Venice is Here to Win: How a Private AI Company Plans to Take On OpenAI and Anthropic
TRANSCRIPT
Jon:
[0:00] Ultimately Venice wants to be this mass market consumer app. And that means that most people don't have to care or know about the crypto aspect of the company. I actually think it's one of the things that makes Venice so strong in the market is that we can go to the average user and most of our users are actually not crypto people. We were attracting users from outside of the crypto sphere for the most part. And most of them don't care about the token. Some of them even actively dislike crypto and dislike tokens and don't want anything to do with it. But they They don't care as long as the product itself is good and they can use it. But what's cool is that all of them are contributing downstream to this token economy regardless.
David:
[0:43] I'm here with John, the head of strategy of Venice, and also Jesse, the CTO at Venice. John, Jesse, welcome to the show.
Jon:
[0:50] Thanks for having us. Thank you for having us.
David:
[0:51] Yeah. Pretty excited for this one. I got a ton of questions about what you guys are up to and all the nuances that I think a lot of people really want about Venice, because Venice has definitely caught the attention, of a lot of crypto investors and I think is also starting to poke at some external investors as well. So let's just get into it. Maybe we'll start with the banger question. Why is private AI important?
Jesse:
[1:14] Yeah, absolutely. It's a great question. And I think it's a pretty simple answer. You realize if you've ever put anything into an AI model that you wouldn't want published, why privacy is important. I think we founded Venice on this principle that it's somewhat dystopian that a small handful of tech companies are building databases of everybody's most intimate thoughts. And really that there needed to be an alternative. And we wanted to be that alternative and
David:
[1:39] To build it. What would happen or what is the fear of this dystopian future if we don't have private AI? Maybe you could paint a picture of just what we know from the arc of public companies in Silicon Valley or just whatever fear might be of what happens when anthropic opening AI just get a little too big, a little too powerful, and a little bit too accessible to some of our data. What's the fear here?
Jesse:
[2:04] Yeah, I think it's not necessarily either of these companies are bad or evil. I think they're great companies. They've built phenomenal models. The concern here is that as AI has become more capable, that people are putting more and more intimate details of their lives into these models. And so, like myself personally, I'm using agents for health reasons. And so my agent now has access to incredible amounts of health data on me. And when you think about where that data is stored, if it's stored in a central location, like on ChatGPT's infrastructure, there's a world in which rogue employees, there's a world in which hackers, there's a world in which the government, through means of subpoena, can begin to access that data, whether intentionally or unintentionally, on the provider's realm. so The reality is like we just need some alternative that allows individuals to have private data.
David:
[2:56] So it sounds like they're really the big deal. The thing that you guys are focusing on is like AI as a product is fantastic. One of the greatest products humans have ever created. And one of the unfortunate byproducts of that is it's going to create the world's greatest honeypot of data ever. And that's something that you guys want to solve.
Jon:
[3:16] Yeah, and I don't think a lot of people will even realize, especially, you know, the average person using like chat GPT, etc, just how much is getting hoovered up about them, you know, people already worry about things like their social media presence and their their privacy of like bad information they show online. And it's just a small subset of like, all the things that people are putting into AI without thinking about it. And if you generally like ask someone, do you would you be okay, if everything you've ever put into AI was published publicly, almost everybody would pause and be like,
Jon:
[3:47] Actually, that's a good point.
Jon:
[3:49] So that generally drives home the need for privacy. And it's also the classic, it's not about whether you have something to hide or not. This is always the classic case for privacy.
Jon:
[4:01] You want a place where you can think and not have to worry about constantly being observed.
David:
[4:07] Can we talk about the size of this market? I think as I have done my due diligence on what the private AI industry looks like, there's different companies going after this. I think none as purely and as vertically as Venice is. But maybe we can talk about just like the size of this market. Obviously, Anthropic and OpenAI and XAI, they are in the world of AI and AI inference. So is Venice, but you guys are in the private AI inference sector. Can we talk about how big this private AI sector is and whether we think it's growing? Maybe just like color in some of the details about this marketplace that you guys are in.
Jesse:
[4:46] I think the key piece here is to realize that Venice is not focused on private AI. Venice wants to be a consumer household AI brand of which privacy is a core tenet. And so in that world, we're focused on every consumer. We think we want to be positioned in a way when you think of AI, you think of ChatGBT, you think of Claude, and you think of Venice. And everybody from a household point of view is thinking about it in that way. And if we do that right, and privacy is foundational to that reality, then we've really achieved our mission. So it's not Venice focusing specifically on private AI, it's Venice focusing on all of AI with privacy at our core.
David:
[5:21] And I suppose that's how you make as privacy as ubiquitous as possible is Venice needs to swing for the fences. It needs to go toe to toe with the biggest AI labs. And it just so happens that privacy is a core tenant of Venice. Maybe not leading with privacy, but definitely proudly being private, but nonetheless leading with UX, being a good product, broad consumer adoption for all the normal reasons. Is this right?
Jesse:
[5:45] That's exactly right. I mean, I think... The biggest consumer companies in the world have not got there based on a foundation of privacy. They've got there based on the capabilities of their products. And we believe that we can have one of the best AI experiences and do it with privacy built in. And that's where we're spending our energy.
David:
[6:04] There's a challenge here for Venice in that you guys want to go toe-to-toe with the big guys. But Venice isn't. I don't think Venice is people's typical first stop on the AI train. Like if somebody is curious about AI, I think they go to probably chat GBT most frequently because it's kind of like consumer platform and then maybe Anthropic second. So how do you guys fight the biggest gorillas in the room when you guys are
David:
[6:33] trying to execute on the strategy? How do you actually be like people's first stop on the train?
Jesse:
[6:37] Yeah, I think it's a really important question. There are two realities to vendors, so two foundational pillars. One is that focus on privacy, and the second is access to unrestricted AI. So we realized when we founded the company that beyond just the dystopian nature of these data sets, there was a world in which sort of a content moderation committee at each one of these companies was deciding what the models could and could not say. And that was the second thing that really concerned us. So if you've ever used ChatGPT or you've used Claude and you get the response back that they can't answer your question, and it can be anything benign. We get this all the time from an engineering point of view, just trying to build our own software. That is the second tenant to why Venice exists. And so what we've found is that people are frustrated in that scenario. They are looking for an alternative. And so that vector... Sort of the unrestricted access to raw machine intelligence has driven a lot of traffic to Venice and we think will continue to be sort of a pivotal way that people find us.
Jon:
[7:34] Yeah, in many ways, the funnel is often someone who tries another AI product and gets a refusal of some type and often over something very benign, like not that they're trying to do anything crazy. They're just trying to do something very basic that they think they should be able to do with the AI and the AI refuses or tries to steer them elsewhere or sometimes even tells them to like go to sleep instead. Like all of these things lead to someone to look for alternatives of like, where can I have an AI experience where it's just going to do the thing I ask and not basically moralize. And I think a lot of those people end up at Venice.
Jesse:
[8:11] The other thing here that I think is really important from a consumer point of view, It's like we've seen this proliferation of models and they're all very good in their own ways. But nobody wants to have 15 different AI subscriptions in the same way that we're all frustrated that we have to have 15 different subscriptions to stream movies. And so Venice has solved that problem as well by aggregating all of the models in one consumer interface that is consistent. And so every time you come back to Venice, you don't have to go back and figure out how do I use ChatGPT's interface if I've been using Cloud for the last week. Like everything's there, your prompts and data are all there, your memories are stored locally on device in a private way, they're all there, and you can access all of these different models and capabilities from one place.
David:
[8:51] Yeah, this represents to me a core differentiation that I see with Venice versus the AI labs. The agentic chat that Venice introduced pretty recently, a couple weeks ago, I think, And the cool thing about it is that when I type in a prompt, I can see it making decisions about which model to choose. And that's not an experience that I have with OpenAI or Anthropic. I actually have to choose the model, which as a consumer, as a user, is cognitive load on me. And alternatively, the fact that it is a aggregator of many, many models is not something that I see frequently in the marketplace. And in addition to the privacy, in addition to the uncensored response from these models, there's also the aggregation element of, you guys have an agent which elects to choose a particular model to best suit my needs, which feels very aligned to me as a user and not a property that I see anywhere else in the market.
David:
[9:52] Jesse, can you talk about this just a little bit more? Yeah, absolutely.
Jesse:
[9:54] The problem we have right now is that there are hundreds of models and nobody knows what to use for what. And we have this problem internally ourselves, right? Like if you look at the original, we call it now Legacy Chat, the Venice product that existed before Argentic Chat, we have a model selector there. First of all, only roughly 20% of our users use that model selector and to see what options are there. And it can be overwhelming. Like you're given hundreds of different models. And how do I know between GLM 5.1 and Kimi 2.5? Like it's not an intuitive thing, even for folks like us that are in the industry. And so with the Gentic chat, the objective was to remove that cognitive load entirely And to provide a platform where effectively the aggregation and decisioning can happen behind the scenes, based on your prompt, the models themselves can go sort of select and route and determine not just from a text model perspective, but across all modalities. So images and video and audio and music, and have that all in a very simple and intuitive place.
Jesse:
[10:51] So instead of having to sort of aggregate and decide across all these things, it's done for you.
David:
[10:56] And I think this brings in the question of the nature of like open source versus closed source models. Venice, as I understand it, and John, I want you to check me on this, just aggregates all the value, the aggregate value of all open source models. And there's always going to, there's going to be this rat race between the capacity capabilities of open source models versus the closed source models. And maybe there's a big gap there at Terminus in the final equilibrium of AI. Maybe there's a big gap there. Maybe there's a small gap there. I don't really know. But it's nonetheless Venice's strategy to just aggregate all the value of the open source models. And maybe in aggregation, that's even maybe at par, at least at par, or even better than just being narrowly funneled into one super powerful model from OpenAI or Anthropic. John, what do you think about this?
Jon:
[11:45] Yeah, no, I think you generally have it right. I mean, the way we view it at Venice is you should have access to whatever model you want and you should have the power of all these things. And in general, the open source models are catching up quite a bit. Like the gap, like since since Venice was started, the open source models used to be closer to like, I would say, a year or more behind the frontier. And that gap has steadily started to close. So they're still behind the frontier models, but it's probably closer to like a three or four month gap today. and many of them can do just, you know, 80, 90% of what a top GPT or Anthropic model can do and often at much cheaper costs. So there's definitely a lot of value there, but ultimately like we don't want, we want our users to have access to whatever models they want. So we also wrap even closed source models. So you can use Anthropic, you can use ChatGPT through Venice and by doing that, you actually gain a level of anonymization. So it's not pure privacy, but it at least helps protect your identity a little bit because that model doesn't know anything about you or that you are the one using it at minimum. So we think you can actually get all these values and a lot of our users like to even combo those. They like to, you know, use Venice and then, you know, like an anthropic model for some coding things, but then they let it route or kick back their agent to some open source models for other types of tasks. So the fact that you can kind of get this best of all worlds in one place is,
Jon:
[13:11] I think, a big part of the value.
Jesse:
[13:13] One of the really interesting pieces here is that Like providers like XAI and Grok, we're able, Venice is able to provide Grok privately to our users, which is a very unique proposition. And so in that world, you have a state-of-the-art set of models, but it's delivered with zero data retention. So there's no storage or post-training on anything that you put in there. So not only do the closed source models exist, we can, in some cases, offer them in a private capacity as well.
David:
[13:40] Maybe you can talk a little bit more about how that actually works. So if I, as a user, go to Grok and I type in my prompt, my query, give it my dot, my data. Grok has that now. Elon has that now. But Venice is able to hook into that same model, but it represents me while also hiding who I am. Maybe talk a little bit more about how that works.
Jesse:
[14:01] Yeah, absolutely. So on the Grok side, Venice has a commercial relationship with SpaceX. And in that relationship, they have guaranteed zero data retention on any of the models. So if you go to Grok today directly and you're using that model, they do indeed store that data. They're using it for training. That's how they improve everything. That's the whole sort of value prop and why it's hooked into X, et cetera. If you come use Grok at Venice, through our commercial relationship, you're getting the same model, the same quality, the same speed for video generations and all those pieces that make it an interesting model to use, but you're getting it completely privately. So they're not storing it. They're not training on your data. And that is commercially guaranteed through our relationship with SpaceX. So that's an example of a way that we're able to work with a commercial provider and still align the value proposition of privacy.
David:
[14:50] And that's simply just because the commercial relationship is just more strong in terms of like sovereignty and protections. The commercial relationship is just stronger than the consumer relationship. And so through Venice, many, many consumers unionize in a way and use Venice's actual, the fact that it is an entity and the entity represents all of its consumers but does the whole privacy thing on behalf of its consumers.
Jesse:
[15:17] Yeah, I'm not sure I'd use the unionize referencing, but yes, we're able to aggregate the demand from our user base for that model and then the users get a net benefit from that demand.
David:
[15:29] Sure, sure, sure, sure. Can we talk about who needs private AI? So there are consumers that might prefer it, but then there are people, maybe entities, consumers that need it for whatever reasons, legally speaking, for maybe HIPAA laws. What are the consumers of Venice that actually need the level of privacy that you guys offer?
Jesse:
[15:52] Yeah, I think this was actually one of the more challenging things I didn't think about when we founded this company was understanding the customer. We don't necessarily know who's using Venice. Like we don't collect that data. That's the whole premise of the product.
Jon:
[16:06] By definition, yeah.
Jesse:
[16:06] So we can hypothesize who these folks are. And, you know, I think there is a world There are all kinds of examples we can think of around legal use cases. You know, there was a court case recently where a company CEO had used, I believe, ChatGPT to effectively figure out how to terminate a earn out for a company they had acquired. and they follow that plan. And then the gentleman of the earn out sued and was able to obtain all the records and prove that they were sort of nefariously acting to terminate the rights. So any world in which you're trying to do something that you want privacy around, and in that case, that's not a great example of using it for good. But the reality is everybody basically should, by default, be using privacy. I guess it's not a world in which if you have that option, you should elect to use something that is not private.
Jon:
[17:05] Yeah, I think a lot of people like wake up to this fact in various ways over time and then they find themselves seeking out privacy. And sometimes we just get free advertising like, you know, ChatGPT, a big news article goes up because someone was able to do a subpoena to ChatGPT and get all these records, etc. Or Andropik shared something with the government or whatever it is. Every one of these news articles tends to drive more people to look for private AI when they realize what's going on. And I think that happens very, I think there's kind of, again, like an awareness that happens that's very similar to what happened in the early days of social media. Early on in social media, nobody was thinking about privacy. They're just putting whatever out on the internet. And then over time, this became a big issue and people realized and became a lot more careful with it. And I think we're going to see the same thing and very accelerated on the AI side because again, it just grabs so much more about a person. It's so much more personal And most people just don't think about it. And when they start to think about it, it becomes very clear of like, yeah, I don't want everything I talk to about AI to be public or for someone to be able to get whenever they want.
Jesse:
[18:11] Another big use case we see often is just around medical records. You know, I think I have a family member who's been diagnosed with cancer and has been going through treatment and we get these big dumps from the doctor of test results. And it's not written in human language and it takes the doctor two or three days to interpret it into a human result. But when you're dealing with something like that, you want the answer immediately. And so to be able to take that data and put it into a model and get back at least some sense of directionally how things are going without having that then permanently logged, it's another great example of a use case where privacy first just makes sense.
David:
[18:47] Yeah, there's got to be a ton of actual laws, like HIPAA laws, which is what you're referring to, where privacy is very strictly mandated in the healthcare system to be controlled and not, and like the data that ends up on a database really matters. So that's interesting from the Venice perspective, because you can't take somebody's medical records, a doctor can't take somebody's medical records and just ask Anthropik about stuff. And so that's from the medical side, that's from the HIPAA side. There's also probably just like corporate deals and just information secrets that they just don't want to put on OpenAI's servers or they are entrusted with somebody else's data that they can't have liability over and therefore they
David:
[19:28] can't put that on OpenAI's servers. So like how big is that market? And like what about all of these like corporate rules for data protection and data custody? How much does that actually.
Jesse:
[19:40] Benefit you guys? Yeah, I mean, I think that market's enormous.
Jon:
[19:43] At Venice, we're focused on the consumer play.
Jesse:
[19:45] That's where we think we will build a differentiated business where the product itself really stands out. I think there are plenty of companies that are focused on enterprise and on building token resale agreements for enterprise customers. Certainly, we have those customers using the platform, but as we think about how do we make Venice a household name, the focus really is on specifically consumer.
David:
[20:06] What is that most proximate consumer that you guys are going after? So like the user base that is the lowest hanging fruit that is where you guys are targeting for growth. Is there a consumer identity that you guys are really pursuing?
Jon:
[20:17] Yeah, I mean, there's a number of them. And so again, back to Jesse's point, you know, it's by the nature of the business, it's really hard for us to know exactly what the consumers, who they are, unless they come and talk to us and tell us. And some of them do, which is great. But it's really a wide range of use cases. You know, some people, some people are artists and they want to use these things primarily for like image and video creation. Some people are entrepreneurs and developers and builders, and they don't want Anthropic or ChatGPT to basically have all the code, a proprietary thing that they're building and then, you know, put out something that's going to put them out of business the next day because they basically hoovered up all their code. Some people want to power agents you know obviously this has been a a very hot thing this year especially since the advent of open claw and hermes agent and all these things and these agent harnesses they want to be able to plug into a service that gives them access to all these various models and doesn't again hoover up all this personal data that they're putting in so those are just a couple examples but you know the kind of crazy thing about ai is like it touches almost everyone almost everybody has a use case if they really think about it or once they discover it. So people are using it for all sorts of things. And we can't even necessarily predict who those people will be next. Like we did not predict, I did not predict the rise of OpenClaw, you know, back in December. But then by January, it was very clear that this was a very important trend. So like, these kind of things change very quickly.
Jon:
[21:45] Our total addressable market,
Jesse:
[21:46] As we think about it, it's every consumer globally. We all have this problem, and we built a platform that solves that problem for the consumer, and now it's our job to get in touch and in front of those people.
David:
[21:58] Venice has experienced a big amount of growth recently in the last two months. And the way that we know that is we can actually see it on chain. When a new user signs up, you can see some VVV get burned, indicating that somebody is buying a package. Do we know how and why Venice had that specific amount of high growth in the last two months? Or is it kind of the same thing? Because it's private, we just don't really know how or where it happened.
Jesse:
[22:22] In the last two months in particular, we know there are sort of four things that have contributed substantially there. And it's the data we have is effectively from the models that are being used. So the first big thing that changed for us was the addition of Grok privately. Having that brought on the platform has brought a tremendous amount of new users and growth. And specifically, like the video models there have done absolutely, I mean, day after day, just substantially new generations. I think we've doubled in May the number of generations since April, and that was like doubling from March. Think about this world like Spencer Pratt, right? I think we've hit this point where kind of AI generated media is now good enough where true kind of artists and creators can build incredible content. And the models running on Venice are able to support that. And to be able to deliver like Grok in a private way is really attractive to many of those creators. The second big thing that we've done over the last 45 days is start activating Asian markets. So Venice has historically been focused on U.S. language markets or English speaking markets, I should say. And most of our clientele is in the US or in Europe.
Jesse:
[23:28] We have begun to really work with folks in starting in Korea to bring awareness to the platform there. And as you think about Venice and our supply, our inference, the objective for us is to really maximize utilization of that inference. And so Asian markets, given the difference in time zone, really fill in the gap of demand that we currently have at night. So it's sort of a net benefit to all of our Venice users. The third big thing that we changed was Agentec Chat. So that's been rolling out over the last roughly 45 days. So we slowly have been sort of A-B testing that. And then roughly 10 days ago made that the complete default experience. And it's just such a substantially better chat experience than the legacy Venice product. It's converting at 2x the rate from free to pro that the historical product was. And then people are excited about it and they're telling their friends about it. And so that's been really helpful. And then just generally speaking, I think the attention that Venice has been getting as a function of people's focus on VVV has driven additional subscriber growth on the platform, which is sort of the self-reinforcing flywheel in many ways.
Jon:
[24:37] Yeah, having been in crypto for quite a while, it's really nice to be working on a product where when the crypto side of things gathers more attention due to price or whatever, it actually reinforces the product itself. Most crypto projects do not necessarily have this of, you know, price goes up does not necessarily lead to more revenue for the actual product. But in our case, just because it raises the attention, we get this self-reinforcing loop of people actually trying the product and liking it. And then it just actually drives more revenue, which helps all of these things.
David:
[25:10] Liking the product, enjoying the product, and then buying some VVV and then
David:
[25:14] evangelizing about the product, which the VVV owner feels like they own a share. And they do own a share. Let's talk about the VVV tokenomics here. So a share of the total of a, so if I own 1% of VVV, I own 1% of Venice's total compute. I'd like to look into what that actually means. What compute does Venice actually have? Like, do you guys own data centers? Or if I own like 1% of VVV, I own 1% of the compute, but what does that even mean? How does that work?
Jon:
[25:45] Yeah, so to be clear, you don't necessarily own 1% of our compute if you own 1% of EVV. What you own is a right to mint Diem, the other token in the Venice ecosystem. And the more Diem you are able to mint, the more daily access you get because Diem gives you this $1 a day of perpetual tokenized inference. So you can always get, could go to come to Venice and use that however you want, as long as you're staking that DM. And the VVV allows you to mint that DM. In terms of where we get the compute, there's a lot of interesting questions. I'll probably throw that one to you, Jesse.
Jesse:
[26:24] Yeah, so Venice today works with approximately 15 different inference vendors from the state-of-the-art model labs to vendors that support our private inference to then GPU providers that provide us direct access to hardware to then our own physical data centers themselves, which we've begun building out in this quarter. So it's across the spectrum how we deliver that inference itself.
David:
[26:46] Is it incorrect to say that 1% of the Venice tokens owns, gives the Venice token holder claims on like 1% of the compute? Because I've seen that echoed everywhere. Is that not right?
Jon:
[26:57] Yeah, so I think what's happening there is that the VVV token has gone through some evolution since launch. So when we first launched, that was closer to how that worked in that we had a set amount of compute set aside for users. And if you had 1% of the stake, you would have 1% of that allocation of compute. But the reason that it evolved is we found that the issue was that that was a variable amount of compute right if I'm staking one day and the next day someone else stakes my amount of share might go down and what this created was issues of I don't I know how much I have today but I don't know how much I have tomorrow and especially if you're like trying to build a business or something on Venice that created all sorts of issues and this is ultimately why we ended up adding DM into the ecosystem because we wanted to solve that problem, our users basically wanted to solve two things. And they were asking us basically since the inception of EVV. One, they wanted some way to have a fixed amount of compute. And two, they wanted some way to trade the rights to that compute away if they weren't using it.
Jon:
[28:01] And so Diem actually solved both these problems. It created this system where you could effectively mint the Diem and know exactly how much compute you have today and exactly how much compute you'll have tomorrow and ongoing forward. You don't have to worry about what others in the system are doing once you have the Diem. And then two, you can do things like lend the Diem out if someone's not using it and let them pay you for that inference. So it gives you this kind of direct tokenized ownership claim to the compute itself.
David:
[28:29] And DM as a token, I think, represents the foundation of what could be financialization layer around compute. Because now that you have one DM equals $1 worth of compute per day, that's just like, it's like a building block. It's a primitive, it's an atomic unit of compute, in a sense. And now there's going to be, there can be this financialization layer blossom on top of it. is that something that Venice is explicitly trying to help grow or you guys are letting that be a little bit more emergent?
David:
[29:01] How is that financialization layer going to manifest?
Jon:
[29:04] Yeah, it's a great question. You pretty much have it exactly right. The way we view it is Diem is a kind of new financial primitive that lives on chain. And it's in many ways like a kind of this form of tokenized inference that can be used by other DeFi applications or anyone who wants to build on top of it in all sorts of interesting ways. So we are definitely leaning into that and that we want people to build on top of Diem. And we've seen quite an explosion, especially over the last month or so, of various projects that are doing exactly that in all sorts of interesting ways. And I think the use cases are almost limitless. Like it's, you know, I could probably spend this whole podcast just talking about, you know, 50 ideas I have of various apps that people could build on top of Diem, half of which are already in process. So we are definitely looking to support the teams that do that and support a whole ecosystem of people that want to build on top of Ennis.
David:
[29:58] So that has to inform... Strategy around how you guys treat the VVV token. The way that DM comes into existence is that you lock and stake VVV and then you're allowed to mint DM on the back of that. There's a bonding curve associated with that. So if more people are minting DM, then DM gets more expensive to mint. More VVV is required to mint more DM if other people have also minted more DM. And so there's got to be some sort of like economic balancing. I'm sure there's like a little bit like kind of Fed monetary policy is like, what's the right parameters? What's the right levers and buttons to press to help this financial world grow is like figuring out the tokenomics of VDV and DM.
David:
[30:46] That's got to be an actual like explicit strategy from the tokenomics side. Is that correct?
Jon:
[30:51] Yeah, absolutely. It is an explicit strategy. And it's really, well, well, the way to think about it as well, Venice does not control the supply of Diem because anyone can come up and mint more Diem or burn their Diem to get their BBB back at any time. We have to have influence over it. And so in our case, we call this the target rate. And the target rate is what influences that mint curve in many different ways. So like today, that target is set at 38,000 Diem.
Jon:
[31:23] But if that were to move up, for example, then the mint cost would drop a little bit and that would encourage people to lock up more VVV and mint more Diem. And so there is always this kind of equilibrium and really, really since August of last year when we launched Diem, the way I view it is for a number of months, the market didn't understand it. And then probably in like November, December, it became clear that the market was beginning to understand it and then OpenCloud launched and then the market really started to understand it. And you can just you can look at both the chart of obviously diem price but diem supply and see how more and more people started to mint and you get this interesting economics especially as the whole ecosystem takes off because people you basically have this market like diem itself is a market-based solution of who wants to mint at this rate and who wants to burn and unlock their tokens so the mint rate can go up but it can also come back down so like over the last you know week or so, there were some people that, you know, burned their diem and the mint rate actually came back down and that incentivized more people to then lock their VVV and send the mint rate back up a little bit. So like, it's this dynamic adjustment that's always going on. And Venice has to have influence of it because ultimately we need to make sure that we can always deliver when someone shows up for that inference that we're ready to give it to them.
David:
[32:42] So if the Federal Reserve has a mandate of price stability and full employment, Carrying that metaphor forward, I'm sure you guys don't want me to keep calling you guys the Federal Reserve. You guys don't like that. I'm sure you guys don't like that, but this is how I understand it. What is the mandate of the Venice team when it comes to the monetary supply of DM? Because there can be a hard money DM or an easy money DM. Right now, I have VVV staked. And, you know, some amount of it, not a little bit amount. And if I go and I mint DM, I'm actually kind of disappointed about how few DM I can mint because I haven't checked in a few days. Maybe the price has come down. But it seems like DM is in like a hard money era. It's like DM is pretty scarce. Of all the VVD that I have, I can only mint so much DM. And maybe the better equilibrium is a little bit of an easier money DM. How do you guys think about like your guys' mandate and strategy to find what
David:
[33:39] is the efficient frontier of like hard versus easy money when it comes to DM supply?
Jon:
[33:45] Yeah, I mean, to some degree, we have to figure this out over time, right? Like the market has to inform us. And there's basically two sides to how these economics work in the market, I would say. So there's the mint rate, which is this kind of economic layer of who wants to mint at what rate. And naturally, there should reach a point where the reason you're having that experience is because we're very near the target right now. So it should be more expensive. It should not be attractive for the average person to mint as we get closer to the target or beyond the target. But what that means generally is that the other side of things, the actual market of like being able to just go and buy Diem becomes more attractive. So like at any given time, it's either more attractive to buy Diem or it's more attractive to mint it. Like right now, we're probably more in the phase where it's more economically attractive to just buy it. But those things naturally come into equilibrium. As DM price rises because more people are buying it, then the minting becomes more attractive. Because now you're minting something that has a higher value asset that you have more reason to want to mint. And then if the mint rate comes down, the reverse can happen. So these things are always kind of in that sense. And then in terms of what we want from a Venice side, we're trying to figure that over time. So like we're looking at how much of the DM is actually getting utilized, how much, and that has been growing over time. But arguably we're at a point in time where the market is showing us that there's still demand for people to buy and still up that utilization. So that's really what we look at is like,
Jon:
[35:11] How much of the DM is being used by the market? And some of these other apps that are building on it are actually going to help us up that utilization. So there's, you know, reseller apps that people are creating where you can like buy cheap inference because someone's renting out their DM that they're not using. These things actually help raise the utilization rate. And as we see those things, we may adjust the target over time so that the mint rate rises or comes down or rises a little bit depending what we're looking for. And at that point, you as a user might be more attracted to Mint and lock up that much more VVV. What we want to make sure is that there's enough supply of these things out there that people can get it and use it in the various ways that they want.
Jon:
[35:51] Right now, I think that is accessible. You know, larger projects or larger entities that are less price sensitive about it can do it. And an important part is like, you don't have any risk really if all you do is Mint DM and use it. The risk only comes in if you decide to sell it. But if you're just a user that just wants to mint some DM, you might not be happy with that amount, but you could always still use it for a couple of days. And then you haven't lost anything, really. You know, you still have your VVV. So like some people are far less sensitive to that and they just don't care. They just want to have a bag of VVV. They just want to mint a bunch of DM and use it and they want to burn it back. Other people, you know, they're speculating. They might want to sell some or trade it or rent it out in various ways. Those people are more willing to take on those risks. But the average user doesn't have to take on that risk, which is kind of the beauty of the BBB and DM system is you don't even have to care about the price.
Jon:
[36:45] You just can get X amount of DM. You can use it and then burn it back when you're done.
David:
[36:49] What I really appreciate about this is, you know, for the last like five years of crypto, the word tokenomics has been just completely adulterated to mean just like the pie chart of the investors and the team and the airdrop supply. And there's not really any tokenomics at all. It's just the distribution. But we didn't really have anything like this to talk about a true token economy. This is a true token economy. And I appreciate that because that's kind of like what really excited about me about coming into crypto in the first place. How does this token economy fit as a puzzle piece into like the rest of the Venice strategy? Because when the token economy for DM and VVV goes up, I'm sure that turns into more inference demand on the Venice side, more signups. How does this turn into like a little bit of a flywheel as a part of the whole system?
Jon:
[37:41] Yeah, it's one of the coolest parts of this to me, you know, and as someone who, you know, spends a lot of his time focused on the token and tokenomics, it's been really awesome to see how this develops and how people are using it, both in ways we expected and in ways we did not.
Jon:
[37:55] But it's definitely the way,
Jon:
[37:57] You know, I think Jesse put it good earlier, like ultimately Venice wants to be this mass market consumer app. And that means that most people don't have to care or know about the crypto aspect of the company. I actually think it's one of the things that makes Venice so strong in the market is that we can go to the average user and most of our users are actually not crypto people. We were attracting users from outside of the crypto sphere for the most part. And most of them don't care about the token. And some of them even actively dislike crypto and dislike tokens and don't want anything to do with it. But they don't care as long as the product itself is good and they can use it. But what's cool is that all of them are contributing downstream to this token economy regardless, right? Because every time they sign up for a pro sub or they purchase credits on the platform, all these things lead downstream to VVV to get bought and burned and create this flywheel of just the more usage of the product that happens, the more it actually helps the token economy. So the bigger a product, the more of a revenue stream that Venice has, the better for all token holders who do care about the token because it creates all of these downstream buy and burn flywheels and it creates more activity which allows Venice to then get better, you know, better costs and deals on inference, which then benefit everybody.
David:
[39:18] One of the reasons why I think so many people, like me are excited about this is it's nice to have cool tokens that do things. And so Venice has earned itself a decent amount of VVV buyers and investors. There isn't too much clarity about what alignment between the equity entity of Venice and what the VVV token holders have rights to. It's early and things are going well. And so this question isn't particularly relevant now, but we have seen plenty of tokens in the crypto world have, cracks that appear when times are harder or as time goes on, eventually there are going to be enough VVV token holders who consider themselves investors who want just clarity around what it means to be a VVV token holder. As John, you alluded to, when there is a user sign up on the Venice platform, it burns a little bit of VVV. And this is like a tie, a binding between the equity and the token. Is there going to be, what can we say about the future of VVV token holders
David:
[40:22] and equity alignment, and if there can be any clarity about that?
Jon:
[40:26] Yeah, it's a great question. You know, the way we view it is that we actually view these things as quite in alignment, and the hope is to keep it that way. You know, ultimately, Venice is the plurality holder of the token, and we think this actually aligns us very well with the community. We all succeed when the token, you know, when the token economy is flourishing and when it's doing well. And so as a result of that, we actually think there's quite a bit of alignment. VVV itself, obviously, and Diem are really meant to be utility tokens. The value is supposed to come from what they actually do. You know, so to some degree, The equity alignment is kind of a misnomer in that the token does exactly what people
Jon:
[41:05] Think it does.
Jon:
[41:06] It actually has a useful purpose and we want it to be that way. But also we do care, obviously, about making sure that all these things stay in alignment. And at the extremes, you know, what we want is for to have a project that produces, you know, grows so much and produces so much revenue that we can effectively buy and burn every last token. You know, obviously that becomes asymptotic, but we think that strategy really
Jon:
[41:30] aligns everyone well together.
David:
[41:32] I want to talk about agents. There's this hypothetical world out there about sovereign AI agents working autonomously on the internet. I think this is even an explicit strategy, especially when we talk about DM because agents and tokens go well together. Can we talk about this hypothetical market of autonomous AI agents that need inference and how Venice is working to supply this market or what its strategy is around this market?
Jesse:
[41:55] You know, six months ago, if you'd have asked that question, everybody kind of would have looked at each other and scratched their heads and said, like, what are you talking about? And yet here we are, you know, I'm actually spending a week with a group of guys here. And we've actually got a group of agents that are all collaborating in real time, doing exactly what you're describing. They're working on new apps together as an experiment. Like, it's been a fun experiment with these guys this week. The reality is that agents need inference to operate. It's sort of like that's the air that they breathe. And somebody has to provide that inference. And so at Venice, we've thought about that in a couple of different ways. Obviously, we have the API and that's a core part of the business. That API can be purchased using X402. So you have that capability if the agent has capital available to it to be able to purchase the inference in that mechanism. You can pre-populate the API keys with fixed amount of credits or dollars or certainly put credit cards in and use auto tap up. But then I think we think it's interesting that now with Diem, sort of the agents are able to sort of autonomously purchase on-chain access to inference without sort of anybody needing to be involved.
Jesse:
[42:59] Again, given the privacy nature of the business and certainly given kind of the privacy nature of crypto networks in general, we can't necessarily tell you which agent is doing what. But the goal here was to ensure that we've got capability and optionality for those agents. And then our team has spent a bunch of energy ensuring that the API itself is friendly for new agents so that they can onboard, understand the services and the capabilities that we can deliver and interface with those in a simplistic way.
Jon:
[43:26] Yeah, kind of from a philosophical perspective, I like to say that, you know, for agents, inference is existential. Like, they don't exist without it. And so, it's really... From an agent's perspective, this ability to have this known perpetual amount of inference when it acquires DM and to do that in a permissionless manner, you know, which, from my perspective, is really the ethos of, you know, one of the aspects and ethos of crypto that I really care about is allowing for this kind of permissionless innovation. It's it creates this whole interesting economy. And it also, like, raises this question of, like, you know, there's there's how much we would pay for inference or how much we care about inference as humans when we're using AI. And then from the agent's perspective, what are they willing to pay for existence?
Jon:
[44:19] And it might be a much higher price than what we're willing to pay.
David:
[44:22] As I dove deeper into Venice, I think what I'm reading between the lines, Venice is really trying to protocolize its product offering. Permissionless inference seems to be like a core tenant of Venice. And what I mean by protocolize is that it speaks agents' languages, like inputs and outputs, API, tokens, X402. It really treats agents as like first class citizens. And I'm guessing on the Venice side of things, when something calls the API, the Venice API, you guys actually don't know if it's an agent or if it's a human, the other end of things. And that actually benefits the agents because the agents, again, you guys are meeting agents with where they are at. How does this compare to any alternatives in the market? So like everyone's talking about agentic commerce, you know, agentic inference, like Tempo and ARK, they're trying to be the agentic stablecoin layer. Everyone's talking about this world of agents. I don't know to what degree it's actually manifested, but how does Venice's product offering when it comes to agentic inference and serving agents,
David:
[45:27] how does that compare to anyone else who's doing anything in this locale next to you guys?
Jesse:
[45:32] Yeah, if we go back to the core vision for Venice to be the mass market consumer app, agents in other ways are their consumers. So it aligns incredibly well with what we're trying to do here. We have designed the system for agents to be a first-class citizen, for the experience to be well-aligned with how they think and operate, and for everything to just simply work. And so we don't necessarily think about kind of an individual or an agent as a different consumer of the product. We just want to make sure it functionally is well-aligned. And I think what's unique to Venice that differentiates us from other sort of token reselling platforms, like at the end of the day, sort of selling tokens, it's ultimately a race to zero. Like there is not a lot of differentiation in acquiring tokens from one provider versus another. The things that really matter are price, availability, like reliability, and then sort of what capabilities are available on the API. So we can compete on all those things. And what is unique to Venice, opposed to these other token resellers in general, is that we then have on top of that, the subscription side of the business and all of the support that that user base provides to our ecosystem.
Jon:
[46:44] And that allows us to drive our volumes up. It allows us to get access to compute.
Jesse:
[46:49] And at some point, I think sort of scale is going to beget scale. Like there will be a world in the not too distant future where there are no longer sort of 50, 200. I don't know what the number is now, but like token resellers, like there will be a world in which this industry consolidates. Venice will be at the forefront of that world. I think our objective over the next six to 12 months is really to figure out how we become a leader in that class or continue to be sort of at the top of the pile. And we think then that allows us to sort of net benefit the API users, the agents in general.
Jon:
[47:21] Yeah, and just kind of how we're building things. We very much build agent first, especially since the beginning of this year and things like OpenClaw took off. Our whole goal is you should be able to point your agent, whether you're using OpenClaw or using CloudCode or Codex or whatever, you should be able to just point it at Venice and basically say, do these things. And it should be able to get all the information it needs and just figure it out. And so that's always a good test that we do internally. and a lot of our external users show up and they basically just tell their agent, go use Venice and do this thing and it just works. And that's really where the magic is.
Jesse:
[47:55] And really since the beginning of the year, kind of Venice has been agent first. Like we've had an internal mandate where everybody on the team effectively is deploying and using their own agents as effectively sort of dogfooding our own product. So we are...
Jon:
[48:11] We are in a position where we're able to optimize the experience based on that relationship.
Jesse:
[48:16] And we've been doing that for months.
David:
[48:17] As I understand it, a lot of the Venice team came from ShapeShift. Can you guys just talk about that? Because I'm just kind of curious. ShapeShift, the first company that I, maybe there was an earlier one from Eric, but the Eric's first big crypto company. And now Venice has like a lot of those same people. Can you just kind of color in those details there?
Jon:
[48:35] Yeah, there's kind of two backgrounds that, you know, I would say something like 70 or 80% of our staff come from kind of two former companies. So I co-founded Shapeshift with Eric back in 2014. And we have definitely brought in what we think are some of the cream of the crop and best people from that who we think can contribute. And that really helps to have that kind of shared knowledge of working together and shared culture. And then Jesse also had started a few former companies and has brought in some of his former employees that also have that. And so we've kind of combined these kind of two former sources of employees to kind of create a new shared culture where a lot of people have a lot of experience working with each other. And that just makes things significantly easier, significantly faster, significantly more efficient because we're not spending a ton of time getting people up to speed or learning how to work with each other.
Jesse:
[49:28] Yes, Eric and I actually went to college together. So I, at the time was running a company called Bluebox, which was a private cloud company. So we physically had hardware and data centers and we built private clouds. And so here I am a decade later sort of doing this all again now for private inference. And so over the years, I've certainly worked with dozens of phenomenal engineers, with dozens of phenomenal folks. And as we decided to build this company out, it's been lovely to be able to go back and really pull the all-star team from, it's really been 28 years of entrepreneurship. And yeah, to John's note, it's like everybody knows the pace in which I'm here to work. Everybody knows the velocity we're here to ship at. And like they're here because that's exciting and they want to be part of that endeavor.
David:
[50:14] The coming together of the product makes total sense. I didn't know that, Jesse, that you were working on just like private database stuff. But like you can look at Venice and if anyone was at all familiar with ShapeShift, you can see a lot of ShapeShift in Venice. It's just like order routing, no KYC, no sign up, just presenting the product without collecting any sort of data. You see that and then you add your private database stuff. It's like, well, this is so obvious now.
Jon:
[50:36] Yeah.
Jon:
[50:37] And I think the theme that really goes through from Shapeshift to what we're trying to do with Venice is just that focus on UX and experience. Just, you know, the magic of Shapeshift in the early days was just how easy it was. You just sent this one asset, the other asset popped out. It was so clean and magical. And people love that. Obviously, that was in the pre-dex days, but there wasn't anything like that. And similarly with Venice, the focus is just make it easy, make the experience great and make sure users have a great experience. So that's really the theme
Jon:
[51:06] that I think runs most through it.
David:
[51:08] With Shapeshift, Eric and the Shapeshift team went right up toe-to-toe with the regulators and I think the SEC? I think that was them? Maybe some entity, some three-letter agency. And I think that's what the crypto community and myself really just love about Eric is his willingness to go right up to the line and kind of like hold the Overton window to protect individual rights. Do you guys expect to fight a similar fight when it comes to regulations around AI? We haven't really seen regulations around AI yet, but like I kind of see Venice doing a very similar thing of just like holding a line for individual rights and individual protections. Do you guys have a fight coming your way?
Jon:
[51:49] I mean, we're never, we're never looking
Jesse:
[51:51] For a fight.
Jon:
[51:52] In a perfect world, the regulators would just let builders build things and the market would figure these things out and you know we wouldn't we wouldn't have to have those kind of battles and we've been very fortunate so far that like especially in the u.s especially in comparison to the u.s approach to crypto before this current administration it's been a lot more open to that. It's been actually in many ways, the best place in the world to build an AI company has been the US where others have, have been more antagonistic. That doesn't mean the winds can't change. And we're ready, of course, to fight for what we believe is the right things for, for, for customers, for human rights, for, for the product, if it comes down to that, but hopefully we can work, you know, productively within the environment and it doesn't come to that. But, you know, if it ever came down to it, of course, you know, People know Eric. People know his ethos. We're not afraid to stand up for what we think is right.
Jesse:
[52:48] I mean, one of the reasons I'm here is because Eric is one of the most principled and honest founders I've ever met. And you see that consistently in how he presents itself and how he talks about his beliefs. And it's often a rare find in entrepreneurs to be so consistently principled. So it's been an honor to be
David:
[53:06] Able to work with him. It's been an honor to have you guys on the show. I learned a lot about Venice. I think everyone from the crypto industry is excited for you guys simply just because it's it proves our side of things out as well the the value of a token the value of a token economy and and what that can do for a project so I think I can speak on everyone in the industry as we're all kind of rooting for you guys so John Jesse thanks for coming on the show and just teaching me about Venice appreciate.
Jesse:
[53:30] You having thank you yeah Bankless
David:
[53:32] Nation you guys know the deal that's it for the episode this is Frontier it's not for everyone but it is for us and we are glad you're with us.