Hi, this is Charles Hoskinson broadcasting live from warm, sunny Colorado. Always warm, always sunny, sometimes Colorado. Today's Saturday, and the work week never stops. The work never stops. The scandals and outrage never stop. So, I figured I'd make a quick video to talk a little bit about the Input Output X account and how we do moderation and social content these days.
For the past few years, Input Output has been a large company. As a large company, every time our official Twitter would release something, it would go through a long pipeline of curation. There were checks and balances, and the legal department would review each and every tweet, which is why it was a sterile, bland, and very boring account that no one really cared about. In fact, it had significantly fewer followers than me, where there is no legal review and pipeline.
As we've been optimizing the company, we've gotten rid of those checks and downsized the team to about two. That account shares all PR media that we do for all the ventures we're working on, including RealFi, Pogen, Midnight, Midnight City, Cardano, and other unannounced ventures alongside Blockfrost and things relevant to the Input Output group as a whole. With the new team, whom we hired from the Cardano ecosystem, they have moderated and worked with other ecosystems. I said, "Feel free to experiment, try new things, and let's see how it works and what people like."
Now, there's some risk in this because when people post things, they can sometimes be interpreted the wrong way and create issues. Midnight City, in particular, has had a hell of a time on X. Our account has been taken down without explanation multiple times for "impersonation," and we're not entirely sure why or how. X will not give us an answer for that. So, I suggested to the Midnight City media PR team, "Why don't you just use the IO account and play around with that account?" They agreed.
A big part of Midnight City is the exploration of the art of the possible with digital twins—owning your own agent, talking to agents, interacting with agents. We have multiple programs; Nix is one program that we have. We've been exploring with partners things like Omega Claw, Open Claw, and Hermes. I'm personally working on some voice cloning technology using Diffusion Gemma and Quan 3 TTS. There's a vibrant and lively area of exploration.
As a city with many events going on, the way information about it aggregates will be agentically led. Some of the people working on that got excited and enthusiastic and have been playing around with AI-generated influencers. Yesterday, I believe, one of them was posted on the IO account. This was done in good faith just to showcase some cool new things and the possibilities of these innovations. Unfortunately, it was received very negatively. A lot of people seemed to think this was an attack on the KOLs of Cardano or that we don't value them and just want to replace them all with AI.
This is just where social media is at. Every person takes the worst possible interpretation, and every person becomes an instant critic the minute they see something they don't like. It's impossible for a startup to move fast and try new things if the standard is that every single time you make a mistake, it's seen as bad faith, and the whole thing should be torn down. The reality is, if Midnight City is successful, it will have hundreds of thousands, if not millions, of users within the next 12 months. Open Claw and Hermes are some of the fastest-growing open-source projects in history, with millions of users. It's a new, high-growth category.
We can't have a community management team or media comms team that scales linearly with the population. We're going to need agents and AI to organize and sort all that out and broadcast regularly what's going on in Midnight City. It's important that we invest in and understand where the state of the art is at.
Patrick Tobler, for example, commented on the video saying we already have this technology. It's surprising to me that he would comment because we've been in contact with him regularly alongside other people in the community about what we're doing in Midnight City. This is not an attempt to do something different from his tech stack, nor is it an attempt to say that what he's been working on is not important. It's a small team thinking like a startup, trying new things. Some things people like, some things people don't like, and every two weeks we release a sprint. We're moving rapidly and have hit all the KPIs.
That is what the new Input Output has to be if we're doing commercialization and building ventures. We're not a government agency. We don't think in terms of years and decades, where every single communication and action goes through many layers of bureaucracy. I personally review all of it and sign off on it. There are a lot of people here thinking like startup founders, moving fast and trying new things. Some of those things work well, and some don't. The things that don't work well, you just stop doing and pivot. The things that are working well, you continue.
We live in a world where a company called MedV, with two people who started it, was able to sell $1.8 billion worth of product in a single year as a direct consequence of AI-enhanced marketing. Regardless of the ethics of that company, that's a shot heard round the world about how distribution and influence are going to work. Influencers, especially in crypto—not just in the Cardano ecosystem but in other ecosystems—have taken advantage of their position of influence and power to push pump-and-dump schemes and mislead people into buying things where they had secret economic relationships. You've seen the criminal allegations, sometimes convictions, lawsuits, and other things. This is not a pure or clean industry by any means. It's very expensive to utilize these networks, and they should be used sparingly.
We're always looking at where the future is going with AI CMOs, broadcasting, and how to make these things lifelike. Recently, I took all the voice files of Baldur's Gate 3 and parsed them using Coqui 3 TTS alongside some other technology. I was able to perfectly clone the voice of the voice actress of the character Shadowheart and have real-time conversations using Diffusion Gemma with it. This was a test to see the latency, intonations, and how lifelike the conversations are. In my estimation, with the evolving category of diffusion text models and where the voice cloning models are at, I believe that in about 12 to 24 months, this technology will be able to create a lifelike, near-perfect copy of a person's voice. On the thinking side, while combining Diffusion Gemma with Hermes, their self-learning agent, if you have a sufficiently advanced personality corpus, you will feel like you're talking to a real person within 12 to 24 months.
On the video generation side, there's probably another 24 to 36 months before there's perfect lip syncing, but the brain and personality are almost there. It's impossible to have these capabilities exist and not imagine that they'll have a tremendous impact on how we broadcast, who we listen to, and what we do. This is one of the reasons we created Midnight and Midnight Passport; we wanted to have an unambiguous proof of human and proof of human-generated content for these systems.
Everything is moving in new directions. I really need the Cardano community to calm down. We can't take offense to every single action and live in a state of perpetual outrage and bad faith. This is an example of an empowered social media team taking the time to try new things for various products in our portfolio, in this case, Midnight City. If you didn't like it, great, we move on. Reading into it that this means we have a position that every influencer in Cardano is worthless and needs to retire is absurd.
Second, not everything that's tweeted these days comes to my desk. It seldom does. I'm not even aware of most of these things. I've empowered the people beneath me to move fast and do things, and I told them I have your back. If bad things happen, I'll take the hits; that's my job as the boss. But they have to be allowed to try new things, and some of the things they do aren't good. They need to pivot and change. That's how we grow. That's how we think like a startup. That's how we get where we need to go.
We no longer have densely curated external communications where everything is pathologically examined 14 times and takes two weeks to send a tweet. You can't do that anymore—not in this environment, not these days, not with these resources, and not when you want to grow. You have to be willing to take chances. This ecosystem as a whole has to start thinking that way as well. Input Output is not the enemy. We're not trying to harm anybody or put anyone out of business; we want to work with everybody where and when we can. We're going to try lots of new things and explore where this technology is going.
I'm incredibly interested in voice cloning and video cloning because impersonation is going to exacerbate identity theft at scales the human race has never seen and creates a massive infosec hazard. This is one of the core USPs of having a selective disclosure system that can prove you are who you say you are and prove you're human, which is a core USP of Midnight Passport.
From the Midnight City side, we have two major KPIs that we chase: watchability and empathy. Watchability is how novel and interesting the game world is. The more novel and interesting it is, the more people passively and actively engage with it, which lowers customer acquisition costs. On the empathy side, it's how lifelike, understandable, and relatable your agents are to you—the Tamagotchi effect. The higher the empathy, the lower the churn rate, meaning less turnover and more engagement on the individual side. If these two factors work appropriately, they can grow the network to millions, even tens of millions of people. Adding in capabilities like agentic trading and affiliate and partnership relationships can bring millions of users into Midnight. That's why it's one of our most important projects, and we're leaning into it and integrating every single AI standard, whether it be X 402 or OWS, from a partner-driven lens.
If there's someone who's already built something, whether it be Ben Goertzel and Omega Claw, or what Patrick has done, or what Sunday is working on with trading strategies, we try very hard to bring these into the game world so we can share in mutual success with our partners. The goal here is growth. You tell me you want more transactions, more TVL, more users. So, we're experimenting with new things to facilitate that. We're not just an infrastructure company; we are a consumer product company as well. We have to enable and empower them to think like startups and work like startups.
Input Output Group has many experiments. This relationship doesn't work very well if every single time one of our startups, one of our teams, or one of our people does something you don't like, you infer that we're now your enemy. We hate you, and we want to put you out of business or harm you in some way. We are in a broader macro environment. AI is growing exponentially. Crypto is in the toilet. It's uncertain if we're ever going to get out of this war with Iran. The regulation side lives in a gray area, and the world moves on. We can complain about it, or we can take a step back and try to be as agile as possible and move forward.
The innovation doesn't stop, and the core of who we are—the scientific backbone—continues to push forward. Just the other day, we were looking at how Star Stream can enable a new class of persistent agents that can live on the blockchain, act as event listeners, and be decentralized, triggered by events on the chain to work with smart contracts instead of doing this in a centralized off-chain way. On the formal methods side, there are already conversations about how dependently typed languages can be used as a heavy specification language for vibe coding. We've made enormous progress with Leos, and in three days, the Leos testnet is launching. The lessons learned there will enable Cardano to improve in speed by 60x, and that's just the beginning. Adding Parison to the fray means we'll also have fast finality. With all the ZK enhancements, we're going to be able to have trustless bridges.
The innovation doesn't stop, but it has to be wrapped into products that can be adopted by the countless millions of people around so the ecosystem can grow. Through that growth, we have more funds available for everyone to grow their businesses. That's the lesson we've learned from the failure of the Cardano dApps and the fatigue of people in Cardano governance. The pie continues to shrink because we're stalled as an ecosystem, and thus there's not enough resources to sustain them. We have to grow the pie again and do that by being empathetic, welcoming, giving people the benefit of the doubt, working with people, partnering with people, sharing the wealth, and embracing the commercialization of the platform and innovative new products.
We don't do that by being in a state of perpetual rage, cynicism, and adversarial behavior. I'm sorry if anyone had any problems with the AI video we released, the AI influencer, on our Twitter account. At least you now understand where it came from and where our thinking is at, and more of these things will happen in the future. They are a direct consequence of empowering people and giving them the freedom to try new things without being punished. Bad companies look for scapegoats and punish people the minute someone makes a mistake, which kills the culture of innovation. If I'm going to be successful and all these things we do are going to be commercialized, I have to do the opposite and say, "Gosh, sorry that didn't work. Let's try something new and different."
If this ecosystem wants to be welcoming to people who want to innovate, you have to do the same as well. You can't make an exception for Input Output and say that we're somehow different. We're not different. The governance debates have proven beyond a reasonable doubt that we are truly just a community member and one of many building infrastructure and commercializing products on Cardano. We don't have any special roles, responsibilities, or rights, and my legacy as the founder no longer matters, as I've been reminded many times by many different actors in this ecosystem.
But you can't have it both ways. You can't say that and then treat us differently when we try things that don't work out. It's not fair, and it's deeply destructive to our ability to grow, innovate, and try new things. Thank you so much for listening, and I hope this clarifies and allows people to understand where we're coming from. We're not trying to replace the KOLs of Cardano. We value them and think they're important. We're going to continue with our Twitter and other accounts to try new things. Some you're going to like, some you're not going to like, and hopefully, these new things will allow us to manage the growth of Midnight City as it turns on and gains a huge population, which will be a net benefit to everyone in the Cardano ecosystem. Cheers.
30:22
8:56
8:30