Miguel de Vega from Nillion, welcome. Thank you, Peter. It's great being here. Excellent. We are a podcast focused on marketing, branding, in all things, web3. And Nillion is doing something quite interesting in the web3 space and promoting and marketing a set of services that's really complicated, but doing it in a very unique way. And so I wanted to really get into that. But before we do, we'd love to hear your origin story, how you as an academic got into web3 and a co-founder of Nillion. Yeah, sure. So it started a while ago. My background is in engineering. So I have a master's in telecom engineering, but also I'm quite deep into math. So I did a PhD on math. I started working for large telco companies, Nokia and Siemens. Essentially, I was attracted to the notion of network, know, distributed number of nodes working together. and something emerging from that collaboration, which is a precursor of decentralization, if you look at it from that angle. But yeah, I was really deep into the math aspects of understanding networks, trying to model them with probability theory. And that led naturally into big data, which led into AI. So later in my career, I worked in AI projects. essentially around affiliate marketing and also for banking sector, trying to recover the understanding of the legacy code that they have. But it was in 2013 when I first came across the concept of zero-knowledge proofs and cryptography, essentially. So I started working first outside of Web3 in Web2 projects around authentication and identity verification. But in 2017, that led me to Web3 because there are so many projects, particularly back then using CKP, you had Zcash, you had Monero. And so I met Andrew Masanto, who is co-founder of Hedera and Reserve and other unicorns. And essentially, we toyed with the idea of creating a privacy coin, but we ended up not doing that because we thought that the market wasn't ready for another privacy coin. I kept on working on privacy enhancing technologies, broader than just ZKP. So I delved into FHE, fully homomorphic encryption, and also multi-party computation. And later on, I think it was 2021, the summer of 2021, we met again with Androma Santo and other co-founders like Alex Page, who is current CEO, and also Andrew Yeo, CMO. And essentially, we came up with the idea of Nelion. We thought that the technology, privacy enhancing technologies, were ready to take them to production and bring to Web3 something that people were not used to, which is a new type of decentralized infrastructure that is able to work with private data, essentially. So that was the origin story of Nelion. Now, Nillion uses multiple privacy enhancing technologies. Is that correct? So multi-party computations, ZK proofs, and it's not a chain, but rather it's a set of services that all chains could potentially work with. Exactly. So those are two very important distinctions, I think. Number one, we're not a blockchain. We are an infrastructure Web3 project, though, that can interconnect or interface with blockchains. So you can, you know, the nodes can listen to both events that happen on blockchains, as well as writing back, sending transactions back to smart contracts that, you know, that are consumed by blockchains. But essentially, you can run, you can think of this as a separate infrastructure. from blockchain that could, and actually some developers are using it besides blockchain technology. So they're just building use cases that require some form of decentralization, but they don't necessarily need to end up in a smart contract as a transaction. But we also have the other type of developer that kind of access our services from a blockchain infrastructure and then wants something to be reflected back on the blockchain. So that's a very important distinction. The other one that you mentioned is that we... We don't think that there is a silver bullet in Web3. We don't think that there's one privacy enhancing technology that will elbow out the other ones and kind of rule over the other ones. You have FHE, you mentioned it, you have MPC, you have ZKP, and you have others like differential privacy, you have function secret sharing, the list goes on. And essentially different PETs are good for different use cases and applications. So that's why we built what we call the orchestration layer, which is, it is a piece of technology that is able to tap into those different PETs so that you as a developer have a wider design space because you can choose the different PETs according to your applications and needs. So essentially we're kind of widening the design space, being able to tap into every different PET. I think that's very interesting. Now I, a developer, if I know how to code and I know how to products and services on chain, but I don't know much about privacy enhancing technologies. And let's say I'm building a blackjack game. How would I know which privacy enhancing technology would make the most sense for my blackjack game? That's an excellent question. ideally, originally we were thinking about providing you with a super smart compiler that is enabled to make those decisions on your behalf. But the roadmap towards that concept is far away because that's a very ambitious concept. So what we've done is to design our architecture. in the following way. have two layers. One is comprised by what we call blind modules, and we have three right now. One is general purpose compute in a privacy-preserving way, so we call that nil-vm. Then you have another one for storage in a decentralized way, privacy-preserving way, so that's nil-db. And then we have a third one which is centered around AI, nil-ai it's called. So that's a step closer to what you want. Because it's already kind of if your application is compute intensive Then you're probably going to be using nil VM if it's AI is going to be using nil AI and so on But it's still very low level. So that's why we're building yet another level on top of that, which is you know, it is full of SDKs what we call SDKs and That's where you have the more application specific things like you have one for data analytics. You have one for LLMs and AI and agents and private racks. you have one for signatures. And so that's where, that's where you interact in a very seamless way with that SDK. you know, like if you were programming, just like if you were programming with web two applications, you have a number of endpoints, you use them. And under the hood, all those choices of mixing the right PTs in the right amounts. And, you know, all those things are, are being done transparently. so that you don't have to do that. If you do want to choose that, of course, you can move down the lower layer and be more like prescriptive as to what to use. But if you don't want to get involved in what PTs are, that's understandable, that's something that not everyone wants to do, then you access this technology from the SDK level that I mentioned. That sounds very developer friendly. Is that that design choice that you that the team made was that what were those discussions like to make it very deliberate to be developer friendly? Yeah, essentially, yeah, it's always this question between freedom of choice versus usability. And it was a data-driven decision. So we started working very early on with different, what we call founding entrepreneurs, which is developers, which are very close to us and to our project, gathering all that feedback. And it was very clear. after a few months of working with them, that essentially they want those choices made for them, most of them. And those who don't, then they want to have the freedom. But the focus went from roadmap talking about a compiler to these different layers. Another aspect that informed this decision was what I said before, that we don't really believe that there will be a single PET you know, controlling everything. initially we were thinking about developing a language, nada, it's called, that would kind of incorporate everything underneath. But those PTs are going so fast. And also the languages that they have are so specific to the different, you know, flavors of privacy enhancing technologies that it doesn't really make sense to try and map everything into the same language. So rather than thinking about, you know, monolithic language with a monolithic compiler able to cater for the different PETs. We went over to these more modular architecture where you have the different PETs inside of these different modules, which then can again be combined into SDKs. And in doing so, kind of the choices are being made for developers. And it sounds like it's the right choice that you've made. How have developers... Go ahead. Yeah, we saw like an immediate change of traction in the developer space because now we could target more general developers who are not really deep into crypto. So we could see that from one day to the next one, really. We're very happy to have made that choice. Now, Nillion is marketing a very complicated product in a space that is really unfamiliar to a lot of people. Now, in Web3, we love to say that we really value self-sovereignty and privacy, but when you look at all of the attention and mindshare, privacy projects don't get a lot of it, but it's so important. How are you? How is Nillion positioning itself in a way that seems to be working? Because it's in the privacy space, it also touches AI, and it's quite complicated with real tech underneath. But it's doing it in a way that involves normal retail people and participants to be very interested in it. Tell us about how you're positioning Nillion. Another quick question. So we also had some learning to do initially because privacy, as you mentioned, particularly like a few years ago, now it's not that much so, but a few years ago, privacy was a very weak concept in Web3. Essentially it was also linked to privacy coins, which is something very, very specific and niche. And it's actually, I don't think that it's the best use case for privacy because it is good to have a token with all the transactions publicly available for public scrutiny, even though you may have like, cryptographic guarantees that all the transactions under ZKP are secure and there's no double spending and all that. But I think like, there's a reason why Web3 started with tokens and transactions. and use cases around them in the first case, because it's the one thing where you don't really need privacy. Everything else, almost everything else, if you think about Web2, requires a lot of privacy because you stumble upon the end users who have sensitive information and they want to hide that from other users, the same for companies. So very easily you find the need for privacy. The reason why Web3... hasn't found that need is because it's been centered around those transaction use cases. And so I think making those transactions confidential is not necessarily the best use case for privacy. But it was in the mind share of Web3 people. It was like the immediate thing that came to mind when talking about privacy. So that was one reason why in the beginning we shied away a little bit from that concept of privacy. Also, because from the other side, from Web2, it's been misused as well by companies, which business model is centered around having access to your data in plain text, even though they store it in encrypted form, but they have the decryption key, so they actually see your data and process it and run whatever they want on your data without your permission. We've known, of course, some cases where they even sold that data to other companies. Again, it's been misused because those companies claim that they are private and they care about your privacy and they don't say that their privacy model is broken because it gives them access to your data. So again, it's been misused that concept from Web2. So it wasn't like a clear choice at the beginning and we went for other things like talking about confidential, confidentiality, confidential compute, talking about high value data. But in the end, and I think it kind of it's linked to AI. In the end, privacy did start to take off in Web3. Because with AI, it's very clear to understand that there is a piece missing in Web3. You can run agents that tweet, OK, that's fine, that maybe do transactions. But even with transactions, where do you store those private keys? You need something like MPC or FHE to do that. And let alone when the agents start learning about, you know, your private life so that they can become your advisors for, you know, taxes and finance and health, or when they start using tools from, from web two and they need credentials and they need API keys, all those sensitive details and information will have to be stored in a privacy preserving wave in, in web3. So it is very clear that you need that. And it's also very clear that the type of information you're going to be sharing with those agents is going to be way, way more sensitive than the things we're used to sharing in Web2. It's going to be the things that you really think about, other people, things you would only share with your therapist, all your health records and financial records. So it is really sensitive information. And it made it so, so easy, I think, to defend the case for privacy in Web3. So that we switched completely our strategy and started using actually privacy as a main word. We are a private privacy project. Yes, we are very focused on AI, but we also cover other verticals and areas. And then I think the magic came from marketing with the use of the notion of blind computer. I think it's a very descriptive term. the word blind comes from the fact that the nodes in the Nelion network do not see the information they're computing on or the information that they're storing. So it kind of, it's a form, it's a very, I think very visual way of picturing how the network works and creating that as a category so that all the projects working on what others call confidential computing or things like that, we see them as blind computers. And our particular take on that, we call it the Nelion blind computer. We've seen a lot of people resonate with that because I think it's very easy to understand, way easier than trying to explain the underlying crypto behind that concept. So yeah, I think it's a little bit of AI bringing back privacy again to Web3 as an important element, a key element, I would say. Without it, I don't think you can have decentralized AI. And then also finding a way to present it in a way that people understand the notion of what our network does and is. I think the introduction of the word blind as a theme in all of your marketing materials and positioning and descriptions was really brilliant. Because it's a word that almost everyone's familiar with. It resonates with people. It makes it a very simple way to describe a complicated tech stack underneath. And with multiple privacy enhancing technologies that that is part of the Nelion infrastructure, it makes it easy to explain. And I thought that was just a very brilliant way to do it. Now you've mentioned under the yeah. You mentioned a number of use cases that were like real world use cases. And that sounds very exciting, which brings me to the question of the key audiences that Nillion aims to serve. It feels like there's multiple audiences. So there's developer audience. There's users that aim to do something with their private data in a privacy enhancing way. and feels like a lot of Web2 maybe enterprises also. Could you maybe speak to the key audiences that Nillion is aiming to serve? Yeah, sure. So essentially, from the developer perspective, I think we have, you know, at a very high level, we have two different types of developers, the Web3 developers who want to, you know, develop from a Web chain, sorry, blockchain perspective. So they're working with smart contracts and they need to run some private computations or store some private data on decentralized, on some type of decentralized infrastructure. that's one type, then the other type would be the Web2 type of developer that for the first time can now use decentralized infrastructure to kind of combine it with their existing applications. They don't necessarily want to be using a blockchain, but they do want to be able to store data and compute on it in a decentralized way. And those two cohorts or groups of developers are quite different. It's very interesting to work with them because the Web3 developers are very deep into technology. They're very kind of back-end developer type. So they're very strong on protocols, some of them in cryptography as well. Whereas the other ones are more focused on UX and kind of also product market feed. So they're very concerned about making revenue and getting users and activating them. And I think it's, you know, I'm hoping for a day where both cohorts or types of developers merge into one, because I think both things are really important. But it's very interesting to work with those two different angles in Web3. So that's kind of the developer community. Then, of course, those developers develop applications on our infrastructure that are used by end users. That's another type of user that you mentioned, the end user. They want to store private secrets. on our technology. So some products that are being built on our technology provide you with that ability to store your private information and then allow others to run computations on those secrets and gain insights from your data without seeing your data and you monetize, you can monetize that activity. So for the first time you're in control of your asset, of your data as an asset and you can monetize it and you can get some... something in exchange for allowing other companies to run computations. But it goes beyond that. It also taps into health data and finance data and agents, know, feeding them with all that. So that's another type of user. Then you have all the people that are needed to run the infrastructure. We have a built-in blockchain, which we call Nilchain, which is a layer one. essentially is built using Cosmos SDK. And we need node operators for that, of course. Then we have the proper Nelion infrastructure, the Nelion blind computer, which also needs or requires nodes to run the different clusters because it's comprised of clusters. And then you have verifiers who verify that the data integrity hasn't been tampered with and it still remains, you know, all your data remains. in the system as it should. And then beyond that, of course, we have the on the partner side, have L1s and L2s because we wanna be seamlessly connecting to every single blockchain out there. So we have partnerships with near Arbitrum, Aptos, Rachel and others. We're building now integration for Solana. The idea is that this technology and web3 and standard blockchain technology are one. in essence, so you can use them together in combination. And then you have people who are interested in the project and they're part of the wider, either developer community or just people who want to follow the news, you know, through Telegram, know, Discord, the hackathons that we organize. So all in all, it's a very complex ecosystem that runs on, sometimes on incentives, but sometimes it doesn't because you just have developers building on top of your infrastructure, but it's a very complex one. I think it wouldn't be possible to build something like this outside of Web3. I think it has to be a Web3 project. How has the recruiting of developers into the privacy space been? My experience has been that almost everyone says that they care about privacy, but actually do very little in privacy because it's very hard. But it sounds like Million has made it a lot easier with these SDKs that they provide to developers. But what I discovered when I was in the privacy space was that many developers still had a difficult time envisioning what kind of applications they can build using privacy. Like they actually needed, they needed some help. How, how has that, has that been your experience too? Yeah, so we are in the process of automating that function, but it's not fully automated yet. There is some degree of help of onboarding that developers need and require from us. So we do have a growth team with business developers and solutions engineers, essentially. guiding developers through the process of when they first encounter Nelion technology up to when they launch and even after they launch their product, we have this support. The goal, the end goal is to automate all that so that you have a self-service process probably driven by AI. So we've already run some experiments using AI to help you actually program and use our technology, which is very exciting, I think promising, but until that becomes the default technology, we do have this manual handholding, if you wish, of developers. That is being automated as we speak, but yeah, think you need that, even if you have SDKs with proper documentation. Because in the end, developers are not used to thinking about privacy, as you mentioned. So they know about... functional features, non-functional features, maybe security is a little bit better known, but privacy is not. So they need some guidance. But I think not much. And I think it's also been made easier for them with the creation of what we call the blind batch. So we've created several scores for your project. that touch on different things, like for example, access controls or the storage of information or computation of information or integrity, verifiability, different dimensions. And then depending on how deep you want to go into each one of those dimensions by using our technology, you get a different score. The score is not meant to be something like a grade that you get, but rather like a roadmap. It shows where you are and we help you improve. But this means that you don't have to use everything across your whole application from day one. You can envision a roadmap where you start using some pieces of privacy in your application, and then you continue improving and adding more as you go, which is probably the best way of doing that. So we turned a binary decision of going all or nothing into something gradual, which I think is more sensible. And we're also seeing a lot of... People like that. Developers like that because they see their roadmap and they can picture it better in their heads rather than saying, I'm going to put everything on top of Neelion and that's it. Now, do you see a trade off there? If I were to use more privacy enhancing technologies from Nillion, what's the trade off if I'm homomorphic encryption plus ZK, plus X, plus Y? Do you see a trade off there in terms of user experience? definitely. There is a trade off in performance versus security. So again, we're making that trade off easier. example, in the context of AI, it used to be the case that either you went for all T solutions, so trusted execution environment, meaning a hardware based solution, which is very performant, but it's not as secure as let's say, FHE or MPC. or you go for MPC or FHE. But we are doing split inference, which means that you get to decide essentially what is the percentage of MPC versus hardware-based solutions that you want to throw to your mix so that it makes sense for you. Because you will achieve a certain point in the performance and security trade of space that is suitable for you. And so we're doing We're doing research on that and a collaboration with Meta and other institutions around that concept, which I think is the feature. So to provide you with a slider that you can move left or right in a seamless way to just get you the right combination, the right mix of these two opposing, it's not opposing, but it's two different flavors of PETs. So that's again, I think something that not everyone, not every competitor embraces, which is the use of TEs. We think there are use cases where they make sense. And now that we're combining them to certain degree with MPC to make them more secure, even more so. So we do see again, TEs as just another first-class citizen in the world of PETs, even though sometimes they have been seen as a lesser. option because the security is not that good. But for some use cases where you do need a lot of speed, maybe in AI, we do see that that's sometimes justified, that the use of TEs. Yeah, what I found when I worked at a competitor of Millions many years ago, we also had a partnership with Meta. At the time it was Facebook and TEEs were actually a very strong interest of theirs, which was curious because it is a lesser, not lesser respected, probably less rigorous privacy enhancing technology, yet it... There was a lot of strong interest in tees from multiple web to Fortune 500s. Yes, indeed. So we did a first collaboration with Meta, which was about improving MPC and FHE for private inference of LLMs using new techniques coming from digital signal processing. That led to a paper called Curl, but also we did another for FHE called Ripple and another one coming now called Wave. That was from that standpoint, but again, now we're continuing that collaboration to combine the use of TEs and mix those two to make TEs more secure and make MPC more performant. So it's kind of a win-win scenario where you're using the stronger privacy properties of MPC to strengthen the security of TEs, but also you're using the superior performance of TEs to speed up. NPC essentially. And I think that's something very interesting for Web2 and also for Web3, and particularly in the scope, know, within the context of AI, where you have to do a lot of computations and very fast. Now it feels like with your Web2 audience, that's more of a business development partnership approach where your messaging is very focused on strong privacy guarantees, especially since these Web2 companies have lots of regulations that they have to follow, lots of customers, and they need those strong guarantees on behalf of their users. To the Web3 audience, you've got a set of developers that are learning and need a lot of guidance on how to use privacy enhancing technologies in their developer stacks. The other Web3 audience that I'm thinking of is also the kind of the, they're interested in privacy, but don't know much about it. They're community members. How do you, I guess what, challenges have you seen, well two questions, how do you communicate the value proposition to that audience? And what challenges have you seen in trying to communicate that to the Web3 retail audience? I think one of the things that opens up their eyes is when we talk about use cases. So in Web2, it is not about unlocking new use cases. It's more about, like you said, using state of the art privacy enhancing technologies to comply with regulations, to kind of do business as usual. Because in Web2, people are used to using privacy enhancing technologies. In Web3, It's about identifying privacy or privacy enhancing technologies as disruptive innovation rather than competing narratives, I see them as continuous innovation or incremental innovation because it's more about doing the same things but in a faster way. So, okay, we're going to architect a blockchain in a modular way so it's more maintainable and you can mix different blocks and so on, or we're going to increase the TPS of a blockchain. All of that is kind of more of the same. It's not opening up new use cases, new possibilities. Privacy is different. Privacy enhancing technologies do open new possibilities for Web3. And I think that's what makes people excited about it rather than compliance or these other things from Web2. Because now you can see that you can build, for example, agents that manipulate personal data without putting that personal data on a blockchain, which would be crazy, or the credentials on a blockchain, which would be crazy. And you can see that you can do things like private DAOs and voting for DAOs, or you can do for DeFi, for example, collateralized or under collateralized as opposed to over collateralized lending, because you don't have to, you can use assets from Web2 to actually back up your lending so that you don't have to put, to get a hundred, let's say a hundred dollars in lending, you don't have to, put assets worth 150 or 200, which is what currently happens with crypto because of the volatility, but you can use Web2Assets and a credit scoring that is computed in a privacy-preserving way from your Web2Assets as the backbone for that lending operation. So that's just many examples. Also the analytics that I was mentioning before, also the custody, all things around, managing your private keys in a way that no one really has access to your keys, yet you're able to instruct the network to sign transactions on your behalf, secure messaging, collaboration of different agents. All of those use cases are what really get people in Web3 excited because it's about pushing Web3 beyond its current limitations in terms of growth. So now you can start implementing things that before were not possible. And that's because of privacy. that's the way we have, that's the way we're, you that's our angle, the way we're kind of approaching this. The challenge, you you asked about the challenges, the challenge is always to put names to those use cases, to make it, to kind of showcase them essentially. So that's why, for example, we came up with this masquerade.nillion.com. website where we show some of those use cases and kind of inspire people to think in these new directions. Because this is kind of opening like a new market. And it's always very hard to let people know or to tell people what they can build. Because the things that by definition haven't built, you know, they haven't been built before. And so there's always this element of creativity that is hard to come up with. But we hope that through examples and demos and actually people building already exciting things and hackathons and things like that, we hope that we will be able to bridge that gap and that people will start creating their own things and surprises, which we're already doing in Hacker. I want to get into the privacy plus AI category, but before I do, you've mentioned hackathons a few times. Tell us about some of the hackathons that Nillion is doing and how are you recruiting developers to these hackathons? Yeah, so essentially with every major event like in Singapore, talking to any report then to then then in then sorry in Bangkok and then now in Denver, we're always doing hackathons alongside those events. We're also doing them separately. Some some hackathons are ritual. So we have now it's global coming up. And essentially, we've seen that You know, we don't need to recruit really. It's something that is happening organically. And we see that from hackathon to hackathon, there more and more projects building on top of us. So as we become well, better known, I guess, there more developers interested in us. One marketing campaign that did, I think, matter a lot is the one in Bangkok, DevCon. So essentially it was this tuk tuk. thing that the marketing team came up with. The idea, think, is brilliant. There was a lot of competition for site events. Rather than doing that, which would be also very expensive, and become the hundredth site event in the conference, what we did is to prepare... We rented 100 tuk-tuks and made them available for people to use as free transportation between events. And of course, we branded them with Nillion. The challenge there was how to bring that physical presence into crypto Twitter. So we created games and with QR codes and things like that to bring that online. And then it has its own challenge because we had to create an Uber app essentially with a map, with the different events and the tuk tuks circling around the different events. It was a very, very good idea, I think, but it was also very well executed. And I think that made it very easy for people to become aware of what Nillion is and does, and also that led into a higher signup rate in the hack. I've heard really great things about that campaign. I'm curious internally when Charlie and the marketing team came to the development team and said, hey, we have this idea, but we need some help building this Uber app. But it's going to change whatever sprint you might be in. We really need this yesterday. What was the internal discussions like having to change priorities and Building, you know, product features to now building this, this random app for a marketing campaign. Was there a pushback? no, there wasn't. I think we're an organization that is very open to new ideas. And I think we recognize the genius of, know, Charlie's genius very early and the value of that idea. do have developers that are not very linked to a... piece of infrastructure or technology or function, they're kind of floating around and they can be used for this kind of purpose. So it's not that we had to move away people or engineers from other functions and dedicate them to these. do have that, you know, there's all the time we're creating demos and we're creating these type of things. And we're very flexible, I think, as an organization. I think... That's the reason why there wasn't any pushback. But on the contrary, we immediately saw that it was very interesting. And we went 100 % on it. I heard amazing, I didn't go to Bangkok, but I heard amazing things about the campaign and you painted Bangkok blue and it really grew the brand awareness for Nillion. So congratulations on that. That was excellent. Yeah, to the extent that sometimes I was pitching Neelion from the technical side and people were just like, oh, you're the tuk-tuk dudes. Yeah, that's it. So people kind of ended up knowing us for the tuk-tuks, which is a starting point. Once I get your attention, then I get to pitch you the rest of what we do. It was brilliant. Not only the idea, but also the execution, as I said. That's kudos to Charlie. That was amazing. You know, you bring up a really great example, you know, as you're trying to describe something from a technical perspective and then they just say, tuk tuk. It's a quick shortcut and you just, get their buy-in right away. And that's the value of marketing. It helps to create those stories and people remember stories, but they often don't forget the technical details. They don't remember the technical details or even care a lot. Yes. they sure remember stories and it's the power of storytelling that I think is really the value of marketing. Let's talk about a, let's switch to, yeah. also the jokes that that went around. there was this, this joke that we were the first, physical infrastructure web3 project with real users who are the tuk tuk, you know, tuk tuk guests or writers. Yeah. It was, it was, it's very good. It kind of creates this whole focus of attention. to it and around it that then you can exploit to actually explain the deeper concepts and the reasons why we're doing what we're doing. Yeah, it was really brilliant. Let's switch to privacy and AI. Now, Nillion is doing quite a bit in the AI space. And I saw a post recently on X where there was a demonstration of DeFi AI or DeFi. Tell us about what Nillion is doing in that space. Yeah. So essentially the first thing that points towards a need for privacy in AI is whenever you have agents doing things on chain, because then they need to be able to have, you know, private keys to sign transactions and to trade on your behalf. So that's the first step towards DeFi, you know, with FAI. And that's the first use case and application that we have. So being able to deploy an agent using one of the frameworks that you have, know, Eliza or, you know, agent kit from base or, you know, a game from virtuals or the agent framework from near. So being able to do that very easily and create an agent that is able to trade, but keeping all the private keys safe. That's the first step that we have demonstrated. And that's, think, a really important cornerstone. Now, we're seeing agents also tweet. So we have trading and tweeting, but there's so much more behind agents that we're about to see in Web3. I think we're going to start seeing agents that specialize on you and start learning things about your life to be able to provide you with advice, personal advice, career advice, financial advice, tax advice, health advice. Obviously, all that information will have to be kept safe. So that demands for this type of infrastructure for a blind computer. And then you will have, we'll see that already with agent swarms, initially start with one agent, then you have two agents, then you have a chain of agents, then you have a swarm. And then essentially you have vertical swarms, is a swarm, but kind of organized around a process so that you have kind of clear steps. We see Nelion as the infrastructure, as the... HTTPS that enables agents to communicate with one another in those swarms in a privacy-preserving way. So we are developing, like we call it the vault, but also the escrow that enables one agent, let's say agent A, to store some thoughts, some information on that escrow and then allow another agent B to consume that information and do something else with that. And you can expand this to multiple agents, kind of sharing different information. It's essentially the rails on top of which agents can collaborate and exchange information in a privacy preserving way, the HTTPS, as I said before. So, and then the final step is when this goes into the physical world, first with IoT devices installed everywhere, and then eventually with robots, which I think are kind of closer than one. might think. And then, and then yeah, the information that that those physical devices and robots are fitting into if you if they're installed in your house, obviously is going to be really, really sensitive. So again, you have the strong need for something like a blind computer to be able to incorporate that information and, know, and work like that, protect your privacy. So we start with something quite humble, which is the management of private keys. for transactions, but as you can see, this very easily expands into something bigger, which I think is at the core of, is at the front field of development in AI, also in Web2. I think Web2 is gonna require this type of infrastructure because we're seeing that agents are specializing in different actions. So you have some agent frameworks for programming, some agent frameworks for language processing, like redacting things, some agent frameworks for different tools. using different tools in the internet, like one is searching, another one is doing something else. So all these specialized agents are going to be run by different entities. And you have the need to be able to connect those entities, just like the internet connected servers in a privacy preserving way. And I think that requires decentralized infrastructure and that requires privacy built into that decentralized infrastructure. So I think all the trends are pointing towards decentralized AI but with built-in privacy as provided by one of these blind computers. Now, the marriage of privacy and AI has been something that academics have been thinking of for a long time. But now that we're seeing it in action, or there's efforts in actually making it real, it's a brand new category. How are you thinking about positioning this category to the Web3 audience, and what's the reception so far? I think we started with the concept of personalized AI. So we launched a video a while ago already, like a year ago or so, picturing a little bit what I've said here with the example of an agent that starts becoming more more intelligent as you interact with the agent, but at the same time starts demanding for more and more of your personal information. it initially gets access to your financial records and then it's helping you manage your finances, but then it's also asking access to all your emails to help you gain back the love of your life and all your communications with that person. So it is like this game, this rabbit hole where the agent starts getting more and more of yourself. to be able to provide you back more services. And it becomes really clear on that video that privacy is a mass have. It's not just a function or it's not just a feature. It is actually core. So it was well received, I think, by the Web3 community. I think people are understanding more and more that with agents it's not that much about the verifiability aspect. which is one aspect of decentralization, is mostly about this privacy aspect. It's mostly about being able to protect all that sensitive information that agents are going to be manipulating. So I think it's been well received and it's resonating with the community, particularly that first concept of personalized agent. And then the three levels of data that we defined. So level one is everything that is public. Level two is everything you're happy with sharing in web two, but level three is all these new categories of information that you're gonna be sharing with AI and agents that you're absolutely not happy with sharing, because that's gonna be really, really sensitive and it's information that could destroy your relationships, your life, your career. And there's a consensus, general consensus with people that that's a no-go. And also another angle that resonates very well with people is... cancer research and research around health, whereby all that data is locked up in silos because of regulations and because companies cannot share that information. But building that on top of a decentralized infrastructure that provides privacy guarantees and taps a lot of potential, both for end users, because then now data, for example, is portable. So I can move from countries and that data will come with me. but also for the future of research and on those different fronts. So feeling that you're helping with this technology to make progress on the forefront of cancer research and the generative illnesses and things like that is something that also resonates very good with Web3, probably because Web3 people, many of them are very strong on different ethos that are very altruistic. you know, the ethos of decentralization essentially is about removing power from centralized authorities. So this again, I think is another ethos that resonates very well with with Web3 people. And it's another angle that I think is very successful. That's great. Well, Miguel de Vega from Nillion, thank you so much. Is there any final words that you'd like to share before we end? just very quickly, maybe if you want to know more about Neelion on our website, Neelion.com on the top left, top right corner, you have all the handles, you have Twitter, you have Discord, if you want to proactively participate in the community, if you just want to watch from the distance, you have X. And also you have news there like documentation, you have docs.neelion.com and you you have news there on X as to the hackathons that are taking place and the events that we go to. So I think that's a very good starting point if you want to get involved. And of course, if you want to develop for us, take a look at the documentation at this other website that I mentioned, masquerade.nillion.com, where you have some examples that can provide inspiration. Excellent. Thank you so much. This was really, really enlightening and a really great job on what the Nillion team is doing, a killer job. good. Congrats. Cheers. Cheers.