29:38SPEAKER_01: You know, the Alex over there is just pedal to the metal, like been an early adopter, has whole had like 70 open claws running like uh 30 days after you know OpenClaw came out.
- 29:48SPEAKER_01: And they have all the data and they're doing the right things, like using a bunch of open models and their own hardware to train their agents on how to trade.
- 29:56SPEAKER_01: And you know, that's what it's gonna take.
- 29:58SPEAKER_01: Like that kind of commitment and experimentation is what it's gonna take to build a product that really works, that does exactly what you just said and is and can be trusted with money.
- 30:07SPEAKER_01: I think that that's like a a jump that has to be made.
- 30:10SPEAKER_01: And I do think what's interesting too is there's some risk, but also for people who get into that early and start playing around with it, there's gonna it's just gonna be like early crypto, right?
- 30:18SPEAKER_01: Where there's gonna be just asymmetric upside for people who start using these products um before everybody else does.
- 30:24SPEAKER_01: I do kind of wonder what happens though, like when everyone's using agents to run their strategies, like they're just agents battling each other.
- 30:32SPEAKER_01: And is it just a model war at that point?
- 30:34SPEAKER_01: Or like does do we reach equilibrium when like all the intelligence is equal?
- 30:38SPEAKER_01: Like, how does that how does that play out from a market standpoint?
- 30:42SPEAKER_01: I think that's a little bit uncharted territory when like the intelligence operates on on-chain rails at the speed of light, and like all of it moves at the speed of light, like the thinking moves that fast too.
- 30:54SPEAKER_01: Interesting times.
- 30:55SPEAKER_01: I don't know what your thoughts are on that.
- 30:56SPEAKER_00: It is interesting, but in some sense, that part, uh I mean, I think there are there are interesting trade-offs there because you know, prop shops and hedge funds are already doing that today with existing markets.
- 31:06SPEAKER_00: So it'll be the the trade-off is gonna be latency versus kind of intelligence.
- 31:10SPEAKER_00: And you know, ideally, if you trade off more latency, you get more intelligence.
- 31:14SPEAKER_00: And if you want to get fast latency, you have to kind of give up on some intelligence.
- 31:18SPEAKER_00: And we'll see, we'll see very complex strategies.
- 31:21SPEAKER_00: But I think the important thing here is that while those those are all available today to like the most sophisticated prop shops and hedge funds, that'll be available to everyday people, like to you and to me, right?
- 31:31SPEAKER_00: And and to that, that I think is the most that's the most important thing.
- 31:35SPEAKER_00: We need to make sure we're working towards a role where like you know, there is a lot of this huge disparity between those that don't have access to technology and those that do, which is what we have today.
- 31:43SPEAKER_01: We're getting towards the top of the show.
- 31:44SPEAKER_01: I always ask this question, but what have I not asked you that I should have asked or that you wanted to talk about?
- 31:48SPEAKER_00: There's a lot of good questions we could talk about, but maybe you could ask me or you could ask about I think the general like AI versus crypto sentiment more directly.
- 31:55SPEAKER_00: I think that that would be an interesting question.
- 31:57SPEAKER_01: I would love that.
- 31:57SPEAKER_01: I actually was thinking, and you would, you know, I think your background is so interesting because you came from Meta.
- 32:02SPEAKER_01: And I know you were working on like a crypto project at Meta, but also Meta is just to me like a more traditional company, and you have like much stronger context and contacts to like regular companies, air quotes.
- 32:15SPEAKER_01: Um, and that came out to you earlier in the interview when you're talking about like the early adopters for Shelby, right?
- 32:21SPEAKER_01: Like that's probably your network has helped with that.
- 32:23SPEAKER_01: Um, and I I was thinking about it this morning, you know, about where the market is today and all this wealth that's going to be generated from these labs going public.
- 32:33SPEAKER_01: And I was like, none of these guys are gonna buy crypto.
- 32:36SPEAKER_01: None of them.
- 32:37SPEAKER_01: Like they don't like crypto, you know.
- 32:39SPEAKER_01: So I love that you brought that up because it was literally on my mind in probably a different framing uh today.
- 32:45SPEAKER_01: But yeah, what is what are your thoughts on that?
- 32:48SPEAKER_01: How do you think about that?
- 32:50SPEAKER_00: So, yeah, I mean, I kind of come full circle for me.
- 32:53SPEAKER_00: Like uh I I used to work, my PhD was in hyperformance, like supercomputers and high performance computing, which at the time which was really funny because it was like an older industry at the time, and people were like, well, there's maybe not much to do there.
- 33:03SPEAKER_00: Um, you know, I worked a lot on the infrastructure side and how do we kind of scale it out further?
- 33:07SPEAKER_00: How do we support thousands, thousands of machines working together?
- 33:10SPEAKER_00: And then I and then I worked at um uh Facebook more on the distributed system side.
- 33:15SPEAKER_00: How do we kind of scale this technology to supporting large amounts of analytics, processing, crunching exabytes of data per day on hundreds of thousands of machines, and then went into blockchain infrastructure as like the core uh financial asset rails of the future.
- 33:28SPEAKER_00: But now, like looking back, I and I along the way I also worked on machine learning as a part of the um the data analytics side.
- 33:34SPEAKER_00: And then if you look now, you know, hyperformance computing is all about AI.
- 33:37SPEAKER_00: And all those techniques we worked on like 20 plus years ago are used heavily uh when it comes to like uh you know zero memory copies, um optimizations for network drivers, parallelization techniques for how we can train more effectively.
- 33:50SPEAKER_00: All those techniques are very used, well used in the in the training space.
- 33:53SPEAKER_00: But all the all the work on agent memory context is like distributed systems work, which I've done in the past in Facebook.
- 33:59SPEAKER_00: So it feels very exciting to me because I think it's an area that we can really have a big impact in.
- 34:04SPEAKER_00: But also it's very scary, like I said, because of the fact that you know there, if you look at the way things are progressing, if they progress as is, I don't think society is in a very good place in a couple years.
- 34:13SPEAKER_00: Like that's actually a pretty, a pretty negative place uh for the world.
- 34:16SPEAKER_00: And so I think it's gotta be up to us to put all our forces forward towards solving this challenge of uh accessible uh intelligence to everyone, uh, making sure that ownership of the data and control of that data is is is kept to individuals, corporations, and whoever wants to have that control of their assets and their collective intelligence.
- 34:36SPEAKER_00: Uh and I think you know, if crypto is gonna really be around, uh it's gonna provide a big role in supporting that.
- 34:42SPEAKER_00: And the other role crypto is gonna play is gonna really be in this uh exchange of assets.
- 34:46SPEAKER_00: And I think that's pretty much it.
- 34:47SPEAKER_00: I agree.
- 34:48SPEAKER_00: And I think like, you know, everything beyond that is gonna be five plus years out.
- 34:51SPEAKER_00: Uh so these are the two areas that you know we're we're not just saying and and and is important, but we're putting our our our resources and our time and energy behind.
- 35:01SPEAKER_00: Um that being said, I think crypto, I mean AI is actually, in some sense, more the game-changing technology.
- 35:06SPEAKER_00: Uh, and it's hard to admit that, right, from someone who's been working in this space for so long.
- 35:10SPEAKER_00: But that is just the facts, right?
- 35:12SPEAKER_00: It is gonna affect, it's affecting every industry.
- 35:14SPEAKER_00: It's it's actually true for every industry, right?
- 35:15SPEAKER_00: So crypto is, I mean, sorry, AI is probably more important than than healthcare, you know, in some sense, you know, traditional healthcare.
- 35:20SPEAKER_00: AI is more important than um, you know, um, you know, manufacturing in some sense, right?
- 35:25SPEAKER_00: It's because it affects those technologies so much and and affects the global world so much.
- 35:29SPEAKER_00: So, you know, it's it's the bigger problem that we need to focus on solving.
- 35:32SPEAKER_00: And crypto's and you know, blockchain technology is an important part of it, uh, but it's not the only part, and there's a lot more to do beyond that.
- 35:38SPEAKER_01: I agree with you.
- 35:38SPEAKER_01: And I was listening to a podcast this morning, um, Diary of the CEO.
- 35:42SPEAKER_01: I don't know if you listened to his his show, but it was about the guy, one of these like big money managers or something, hedge fund guy or something like that, that called the dot-com bubble.
- 35:51SPEAKER_01: He was two years early, apparently.
- 35:53SPEAKER_01: And so everyone told him he like made the wrong call, but it it ended up being right.
- 35:56SPEAKER_01: And um, he was talking about the railway and how like every important technology has a bubble associated with it because everyone dumps their money into it.
- 36:03SPEAKER_01: And the same thing happened with with the railway.
- 36:06SPEAKER_01: But AI is is the biggest, both the most important technology in history and also probably the biggest bubble in history.
- 36:14SPEAKER_01: Uh, and it's you know, time will tell if it just number keeps going up or when when the party's gonna stop for a while.
- 36:20SPEAKER_01: But I had the same thought you did just now about and the same like acknowledgement this morning.
- 36:25SPEAKER_01: It's funny how like close it was to this show, but like, yeah, I mean, crypto is important.
- 36:30SPEAKER_01: There's a lot of interesting things about blockchain technology, but AI is by far and away or is the magnitude more important um and changing the world in a much bigger way.
- 36:39SPEAKER_01: And I I have hoped that the intersection of crypto and AI would bring some of that ownership and equality um and openness to the space.
- 36:48SPEAKER_01: Like that, that is my hope that that that is because I do think it's fundamentally important.
- 36:53SPEAKER_01: And and it feels like we really have short time.
- 36:56SPEAKER_01: Very short, very short time it's um it's it's measuring a few years, a few years ago.
- 37:02SPEAKER_01: Maybe maybe less.
- 37:03SPEAKER_01: I mean maybe less.
- 37:04SPEAKER_01: I really worry about I really do worry about a few people having like super intelligence and being like you know, even six months ahead, right, is is light years.
- 37:14SPEAKER_01: And that kind of I we just don't know.
- 37:16SPEAKER_01: We don't know how powerful that's gonna be.
- 37:17SPEAKER_01: It's kind of kind of hard to imagine how big of a deal that will be and how much control they will have over the world by having that advantage.
- 37:26SPEAKER_01: It's just yeah, it's scary in our and kind of Orwellian.
- 37:30SPEAKER_00: Well, I think it's also very motivating, right?
- 37:32SPEAKER_00: We've got to move quickly, right?
- 37:33SPEAKER_00: And I think as you said, like there's a lot of companies that have tried to do things in the crypto and I space and you know, with with measure limited numbers of success, honestly, because it's not obvious.
- 37:43SPEAKER_00: The intersection is not obvious, but the intersection has become very obvious to us in the last couple you know months, which is that that that ownership piece is especially important, confidentiality is important.
- 37:53SPEAKER_00: And I think blockchain is the only thing you can guarantee your block, your data will always be available to you, nothing else.
- 37:59SPEAKER_00: So, so far, this is the best solution towards kind of solving those challenges.
- 38:02SPEAKER_00: And I think uh it's a good solution.
- 38:04SPEAKER_00: Um, of course, something better that's great, but it's uh it's the only solution that we have in the world today.
- 38:09SPEAKER_00: And we've gotta focus on what we think is gonna win.
- 38:11SPEAKER_00: And I think this is the only way.
- 38:12SPEAKER_00: And you like you said, we've got to move fast.
- 38:14SPEAKER_00: It can't be something that we take years to ship.
- 38:16SPEAKER_00: That's one thing about the industry.
- 38:17SPEAKER_00: Like our boxing industry has taken years to build, build, build, build.
- 38:20SPEAKER_00: And then like the use cases are very limited today, right?
- 38:22SPEAKER_00: But now we're just seeing the Clarity Act passing, hopefully, hopefully, you know, 70% maybe passing and driving forward a wave of enterprise adoption, which is gonna be massive.
- 38:30SPEAKER_00: But you know, it's it's it takes time.
- 38:31SPEAKER_00: Whereas like AI time, we don't have any time.
- 38:33SPEAKER_00: We have no more time.
- 38:34SPEAKER_00: Yeah, uh, so we've got to move quickly here and and advance the field as fast as we can.
- 38:39SPEAKER_01: Well, Avery, thanks so much for being on the show.
- 38:40SPEAKER_01: I appreciate your time and I wish you the best of luck.
- 38:42SPEAKER_01: Let's get on again in like six months and see where we're at.
- 38:45SPEAKER_01: See if we've see if we've already been surpassed or if we're winning or not.
- 38:48SPEAKER_01: Um, I would love to check in with you again.
- 38:50SPEAKER_01: Plus cycle would be perfect to check in where where we sit.
- 38:52SPEAKER_01: Awesome.
- 38:53SPEAKER_01: Thank you.
- 38:54SPEAKER_01: Thank you.