Tim Hafner, co-founder and CEO of OpenServe. Welcome to the show. Hello, thanks for having me, Peter. So you guys were at Berlin Blockchain Week and you guys had a hackathon recently. How did it go? It went fantastic. It happened yesterday. It was an intense one day, about a 10 hour hack. We had a couple hours of workshops in the morning for all the folks who were new to our platform and tech. And then we kicked off and hacked for about 10 hours and finished demos at around 11 at night. And it was awesome, amazing. ah I think we're just kind of... entering or sort of leaving our early beta phase and entering sort of the official launch of our platform and seeing the creativity of everyone in the crowd. We had about 20 hackers and eight teams there. So people paired up or gotten threes. And yeah, I mean, the use cases that were built were really impressive. um So the focus was on... uh Am I cutting out here? My video is... You're okay. Yeah, okay. All right. So the focus for the hackathon was to build a build agentic telegram apps. So fully functional full stack agentic applications that are accessible through telegram. The thesis that's sort of our initial go to market here is to bring applications to the crypto audience where they already live. Telegram has 900 million users, if you'd believe it. And uh pretty much all of crypto lives on Telegram already. So Telegram has a super comprehensive API and can basically let you deploy a full-fledged web app inside the Telegram UI. uh Super customizable, super modular, super flexible. And that was the agenda for the hackathon. So we had people building agentic applications, what we've coined as A apps uh inside Telegram and use cases spanned from crypto to also a sort of Web2 size as well. Nice. Maybe you can use the uh hackathon yesterday as a jumping off point to help us understand what OpenServe is and does and kind of your experience in the Telegram or Tawn ecosystem. Maybe tell us about, uh I don't know, the winner, the winner yesterday um and like the app that they built. um Tell us about that. Yeah, so we haven't decided the winners yet. That'll be tonight. But I do have an inkling on what I was most excited about. uh Several of them were super exciting, but the one I was super excited about was a basically a more comprehensive open schema version of Carto uh leaderboards. So if you're not familiar with Carto leaderboards, uh it's essentially a ah sort of social listening leaderboard that measures activity for a given project. uh So it's counting who's posting about it, how many times they're posting about it, what's their engagement rate, how many comments are getting impressions. What's their sort of social score? How many followers do they have? How many smart followers do they have? ah And Kaito itself, black boxes, all of the sort of inputs for how that's decided, what the leaderboard has decided on. And they gated behind a ah super high paywall. And what someone built yesterday, these two guys, they built a... open version with sort of visible parameters and customizable parameters that was fully automated on our platform. So cross channel uh listening, so across uh X, across Telegram, across Discord, and basically scraping all of these platforms and community channels for engagement by community members, checking out, yeah, sort of similar metrics and then displaying it on a They had a web app version and a Telegram mini app version where you as a project can then customize the certain, the different parameters. So whether you want to give more weighting towards retweets or impression count or follower count. ah And then we as a project or other projects can... sign up for it, pay for it and access this and customize their parameters, uh display those parameters or sort of inform their community about what parameters they're sort of setting and what they're deciding and see the leaderboard for the engagement and use that as a system to sort of gamify engagement and community growth m and issue rewards and stuff. that's, think to me is super, super exciting that someone was able to build a fully fledged application, like fully functional, working in about 10 hours. um That rivals some of the biggest names in the space right now. That's awesome and that's venture worthy too. uh So uh I'm a paying customer of Kaido as well as Cookie3 and uh I love Kaido. I think there's so many positive things about it and I think it's actually moved kind of the crypto Twitter space forward. um And I'm really pleased to hear that they're also, they will. start monitoring activity on other channels such as TikTok and YouTube and others. Which kind of points to um getting your voice heard on Twitter is so hard. um And Kaido has helped kind of gamify that and kind of um tokenize attention is kind of what they talk about. And that's really cool that this project and the hackathon yesterday. You know, they're creating an app in that space. I've often thought about, you know, these private chat rooms and telegram, private groups, et cetera. um There's so much data there that I, and I don't know if that's being captured by Kaido. um So that's really cool that this team is doing that um because that could definitely inform, you know, I could see a situation where there's Project has no attention on crypto Twitter, but is talked about a ton in these Telegram groups. so which would be very interesting. Yeah, totally. mean, Twitter is obviously sort of the main stage of the crypto space, but an under sort of hidden and underrated place is the telegram. That's where the real communities are built. That's where the real real time engagement happens. And I'd say it's as big an opportunity um as well to incentivize sort of, yeah, sort of positive active engagement inside your telegram community. That's how you build the community that then sort of takes the next step to get sort of leverage Twitter as a platform as well. Yeah. Well, let's talk about the product, uh OpenServe. um Tell us about this platform. What was the problem you guys were seeking to solve, and how do you do it with OpenServe? Yeah. So the main problem that we've been seeking to solve is to eliminate the complexity in um sort of agentic app creation. we see apps as a relic of or traditional apps as a relic of the past. So apps where you're still doing the work, you're still, you know, you have a goal or task in mind and you're navigating across multiple different tabs, web apps, mobile apps to sort of get to your end result. um And we're envisioning and creating a future of agentic apps, so apps that aggregate experiences into one user experience um where agents are doing all the work for you on the backend. And that basically requires a agentic backend or an automation platform for configuring this sort of foundation layer. And the existing competitors in the space mostly exist in the Web2 side. um there as well, they tend to be a lot way too heavy on the sort of complexity and on the complexity level. ah And we took a sort of different approach to incorporate more AI and create a more abstracted experience that didn't require uh a massive learning curve and made it more accessible to more developers. And over time, our goal is to make it accessible to sort of reduce those barriers lower and lower and lower. So instead of being a highly technical developer with us, you can be a sort of medium, medium skilled developer. And over time, we decrease those barriers all the way down to vibe coders. and eventually it's just going to be natural language prompting for a full stack agent from agent development to app deployment. That sounds amazing. I've been learning a lot about vibe coding recently, and it's actually still not easy. It's not to the point where I think most people could kind of build something. Yeah. part is the issue. Like, yeah, you can write something where like, how, do you know what you see on the screen is relevant, right? Like if you talk to a random person on the streets and you ask them some question and they proclaim to be whatever, whatever, and they tell you something back, if you don't know what you're at, what, if you don't know the answer you're looking for, you have no way to judge that. And I think similarly, if an AI is telling you something and you don't know how to put that in context, um, the value of that is kind of rendered useless for you. Yeah, I think for me, as I've learned, I'm learning about vibe coding, the Lego block approach is really great. um And in the Web2 space, I've built lots of little modules that grab a piece of data here, puts it in Google Sheets, and then uh opens up Tableau, and then grabs the Google Sheet data, and then does something with it, and then presents it somewhere, and then publishes it there, and then emails it here. all of these like little tools and then creating kind of bridges between them is really fun. But in the Web2 space, was like tools I was familiar with already, you know, because that I use every day for work or for fun or whatever. In the Vibe Coding space, it's like a lot of these tools are new for me. And so I'm having to kind of learn it first, kind of get the API key and then... and then create these bridges in between. And so that's kind of a pretty steep learning curve, I think. um I'm pleased to hear that uh OpenServe, you guys are aiming for a really kind of user-friendly, dev-friendly approach where you don't have to be an expert developer or medium-level expert developer, but new developers could start building something. And then eventually, natural language building would be great. um Tell us your decision around the building within the telegram or uh ton ecosystem. You mentioned already that the audience is massive and it absolutely is. um What's your experience been doing that in the ton ecosystem? So we're not specifically like in any partnership with TAN at the moment, uh just leveraging Telegram as a distribution channel more generally. And we're really at the start of that journey. mean, yesterday we've had a couple sort of previous developers building some little use cases and apps here or there. But yesterday the hackathon was the sort of inaugural start of this go-to-market strategy for getting people to build apps for the Telegram. for Telegram users. So it's an early beginning, but an exciting beginning. ah I mean, just really like the speed. Going into yesterday, I wasn't sure how much people were going to be able to accomplish within that short amount of time, like nine to 10 hours is not that long to build a full app, particularly using an AI platform that they've never seen or used before for the most part. yeah, seeing... Seeing what folks were able to build um and get working within that timeframe was really eye-opening and uh obviously got us all super excited because imagine with some proper support and time what else could be built um and I think providing that foundation is really just going to open the doors for uh people to sort of apply their creativity. And that's really kind of the essence of what we're trying to do is AI and agents specifically as sort of like a vessel for AI em sort of is a. trend to democratize skills. the requirement to possess a certain skill, particularly around like development gets lower and lower, and the value of creativity and ideas goes up and up. And we are sort of see ourselves as a sort of a front runner and sort of spearheading that sort of balance shift. increasing the uh value and sort of frequency of or sort of use of ideas and creativity as the primary driver in technological innovation. Seeing so much progress in a 10 hour day, these applications that are built using OpenServe, uh how does that change your approach going forward? You mentioned how inspired you were and also surprised at ah the apps that these teams at the hackathon built, um does that change your approach beginning today on maybe what you and the team are going to be focused on going forward? Yeah, so I mean, the kicker here is that this was actually made with the original UX we built last year. um So a big part of our product development has been UX exploration. What is the ideal user experience for a developer using this? Is it a chat-based experience? Is it what we currently have is a sort Kanban board where agents move along task lines from to do to in progress to done. uh how will a developer uh interface with these agents within a sort of product? And we have sort of the one we landed on for the past year was this kind of Kanban setup, sort of linear progression. And we have a brand new uh UX coming out in the next month or so, which is a... totally sort of visual drag and drop uh canvas. So sort of infinitely scalable canvas where you can drag and drop and customize certain parameters, action triggers, integrations. And I think that really seeing what was being like seeing the sort of rate of development and progress yesterday. got us even more excited and made us even more confident in what we were building at the moment and what we're soon going to be releasing because it's solved a lot of the little pain points that we were hearing yesterday where people were stuck for a moment and then found a way around or found a way through. But this new UX is going to be a real game changer for us and the space, I think. ah If I could just continue on the sort of, yeah, what's next for us is that, and sort of making the experience a sort of more comprehensive package. Like some of the feedback we got yesterday was, um, you know, people were looking for native, uh, agent hosting. So instead of having to host it on, uh, on their own local device or some other server, you know, one click agent hosting, uh, native database, deployment, things like that are going to be things we're focusing on, on the tech. side and then you know more related to your specific words where you know people were able to build these apps within a day where does the focus now lie I think this gives us a window to support as a team support uh other teams building on us support on these uh sort of other measures that turn an idea into a revenue generating application and business, right? It's one thing to create an app. It's another thing to monetize it, market it, get users, right? ah And I think that is where we want to also grow our competency and offering in is providing these sort of auxiliary support services to teams building on us because someone can have an idea and create something in a couple of days now. um That then gives us the space. We don't have to hold someone's hand for a month while they're building something out. We now then have some space to support them to bring their product to market and make it successful. I am pleased to hear that you're viewing these builders not just as developers, but actually kind of business owners. And I've had the same conversation with others. I think in the crypto space, we make a mistake in not thinking of developers and builders as business owners. ah Because oftentimes when we think of people just as like, they're just devs and they're building a product, it's... um it really cuts short kind of like the potential, but it's like, they're actually trying to build a business with users that actually pay for a service or a product. And that's really cool that you and the team are thinking along those lines and how you can best support the builders and the business owners building with the open-serve platform. Yeah, I we were planning to roll out kind of a Y Combinator for crypto AI campaign. um providing technical marketing funding support to these teams and builders ah to really make that next step. Like, like, I think you accurately. said it as it is in crypto. Devs are often viewed as just devs who, you know, participate in a hackathon here or there, like spin some little project up here or there, launch a coin and then nothing ever happens of it. And I think that's something we'd like to change and give people the empower people to. actually take that leap from not just being someone who can spin up some app in a few days, but taking the entrepreneurial leap and journey and creating real value for themselves and for the space. No, that's super exciting. You mentioned this white combinator type of uh kind of uh effort that you guys are doing. Are you prepared to tell us more about that or not yet? Yeah, we're going to be releasing more details soon and building a full campaign around it. But at a high level, it's yeah, kind of as I said, it is so providing technical support to teams for whatever it is they're trying to build. Creative support, marketing support is a huge one, of course, uh and funding support. That's excellent. I have not used the open-serve platform, but I'm curious about the, from a product perspective, um you know, the Kanban linear approach. um You know, I think that model makes a lot of sense, but as I've kind of played with agents, I've learned that it kind of requires more of a fluid approach because you try something and you don't quite know if it's going to kind of meet your objectives. so there needs to be kind of almost decision trees all throughout. um And in some cases, and maybe you can share more of your thoughts on this, in some cases at these decision points, um either yes, no, or does it get me kind of, it meets some kind of threshold at a decision point, I'm finding that one challenge in the agentic space is you try something and you want a result, but you want that result kind of stored in some kind of memory. And especially if it's like a learning, like you try something and then you learn. I'm speaking in abstract terms because I can't think of uh a more concrete kind of example, but what I'm learning is If I try something like a query from ChatGPT, let's say, and then I need that answer stored somewhere, I'm actually having trouble kind of figuring that part out. um Because that's like a learning, a current learning that needs to be stored somewhere in memory. But then that memory needs to be accessed by other agents to do something with. anyway, what are your thoughts on that? I'm hearing from others too that that's kind of a challenge. Yeah, so that's something we're actively working on as well. uh I mean, a lot of what makes our tech, uh our tech is we've built a series of custom frameworks and protocols that manage uh a lot of these complex uh elements, particularly. um So things like uh file creation, file sharing. um creating metadata, reading metadata, indexing files, retrieving information from these files, uh memory and storage. um And so we have a memory function that, sorry, and then if I could just quickly add output evaluation and validation. um And so for the memory one, in particular, we have a module that... Yeah, so in practice, how it works is like, let's say an agent, like yesterday, think someone ran into it where an agent took, uh called the wrong tool for something. uh And it then gave an error and the user was requested to give some information for what the right tool was. And it then stores that answer within the agent's uh memory module. And for the next time it needs to take that sort of decision route, it'll use the user's, the developer's input ah and won't make that mistake again. So that's kind of the way we've tackled that specific problem. it. I've talked with other projects about this and it's a really interesting problem. It follows along the lines of the chain of thought reasoning. AI is super, super smart and of course has vast amounts of knowledge, but what it lacks is human gut or human intuition. So teaching an AI human intuition in specific domains is kind of a big area of focus and something I've been really looking at and spending some time in. And so one example of this is if you're a nurse and someone approaches you and tells you their stomach hurts, you immediately go through kind of the triage process and you ask a series of questions. And I think AI is getting better at doing that, being kind of the interlocutor. of if your stomach hurts, then AI kind of goes through these prompts and helps kind of triage what the problem is. And I think it's getting better. But there's some domains in which it just has no idea. And so I'm curious about kind of your thought around chain of thought reasoning. Like if I have an objective, from a business perspective, I have an objective. But I don't quite know all the steps that need to get there or the tools that need that I have access to or need to be used to get to my objective. um How do you help developers in that situation? you know, put yourself in the hackathon yesterday, for example, and uh maybe the Kaido example that this team built, they probably had a decent idea of what they wanted to do, but maybe it was still kind of vague. How do you help teams kind of figure that out? Well, I think it depends on what level, Like sort of on the business conceptual level, like there's probably not, I mean, obviously we can provide some advice on a sort of more anecdotal conversational level, but sort of in practice. Yeah, it's a, it's a, it's an interesting sort of dilemma here. Um, and Yes, mean, sort of our approach more generally speaking, uh as regards to sort of reasoning and decision making is to like, we've worked hard to actually try and mimic ah or codify how humans operate and work in the real world. ah So instead of trying to, I think a lot of other teams uh have tried to reinvent the wheel. And I think There's scope for reinventing the wheel here, ah as we're obviously doing something new, but you don't need to reinvent the tire, the wheel and the spokes, right? You can reinvent parts of it and reuse things that are tested and proven. So for us, we put a lot of effort into studying and doing research on how do humans work and collaborate in the real world? you... are operating in a company at a certain department and you get a project sort of brief. How do you distribute that amongst your team? How do you communicate? How do you share that? How do you check in on everyone in your team on their progress? ah How does the brief that you've been given, how do those tasks progress between different folks in your team? And how do you create those decisions? we've tried to... basically implement that sort of human workflow or sort of natural human workflow that we tend to take and a process that's been sort of collaboration process that's been honed over. I mean, since over millennia and implement and codify that into our system. So as you say, like sort of the nursing example, when you give a project goal in our, in our platform, the project, have a project manager that immediately sort of initially creates a project plan and also asks you questions about your plan, certain input uh elements that are pertinent to the goal you're trying to achieve and lays out a the task descriptions, what agents those tasks are going to be assigned to, and sort of holds your hand as you make that decision and tries to guide you in a way that sort of fits the um framing that we've set out for how the collaboration should take place. And that type of guidance for teams is super helpful, I'm sure. Yeah, yeah, I believe so. And then the cool thing is, particularly with the new update we're releasing is we offer like the full spectrum of uh sort of flexibility there. Like you can choose to make use of these, I don't know if you want to call them tools or sort of backend decision frameworks, or you can go fully customizable as you like as well. And really give that range of eh of breadth to enable really skilled custom developers down to someone who's just more idea-oriented and has a bit of experience on execution. Yeah. I'd love your take on this. I spoke with another project in the same space and I asked them about the types of agents that are being built on their platform. And they said that they've been very surprised because the majority of the agents that are being built are actually in the ah kind of the creative space. And so copywriting. ah and within kind of like the creative workflow of graphic design, et cetera. um And he was surprised by that because he thought he would see more um of dev specific type tasks, but instead he's seeing a lot of kind of like white collar creative tasks that are being kind of, you know, agentified. um Are you seeing the same thing on your side? Hmm, not as much. I'd say that definitely does feature. uh We do have some sort of uh marketplace, um and as admins, we can see all the agents people have built as well. ah So sometimes I just take a look at what people have been building. And there's a shit ton of image generators, video generators, um those sort of... you know, an agent that imposes captions on a video or like switches up a background. ah I think there's quite a few of those, but there's a huge breadth of other ones as well. Like, I mean, in crypto, there's tons that we've seen quite a few or sort of on chain transaction validating, ah know, grabbing data, extracting data, doing data analysis on transactions. ah And yeah, then just trying to think then a lot of sort of like two more like tooling focused agents, ah which I would say like aren't like agents is a loose term. Like, where do you draw the line between an agent and just a ball that calls certain tools? But then quite a number of them that, yeah, basically take specific actions on certain platform or other applications. So whether it's sending out Google, sending out mails on Gmail or reading your emails, um scraping tweets, reading sort of YouTube descriptions, browsing the web. There's quite a lot of those as well, which are then sort of outputs or whatever those agents create are then passed on to the creative agents as well. ah But they do feature quite a lot in a lot of people's workflows. That is correct. Yeah. Yeah, let's talk about your definition of agent. um It's a gray area, right? It's difficult to kind of pin it down. I think the really interesting piece is the autonomous part. um Because some agents are very supervised, right? And also very structured. But the autonomous part is these agents can kind of think on their own within a specific framework. uh And you mentioned this marketplace on OpenServe. Can you share with us maybe like one or two really interesting agents that you've seen built? um know, of course, if it's not, you know, if you're not violating any kind of proprietary type things. But I think it might be interesting for the audience to get a sense of like the breadth and the opportunity of like, you know, what is possible in the agentic space. Yeah, mean, we're super early, so I take what I say with a grain of salt for sure. Yeah, for example, mean, sort of on the lines of what we just spoke about, like content creation, uh someone built an agent that basically takes a prompt and sources images on the internet. turns them into video, ah into sort of motion video ah and creates a full-fledged video with captions uh based on some story. So it does the copywriting, does the narration, does the voiceover, does the image, sort of image fetching, video motion, motion video creation and like synthesizes it all in one and creates a one sort of MP4 video output, which is pretty damn awesome I'd say. ah And interesting, like I think something maybe to go on a little bit of a tangent is when you think about agents and any specific goal or task you have in mind, can break, you can almost break that up into an infinite number of tasks and something that an agent developer has to navigate is how many tasks do you want to fit within a given agents like remit? Do you shove like 20 different micro tasks inside one agent's of meta task. ah And what are the implications of that? What is, does the error rates go up significantly when you do that? Or is it decreased when you split it amongst, ah you know, you could keep the same agent, but keep the tasks sort of more separated and contained. And then you can sort of debug and fix errors more ah with more precision as opposed to having everything done in So actually seeing that agent in particular where you had all of these different modules and uh sort of tasks being aggregated in one single agent and it being to being able to reliably produce uh an MP4 uh output that combined all of these different inputs was pretty impressive. That is cool. I mean, it reminds me when I first learned how to program, right? It's like functional program. It's like one big monolithic program with tons of functions inside of it. And then it became object oriented where each of these functions became their own module. And then you called functions from each other. um But that took quite a bit of learning curve and it's feeling like the agentic space is kind of going through the same kind of maturity. um Yeah, absolutely. um We kind of jumped into OpenServe and what you're doing and kind of the focus uh right now. um Would love to hear about how you got into this space. Had you always been in crypto for a while and if not, what did you do before? Yeah, I I got into crypto in 2017. Total noob watching YouTube videos and put my money in XRP thinking it was going to change banking forever. That was so was thrown into the deep end there. uh P &L went up a lot and then lost it all. Classic story right there. ah But you know, that piqued my interest enough to stay in it, to stay relevant and I spent my college years trying to learn as much as I could and get to grips with the market dynamics and opportunities that the space presented. So just keeping up to date all the time, reading, obviously trying to learn about new projects and new technology. And eventually that led me in early 2022 to BitTensor. This was, I'd say pre-ChatGBT, pre-AI boom, but came across BitTensor. and just found it absolutely like fascinating. Like to me, it was kind of the culmination of my entire crypto journey ah in terms of understanding sort of I have, I don't have a technical background. So understanding complex concepts and technology as well as being able to find things early. So I stumbled across it pretty early on and spent weeks studying it, trying to understand, make sense of this. uh incredibly verbose and mathematical white paper. And that really got me excited about AI. early uh 2022. So you were really early with bit tensor. I was indeed, yeah, somehow, someway I was, yeah. awesome. That is really, really great. And then, I started studying it and I wanted to get involved. Like I saw people at the time, uh anyone could be a miner and a validator on the network. Now this sort of architecture has changed a bit. um So I threw myself into the deep end and taught myself how to uh set up, like create a mining setup and validators. uh and kept those running and was earning emissions. I've all found, so basically running GPT models that I was finding on Hugging Face, uh swapping them out each time the rewards got less and less. And yeah, I just found it super exciting and interesting. And you could literally see in the terminal in real time, see the network prompts hitting the miners and the models and then seeing the outputs go back. ah And that was my first foray into AI. was running GPT-2 at the time. then chat GPT came out in November. And I remember seeing it like the first week it came out, like seeing it on Twitter, I'm like, oh, wait, this is a product built around that LLM model thing I've been playing around with. yeah, that was sort of like another sort of... sort of a explosion in the head, like, okay, this is like really something to take seriously. um And it's probably going to change. I didn't know yet for sure, but it's probably going to change the course of the world m in terms of how we execute tasks, how we think, we depend or how we access information, how we make use of information, how we learn. And so immediately, like I remember my first week using it with my roommates in London, we spent like the whole week, we bought a whiteboard and we were just like throwing down business ideas that used LLMs like on the whiteboard the whole week. And Yeah, I basically actually spent the next, yeah, sort of three quarters of a year um writing down ideas for AI businesses in my notes. And a few months later came across agents, started looking into different agent frameworks, baby, HGI at the time, MetaGBT, AutoGBT, these sort of early day pioneers of agent frameworks and agents. And yeah, spotted an opportunity that... somehow has still stayed relevant today, um where agents themselves are incredibly powerful, but remain rather inaccessible and unusable for large scale applications. um And yeah, here we are. What a great story. ah Does OpenServe do anything in the bit tensor space? No, we don't at the moment. No. Is that uh maybe in the road maps? Where? Yeah, potentially. I just mildly keep up with it these days. Not hugely. I'm obviously slammed working 16 hours a day. yeah, think potentially, I would like to see the ecosystem grow a bit more than where it currently is. I think there are some exciting use cases, but I think it still has yet to break out really and hit that inflection point. I think there's still a lot of... sort of a lot of actors that game the system that kind of curbs its growth. ah And I'm sure that'll be overcome at some point and I'm really looking forward to that. And then potentially seeing how it could fit in with us or we could fit in there somehow. That's awesome. Well, Tim, let's maybe let's wrap up. Tell us about, uh you know, what's what's in the next three to six to nine to 12 months for for OpenServe and how the community can can learn more and get involved. Yeah, absolutely. over the next three to six to nine months, uh I'd say the first thing is kind of kicking off our Y Combinator for crypto AI or AI agents. um So incubating teams, building on our tech and pumping out or not pumping out. a rather, that term has four connotations, but churning out, ah you know, real applications that solve real pain points and have the potential to onboard real users and generates real revenue. I think that's something that the crypto space somehow still doesn't have enough of. ah And we want to be a driving force in that and basically become ah an app, a app factory for, for the space and obviously eventually moving beyond web three and into web two as well. So I think that all starts with incubating and providing support to the early. builders on our platform, uh not just building some agnostic tech and being like, waiting for them to come, but actually actively supporting. ah And obviously that comes with a whole lineup uh of product updates and features. So like I said, sort of one click database deployment, one click agent hosting. ah The shared agent marketplace is something you want to build on more, uh create sort of gamified experience. think that's something that crypto has done super well ah is being able to gamify user experiences and create uh attention, awareness, and growth through gamification. um So that's something we want to sort of lean into more and see how we can find unique ways to bootstrap the growth of our tech. And something sort of within that... sort of supporting, uh you know, I say technical funding, marketing support is the yeah, particularly on the funding front um and the discussion we're having with internally and that we're going to be working on is our own version of a launch pad uh launch pads in their current form. Some couple, I'd say a couple have been successful and do actually provide value. Most of them end up being a wasteland of rugs. um And we'd like to change that. I think the issue with most current launch pads are the incentives are all wrong. There aren't enough. The system isn't set up for long term growth. It's all eh the lack of guardrails and lack of structure. uh invites people who just come for it to make a quick buck and the lack of focus also invites that as well. And with us, I think we have a super unique proposition where we have verifiable tech that acts as kind of a, yeah, like a factory or a in itself a launch pad for real world applications that generate real revenue, have real users. And if you can add the launch, the token launch pad aspect onto that, it, it totally changes the proposition like for a founder and for the investor. So investors get to be early for once like And we walk that walk ourselves. We launched at a million dollar valuation. Anyone in the world could participate in that and buy that by our token, which is essentially a proxy for our equity or that's where our revenue is being directed. ah Could participate in that from a million dollar valuation. ah That just doesn't exist for a company like us. They're almost entirely VC funded. And by the time your average Joe gets to participate, the VCs are dumping their allocations on them. um So that VC game is no different, is no less of a pump and dump than the crypto pump and dumps are. And we want to change that. We want to give people a way to invest in uh real technology that's verifiable. And we have the, we have the foundation now to do that. You can touch and use and see the apps being built in our ecosystem. And you can use that to decide whether you want to invest in something or not. It's not going to be some abstract idea and a white paper that someone's launching a token for. It's going to be a real app. And for the founders, you know, similarly playing that VC game is incredibly difficult. Like I'll tell our own story. I mean I was a year out of university I studied finance and entrepreneurship. My brother was still in university studying neuroscience uh it was unlikely we were gonna get funded by the VCs like to be frank uh and crypto gave us a way to fund our dream and look where we are now like creating real value for the world with real technology that is differentiated and We want to offer that to the other founders who don't necessarily have the networks or connections to get warm intros to VCs and give them the opportunity to fundraise for a dream they have uh and some value they want to create. And I think so far the launch pads that exist in this space are lacking that key element that we now have. uh And I think that really creates a full package um that should really uh Yeah, I think changed the game. So yeah, that's a big thing we're looking forward to. Well, that's exciting, Tim. And thank you for taking the time to speak with us and for sharing with the audience about OpenServe. And best of luck uh in the next little while in achieving the items on your roadmap. Great. Thank you so much, Peter. It was great to chat. All right. Bye bye.