AI is transforming gaming — not incrementally, but fundamentally. In this episode of The NFX Podcast, James Currier sits down with Nick Walton, CEO of Latitude (AI Dungeon, Voyage), to break down what it really means to build an AI-native game company — and why traditional AAA studios may be unprepared for what’s coming. Nick was experimenting with GPT-2 in 2019 before most people had heard of it. Since then, Latitude has become one of the earliest and most advanced players in AI-powered games, pioneering dynamic storytelling, infinite RPG worlds, and multiplayer AI experiences. This is a deep dive into the future of interactive entertainment. Topics We Cover: - What “AI-native” actually means in gaming - Why AI games improve overnight (and traditional games don’t) - The shift from scripted content to infinite, dynamic worlds - Building Voyage: a genre-agnostic RPG platform powered by AI - How world models (like Google Genie) fit into the future of games - Why mental models must change every 1–3 months - AI-native software development and ultra-high talent density teams - The resistance to AI inside gaming studios - UGC, long-tail IP, and unlocking infinite fandoms - The magic of multiplayer AI storytelling - Why AI games will feel more personal — and more emotional If you’re a founder, game developer, or AI builder, this episode breaks down how fast this space is moving — and what it takes to win.
Nick Walton (CEO, Latitude Games)
AI isn’t just improving games — it’s fundamentally changing how they’re built.
In this episode of The NFX Podcast, James Currier sits down with Nick Walton, CEO of Latitude (creators of AI Dungeon and Voyage), to explore the rise of infinite, AI-powered RPGs and the shift toward self-writing codebases.
Nick was experimenting with GPT-2 in 2019 before most people had heard of it. Since then, Latitude has operated at the frontier of AI gaming — building systems where worlds evolve dynamically, stories adapt in real time, and small, high-agency teams ship at unprecedented speed.
This conversation goes deep on both the player experience and the engineering revolution happening underneath it.
What “AI-native” really means in game development
Why AI games improve overnight (and traditional games don’t)
The shift from scripted content to dynamic, infinite worlds
The architecture behind persistent AI-powered game engines
Building multiplayer RPGs with AI at the core
The “world engine” layer vs. raw AI models
Why mental models in engineering must change every 1–3 months
AI-native coding workflows and self-writing codebases
How small, talent-dense teams move faster than AAA studios
Resistance to AI inside traditional gaming companies
Why players will embrace AI when it delivers real value
Nick Walton is the CEO and co-founder of Latitude, the company behind AI Dungeon and Voyage. Latitude has been building AI-driven interactive experiences since the GPT-2 era and is one of the earliest companies operating at scale in AI-powered gaming.
Subscribe for more conversations with founders building what’s next.
The NFX Podcast explores how technology shifts create new categories — and how founders can win in them.
[00:00:00] Nick Walton, CEO of latitude, makers of ai, dungeon and Voyage. Welcome to the NFX podcast. Thanks for having me, James. Excited to be here. It's very excited to have you here, buddy, because you and I have been working together now for over five years, and, uh, I think we invested in August of 2020, uh, and have been working alongside you since then.
And, uh, you know, latitude from what we can see is still the leader. And, uh, this, number one, the most advanced player in AI games. What. Talk to me about what is AI games to you? You know, you've been at this for five years, you've become the leader in this space. So tell me about AI games. Like what was the idea five, six years ago when you started this?
I was always excited about the, the opportunity that AI would bring for the world more generally. I originally was in self-driving cars, so sort of seeing the opportunity of AI to make a better world where, you know, [00:01:00] the, the roads are safer and, and there's more opportunities. Um. I was playing around at a hackathon and tested GT two for the first time.
Um, had just come out. It was not very well known. It was kind of a, a niche researcher thing. This is November, 2019. Yeah, I think it was March, but it was, they hadn't released the full version. So this was the a hundred million parameter version. Yeah. So like ridiculously small by today's, you know, era of models.
And it was completely nonsensical, right? It could barely write a coherent sentence. Um, but. Just that seed like planted in my brain. Yeah. And um, I built, I, I'd recently, I'd recently started playing d and d my family. And there was this magical experience in d and d that we've never replicated in a digital experience, which is the ability to have the freedom to create your own personalized story and have it go anywhere and just really exercise your, your creative freedom to have these amazing stories and experiences that have never possible before.
And. When I, when I played around the GT [00:02:00] two, that was when I first realized, wait, like this is possible. It's going to be possible to have dynamic stories and dynamic experiences that enable a entire revolution in how we create and play games that we can barely even imagine. Um, and I saw the, I saw the promise of that back then, and even still.
The, the vision I had was far below what, what it's turning out to be. Um, and I think it will still take some time for games to reach there because. Um, they are a more complex media type than, than text and images, but their effect is on the gaming space is gonna be just as pronounced as it's been on text and image and video.
Got it. And so you had on G PT two, you had downloaded it and were running it on your local server, right. Like in your basement or something? Yeah, so I was in a, in, in an AI lab at, at university and um, I downloaded it. I built a first [00:03:00] version. I started like figuring out how do I train this model so it can actually tell adventure stories.
Um, and then trained it. I remember the first time I trained GT two on the right data and it could kind of do this like interactive story experience. It was the most amazing wow experience for me. 'cause I felt like for the first time, it's kind of like the experience I think people see looking at world models like Genie, where you're like, wow, I could be in a world and do anything and move around.
And it's like, it, it, it feels real. That was my first experience. Um, and it really just. Gave me this taste that I've been like craving to build out the full version of for a long time. Got it. And so then you called it AI Dungeon. Yeah. And uh, you just put it up and then suddenly there was 400,000 people using it.
Yeah. So I, I. It, it was kind of funny. I was a student. I was like, I don't know how to host GPUs. I don't have the money to host gpu. So I launched it as a Google collab notebook. I was like, let Google Pay for the gpu. I'll just [00:04:00] make sort of a fun, fun, uh, uh, scripture how to do it. And I did not expect, like, how much it would just blow up like within a day.
There was, you know, uh, a hundred thousand people played it. And, uh, it was clear there was something there that, that something people were really excited about. Got it. And then a little bit later, uh, OpenAI launched, uh, GPT, uh, GP PT three and that that was hosted. Yeah. And that was different. Yeah. But it was a better model.
Yeah. The language was better. The, the language it created was more realistic and, and then again, there was an inflection of users after that as well. Yeah. And that was one of the things that is remarkable about the AI gaming space. It's so different from traditional games, so. Um, when we went from GT two to GT three, we silent, AB tested it and the users knew immediately they were like, something is going on because the experience is just so much better.
And all we did was change, essentially change a string, and suddenly our experience was drastically better than [00:05:00] before. You don't get that with normal games. You don't like take Skyrim or or another game and, you know, change a string and suddenly the game is. So much drastically better at a fundamental level.
And so that's one of the things that's really powerful about AI Native games that have AI deep in the runtime, is you can upgrade this AI engine and have the experience improve at a rate that is impossible with traditionally developed games. Got it. And I think I talked to the open AI guys and after they launched GPT-3, they thought everyone was gonna try it out because now it was just an API call, but for the next six, eight months.
AI Dja was still the majority of of inference, is that right? Yeah, for, for a good amount of time. We were the largest inference provider as far as I'm aware, on open AI user. Yeah. Inference user. Yeah. Because, um, and you were meeting with them two or three times a week sort of giving them feedback on how they needed to improve?
Yeah, because, because Chat, CBT. You know, they, they hadn't nailed, there wasn't good accuracy, which is what you needed for chat. GPT Ja chat, GPT is like AI Google, right? Like it's answering questions [00:06:00] and if it's not accurate, it's not useful. And coding was nowhere near yet. But what it could do, its biggest problem was also sort of a strength for us is the hallucinations.
Its ability to like say crazy things. And that actually like provided some of the charm of early ai. Dungeon was like. Just like hilarious, like really fun experience that was like zany and wild, but really compelling. Um, and so yeah, we were, we were a lot of the early traffic, uh, of ai, of open AI before they figured out some of their other, um, got it.
Product avenues. And then your, your inference costs started going up as you had more and more users. So you raised some money from NFX, from album vc, um, from Griffin, and, uh, you kept going, you built the team up. Uh, and then, and then what happened? Yeah, so we, we kind of went on this route of, of growing AI dungeon and growing the team.
And I think we made, there were a couple mistakes that we did make. Um, one was a classic founder mistake of like expanding too fast. And, and this was something where, you know, I, I sort of like stumbled on this right out of [00:07:00] college and it was like. You know, there's a lot of challenges with learning how to run like a startup.
And we grew so fast that, you know, I went from like graduating to like managing people or managing people in like a year. And I was like, I don't know what the hell I'm doing. Like this is, you know, this is crazy. And so there were a lot of things, and at the same time, we were the cutting edge of like doing things no one in the world had done before.
And so here I am managing, you know, all these people trying to figure out how to deal, deal with questions like. AI safety and privacy and dealing with a technology that's still very new and unproven. So, um, and having to invent new technical things. Yeah. Having to invent new technical things, new sort of like paradigms.
Um, and so we hit, we hit some snags with like, how do you deal with like privacy and safety and, and you know, we, we were sort of like caught between open AI and the community trying to find out like the right approach to things. I think where we eventually landed is actually where the industry has moved in general.
Um, 'cause some of the early debates were, you know, how do you deal with like, private, uh, usage of [00:08:00] ai? Um, and I think some of the earlier approaches, you know, uh, people wanted to follow a social media model where you say. We're gonna like flag things you do. And if it's like, you know, against the policies, we're gonna ban you.
But, but what that turned out to be is like, uh, it felt like a huge invasion of people's privacy. Um, so eventually what we realized is like, yes, there are, it is good to have some guardrails around the ai. Like there are certain types of content you may not want to generate, but within that space, especially in a game environment, you need to give people the freedom to just like.
Express themselves and have fun in a carefree way where they can just play and test the boundaries and have fun, um, without worrying that someone's going to like read their story or invade their privacy. And so eventually we figured it out and, and figure it out the right way. And now that the community's very happy with our approach to how we balance saving sa eventually, now that.
Now the community is very happy with our [00:09:00] approach and how we balance safety and privacy and. And grow a healthy product. And I think we've done a really good job striking the right balance. Um, but it was, it was definitely like a hard kind of rock in the road. We figure out, took a couple years to figure out that's, that's a, yeah, that's a really, uh, sort of, uh, intellectual property.
You guys have learned over the hard knocks Yeah. To figure that out. And it's partly that UGC platforms are just hard, right? Like doing a good UGC platform is hard. Um, so we, we went, you know, it, it kind of what took, took a little while. Uh, since we figured it out in the last couple years, the community and the user base has grown a ton.
Yeah. As we've just accelerated. And you've got great communication with the community now, right? Yeah. We built up, you know, at one point it was a, like, communities can be hard to manage, especially in the gaming space. Um, but we realized that we just need to take out an approach of. Listen to them very deeply at their concerns, understand where they're at, and just have a humble, we're gonna, you know, work our butt off to solve all these issues and make this product great and we're gonna listen to you along the way.
And [00:10:00] over time, that built up a lot of trust because you have regular communication with Yeah. And the community knew that we cared about them. And, and our first value as a company is obsession over delivering value to users. Like we just obsess on how we can deliver the best experience to them and just trust that results come out of that.
And like paradoxically, our best growth as a company has come since we've learned that and just rabid focused on delivering the very best value, uh, in our products can. And the way the product works is, is people are paying a monthly subscription fee. Yeah, and so you're just trying to build more and more and more value without raising that subscription fee just so that they get more and more.
Exactly. Yeah. So talk to me about AI coming into the world of gaming. I mean, you're coming at it from an AI native, like the most AI native guy out there because you started with GPT two and have been with it for the last six years doing that. Then at the same time, the incumbent gaming providers are starting to learn how to incorporate ai.
Talk to me about the differences. I think one of the big [00:11:00] challenges is when you're building complex systems and games and organizations that make games are extremely complex systems. You build them and you start 'em from a small seed and you iterate and, and you make them more and more complex based on set set of constraints.
And so you build up this very large, complex set of abstractions. And organizational structure that is highly optimized to make a certain type of thing. But then when your constraint space changes, when suddenly, like things that were not possible are possible, it, it becomes very, very hard to shift your mental models.
How you work, what your conception of great products are, how your organization builds. So I think that's one of the challenges and one of the advantages of being AI native is we were able to come at it from a, an entirely fresh perspective. Say, let's restart, let's redesign how to fulfill fundamental human needs.
'cause I think there are first principles that you can, you can base on, right? Like humans. Are the same and have been for thousands and thousands of [00:12:00] years, right? Like we, we have a love of, of deep, enriching stories. You know, kids instinctually role play. Like, you know, kids naturally say, I'm gonna play house, or cops and robbers, or I'm a knight, or I'm a, you know, like they put themselves in the shoes of another character and make choices in a world and play.
With their friends in this sort of like immersive story together. And even as adults, we do that. And even as adults we do that, we just don't get enough chances. That's why we like games. Yeah. And so like there's these fundamental human desires and, and that's how we think is like our job is to fulfill those fundamental human desires in new, innovative ways.
And AI gives us an opportunity to deliver on player dreams and fulfill those values in a way that have never been possible before. And we can, we can. Build an AI native version of that that's not held back by the past, you know, ways of thinking and ways of doing. Got it. So you've, you've had this challenge of having to change your mental models and your organizational structure sort of every six to 12 months based on the new capabilities as you've rode this very early six years of ai.
[00:13:00] Yeah. And it's getting faster. Yeah. I would say it used to be we had to change our mental models every 12 months and then it went to six months, and now it's like every. One to three months, like even in the last month, you know what we, what you could do as an AI native accelerated development company is drastically different than what you could do in November of last year.
Yeah. And so what we're finding is, um, so there's a couple pieces of being AI native, but one of them is having an AI native. Like way of building software and this is one of the biggest things that's gonna happen. It's already started happening, but it's gonna happen a ton in 2026 and 2027 is how companies build Software is just going to change dramatically.
It's not, it's not gonna be a marginal productivity gain. Per engineer, it's going to fundamentally transform how we even think about making software. And I think most companies that are not AI native are completely unprepared for that. Um, their mental models haven't adjusted. Their mental models haven't adjusted.
And so one of the things you guys have learned over the last six years is how to change your mental models more [00:14:00] easily. Yeah. And you, you only have people on the team now whose mental models can adapt without a lot of emotional or psychological friction. Hard. Yeah. You have to be very, that's hard, adaptable, hard and rapid.
Yeah. And you have to be very first principles based because. When you're first principles based, you're able to say, okay, here's a new set of constraints. How do we first principle think about it? Whereas people who are not first principles tend to just like take existing patterns, is just assume that they are the patterns that should be followed without questioning them and whether they still apply.
Got it. And, and so, and so that's the AI native approach. What, what are we gonna see from the incumbents using ai? How is that gonna be different for them? Yeah, so I think there's, I think you'll see that in a couple realms. So I think number one is it's gonna be really hard to turn the ships if you're a big incumbent, if, if you've, you know, spent your whole organizational history building up these massive, complex, you know, pipelines of like, you know, waterfall design leads to art, leads to, you know, implementation.
And we have this like really complex flow. That's a big ship that's very hard to turn. So I think the big companies, it made them really [00:15:00] good at what they did. It did, it was the right thing to do for the set of constraints when they built them and the constraints changed. And so now it's, it's no longer gonna be the right thing to do and it's gonna be hard to shift that, that boat for them.
Um, are, is there anyone you think is doing a better job of it? Yeah, I think there are some companies that are leaning in, um, you know, I think one example in a bigger company, and I think they've always had a very scrappy startup approach, so that helps them more is like Supercell. So Supercell has a massive, um, AI games innovation lab where they're bringing in people constantly and they're giving them stipend and they're giving them a space to just like rapidly explore ideas of how to make AI native games.
So I think that's one example of, and they're bringing people to the Bay Area. To be around the ai. Yeah. And they're funding them here and they're mentoring them here. Yeah. And that's all going on. And you've moved here. Yeah. You're now in the Bay Area. Yeah. Yeah. You started in, in Salt Lake, but now you've moved here two, three years ago.
Yeah. Got it. Yeah. Yeah. Okay. So they, they're leaning in and they're, they're bringing in people who have the new mental models and then sort of [00:16:00] slathering themselves with that so that they as an organization can change. Yeah. Yeah. Yeah. So I think that's one, but I think. We've, we've talked about the AI native aspect of development, which is one piece.
Um, and I would say there's maybe like two other pieces of how, you know, AI native paradigms will transform games. The second piece I think, you know, we saw with Roblox the power of user creation, right? Um, and, and that was the, the, you know, the UGC platform, era of YouTube and Roblox and Instagram and, and all these platforms that said, Hey, what if.
You know, creation is way easier. How can we supercharge how people find, discover, and create content? Um, and Roblox drastically lowered the bar of creation. Um, AI is doing this in entirely new way with AI native companies. Um, so we're seeing this with, you know, images with Mid journey, um, you know, with text, with Chad g Bt, or Claude.
Um, with videos, with, you know, [00:17:00] so many new video tools. Um, and, and on the code level you've like lovable and cursor, and this feeling of creation with AI is kind of magical. Um, anyone who's, who's been at the forefront, like I I am working a lot more in the last couple months. Not even partially because there's so much compelling things to build, but also 'cause it's just way more fun, the feedback loops of.
Prompting AI to build something and seeing it manifest is, is just such a compelling experience. So I think company, AI game companies, so you can give that to your players. We can give that to our players, right? Yeah. You can give that player the experience of Create this world for me and then see it in instant shape in front of you.
Right? Like that's a magical experience that hasn't been possible with games before. And that's gonna be, I think, the best AI native games are gonna lean into, or AI native game. Companies are gonna lean into that approach and say, how do we deliver that magical feeling of creation to players? Right? So you can imagine that the, the mental models for the incumbent game developers is, we are gonna create an experience you will experience, and it will be so good.
You [00:18:00] will love us for it. Yeah. Whereas you're saying, we're gonna give you the AI powered tools for you to be the creator. Yeah. And, and, and relish that experience. As well as the game experience that comes on top of you creating Yeah, yeah, exactly. Yeah. And, and, and that game experience is the third pillar of, of what it means to do AI native game development.
So the third is like, how do you make actually new, innovative type of games that were never possible before? Right. Um, and we're still very early on that. And, and like I mentioned that the reason, uh, we're very early on that is that games are just a very complex media type. So, you know, humanity first started making.
You know, text and stories, right? And then, and then it's like they started getting better at painting and visuals and then later video came, right? So there's sort of this progression of complexity of media type. And games came long after vi digital video games came long after we were making videos. Um, so the same is gonna happen I think with AI native, like really good quality like AI created games [00:19:00] is, is, is going to be a bit later, but it's gonna be no lesson.
You know, it's gonna be AI native games are gonna come later, but they're gonna be just as impactful and compelling. So what does that look like? Um, I think we've already seen some signs of this with things like character ai, which, you know, had 20 million DAU showing that there's something compelling here.
Um, chatting with characters. In a way that feels like they're really, there is a very, can be a very compelling experience. And it's so different than like a scripted dialogue tree. A scripted dialogue tree is like, you know, it's not real. You can sort of fool yourself into like getting immersed, but, but you know, at the end of the day, it's like you're just, you know, reading a handcrafted thing, whereas having a real conversation, you feel like they're there.
You feel like you're, you're in that experience and that is gonna be way more compelling. Um. Executed well on. Um, the other thing is, is bringing that magic of d and d and tabletops, right? And, you know, the d and [00:20:00] d market is like, in some ways it's niche in other ways. It's crazy how big it is given how hard it is to get into, like, if you've ever tried to play d and d, it's like you've gotta get all your friends who rely on their schedules.
You've gotta have an entire four hour session just to get set up to play. Like it's this super complex thing and yet so many people do it 'cause there's this magic in that experience. And that magic comes from having the like freedom to do whatever you want and have the story or out of it. That magic comes from having the freedom to do whatever you want and having a unique, personalized story of you and your friends that comes out of it, right?
So if we can capture that magic in a digital experience, I think you can. You can build something that is compelling on a level that. We've never been able to do before with traditional games. And, and you, you guys have been profitable for like two and a half years, right? I mean, you've got a lot of users, got a big community Yeah.
And your main interface is text. Yeah. Yeah. Right. I mean, that in itself is sort of [00:21:00] a, a, it's going backwards in terms of the production value that we've come to expect from, you know, red Dead Redemption and, and you know, uh, GTR and all that GTA and, and, um. But, but it's still a beginning to build new experiences for people.
And that's sort of an example of what the interface is like and, and you're now making it a multiplayer experience with Voyage. Yeah. Can you talk to us about that? Yeah. So, and Rrp G 'cause you're not just going after the dungeon crowd. Yeah. I mean that's, that was called AI Dungeon and maybe people think, oh, that's just for Dungeons and Dragons players.
Yeah. And that experience of being together in a multiplayer environment can extend to all sorts of different. Environments or IP with something like Voyage. Yeah. And you know, RPGs, um, for those who don't know it's role-playing games, um, and at their fundamental, they're actually quite simple at, at a fundamental level, it's you get to step in the shoes of another character and make choices and interact with [00:22:00] other people who are in this shared world with you and have a story rise out of it.
And other, other big gaming companies that we all know of are. Like, you know, you've got Grand Theft Auto, one of the, you know, most expensive media productions ever made. You've got Cyberpunk, you've got Skyrim, you've got Star New Valley. You've got, you know, if you look at it like the vast majority, like the largest market in the games category is RPGs because it's a very compelling experience to get to be someone else in another world.
Right? That's, that's so fundamental. To like human desires of play and stories and creation is to imagine being in another world. So RPGs is just like one of the most fundamental play experiences we have. So it's extremely broad. So what we're building with Voyage is a genre agnostic RPG platform where you can come build any world, any experience you imagine on top of it, in.
Minutes to hours with AI and have it be an experience that is not scripted. It's not a [00:23:00] scripted experience where you have a few options, the developer rewrote, but it it's script list. It has infinite freedom to take the story in any direction. It's infinite content and infinite content goes on forever in any right now fandom or any interest that you love.
Right? And so I won't have to wait. Eight months for the next release. Yeah, exactly. And, and um, you know, it's very much a, a progression point from AI Dungeon where AI dungeon is like the very simple version of this, right? Where you've got like simple text, um, and, and, and the seed of these concepts. But this is going, you know, taking it much, much farther to what happens when you have, you know.
Complex multiplayer where you, where you have a rich full world that remembers you, right? Where you have an entire world engine underneath the ai, because that's something I think people don't understand. You know, people look at things like Genie, you know, these world models where it gives the same promise of like you can, you know, prompt any world.
And I think that's a really important layer, Google's genie that they just released. Yeah. Google's genie word model. Yeah, I think it's a really compelling layer. [00:24:00] There's, there's a second layer you need that's missing from that because that by itself, you've got 60 seconds, and then the, the memory sort of dies out.
There's no stories that you know, that, that really grip you. There's no like, rules and, and way to progress and get mastery and get stronger, you know, and, and learn a, a thing. There's no multiplayer. So you need, you have an AI layer. Yeah. And then underneath that AI layer, you need, uh, uh, what we call world engine layer that grounds the ai.
That lets you have deterministic state connected to it that lets you have persistence in the world. So you can leave a location and come back a thousand turns later and it's still there. So what that does is that makes your choices matter. That makes that unlocks like creation and. Persistence and long-term play in a way that you can't with just the AI model layer.
And that unlocks multiplayer, which I think is something that we haven't really seen a super compelling AI consumer product on multiplayer yet. And I think Voyage really delivers a, uh, uh, insanely fun multiplayer experience. Like we, we play test all the time, and it's one of the, the funnest [00:25:00] things that we do is just.
It, it really captures that experience of being around a table with your friends and going on adventure and having an amazing story come out of it. And you just like finish and you're laughing and you're like, oh, remember that thing that happened? And, and it's just a really magical experience. Um, but you need that world engine layer.
And so that's what we've been building in the last few years. So, so Jeanie is good. It's just a component. I mean, Google has done us a good thing by saying, here's a. Here's a technology, here's a fundamental Yeah. A foundational technology that is going to be a component Yeah. Of what we all wanna do.
Exactly. Just as an l, LM, as a component. But it's creating the more magical experience. Exactly. And as world models get better and cheaper, we'll be able to build that connected to the world engine and deliver the full cycle experience. Um, at least that's, that's kind of where we think the technology will go, um, to enable the end.
Vision and promise that I think people are excited about. Got it. And you, and you know, because it's AI native, instead of it being about one set of IP or one set of [00:26:00] characters, voyage could expand to all ips. What, how is that gonna work? Yeah, I mean, I think you see this with intellectual property, ip, yeah.
I think you see this with things like YouTube or fanfiction sites where when you have a UGC platform, you're not limited to. The few things that the, the developers can create and are interested in creating. So one of the things that's like really fascinating about UGC platforms is you see the long tail of, of things people are really excited about, right?
Um, because especially in games, like we've never captured the long tail, not really, we've captured the long tail in YouTube and, and all these other platforms where any interest, any fandom you're excited about, you can create content on it and, and share it with other people and really get connected to it.
But games. You know, except for the biggest ips, they never get, they never get games made out of them like Star Wars or, yeah. Star Wars gets games. Of course Harry Potter gets games, but, but there's, there's, you know, so many amazing worlds made by so many amazing creators that never get away for players to jump into them and feel like they're [00:27:00] part of them.
So I think that's one of the, the, the real value add this gives is it, lets. You jump into any world, right? Not just the one world that someone like had a big enough budget to be like, any world you love and imagine it could be on this platform or something you create, right? Like, I think one important thing is, you know, uh, we, we very much don't see ourselves as replacing human creativity or replacing, you know, human dungeon masters are, are replacing creators because I think human creativity is this beautiful thing that that is.
Is very necessary. And uh, what we see is these, these people are creating worlds. They're creating ips, they're creating stories, and our job is just to. Let those get amplified. Let those become a real world that anyone can jump into. Sure. So maybe the next great world will not come from an author, uh, writing a book, but it'll come from people playing on voyage and creating these worlds themselves.
Yeah. And, um, and so you, you can let them do fan fiction and let them evolve into worlds they [00:28:00] already know, but. But there are IP holders that could partner with Voyage to actually create new worlds. Yeah, definitely. Yeah. And, and that's one of the cool things is we can, we can go so rapid because the engine.
Does so much of the lifting, um, itself, you don't have to, you know, spend years creating this content. You can say, great. Like, let's instantiate this immediately and let's start iterating. And like we can have a, a, an experience that's super fun very quickly, and that gives a new way to experience a world in IP that that isn't possible with traditional.
So, so voyage can ingest all of the, the content that they already have. Yeah. Look at the videos, read the texts, all all of, you know, user comments. And from that, be able to bring out a world very quickly. Yeah. In a couple, in a couple days. Yeah. Yeah, exactly. Yeah. And then, and then the IP holders can look to see if it fits with their guidelines and Exactly.
And they, they can approve it and that kind of thing. Yeah. Got it. So you've mentioned that building, uh, a unique AI native product like voyage requires a certain mindset and a certain group of people talk to us [00:29:00] about the team that is necessary to make this happen. Yeah, so I think the traditional way is you often have these very large teams that have these very complicated, like, you know, content production pipelines, and I think those have struggled more and more even in recent years.
You know, AAA is, is struggling. Um, I think in the AI native world it shifts dramatically because. Number one, you're not creating content, right? Like the, the game engine creates the content on the fly based on what users express. And number two, and the users are creating the content. Yeah. New users are creating the content.
And number two, um, like even the software will become less and less written by humans and more and more written by ai. And, and we've been able to be at the forefront of that in, in a way that I think we see the future where, you know. It's kind of crazy how fast is accelerator. Like even in the last month, like I've been personally like rapidly rewiring all our systems to move way, way faster with ai.
Right? So it's not, I think a lot of people kind of see the [00:30:00] cursor level of like, okay, I chat with ai, it make some changes. I chat with ai, it make some changes. But there's several levels in this sort of like getting more and more agentic that you have to climb. And I think we've, we've climbed pretty far, um, as an organization.
As I, as an organization. Yeah. So, so what, let me paint you a picture of what that looks like. And then you understand like what the team's required to execute on that. Yeah, because, because the talent density required for this is much higher. Really high. Yeah. Because in the past you have people where you're like, here's this feature, let me hand you off this feature and implement it.
Right. And like, AI can do that. So, so where it becomes is actually like. Instead of just handing off features or development, what you do is, you know, each individual is actually like a PM engineering manager in one and they're writing specs, right. So like, they're writing, this is what I do. I I have like, the most commits and prs in our entire organization 'cause I'm just doing this so rapidly.
Um, so you make like a, a, an entire spec of like, what should this look like? What, what are the. Objectives and you have a conversation and [00:31:00] you sort of like co-design and get it super, super clear and then you hand off to an engineer, an AI engineer and just let it work for a couple hours and you have the right systems in place, the right company infrastructure that works.
You can't just do that outta the box because what you need is you need really good AI native testing infrastructure. You need good AI native. Like project product planning, tracking infrastructure. You need, you know, really good code review processes that catch the AI in a way that minimizes how much humans need.
Because these ais are gonna create so much more code. If you're bottlenecked by humans reviewing every line, it's not gonna work anymore. Right? So you have to figure out how do you optimize this pipeline. The other thing that we've done, um. Is using AI to, to parse all the human feedback. So in our game we have ways for users to submit feedback and have their game state.
And in the past it's like we see a few, we, we look at a few feedback ideas here and there as there's clear bugs, but we don't read the vast majority of that. But now AI can read a thousand feedback Id. In two minutes and it can triage the top 10 issues for me. And then I can say, go do a deep dive on those and it can [00:32:00] load every single feedback.
Id can inspect the game state, it can read the code, it can deliver a clear, concise explanation of the problem. And then I can take that and deliver it to an AI engineer. So you start to see how. You can start to close this gap, right? It's like human feedback leads to AI development, leads to prs that get reviewed by ai.
And then, you know, one of the things we've also built is like preview deployments where, um, you know, you can create a PR and immediately have a link that anyone can test, right? So it's like the AI creates something, you review the link, you test it. So like you, there's all these things you have to do to build your infrastructure to, to grease the wheels, to make this flywheel of like.
Feedback and, and design insights into implementation. Make it as fast as possible. And eventually you get to this level where you know, less and less humans are needed on the low level, and you're just providing the vision and strategy. And maybe they're like high level project direction and so, and that point, and so each, each employee is doing that.
Yeah. So that's where you need a different type of team. Yeah. Right. So they have to be extremely high [00:33:00] agency, extremely high agency, extremely high product judgment. You want people that are obsessive, they have to be really good systems thinkers 'cause they have to understand how all the systems connect to each other to create these like architecture or like strategy specs.
Um, they have to be, you know, really curious and adaptable to constantly adapt to how things are changing. So they have to be very comfortable with speed. Yeah, very comfortable with speed. So we find that like, you know, our last set of hires. Like half have been ex founders, right? Or, or almost ex founders, right?
Like, you know, maybe they didn't start a company but they did a GitHub project on their own. They got like 3000 stars, right? So it's like these are the types of, you know, talent that we're getting. It's a combination of that. And then also, you know, sometimes there are these like young cracked AI native folks who have such a high learning rate that they can keep up with, like the pace, and they're able to learn how to, how to leverage and wield AI so effectively that they can climb that ladder.
To have the judgment needed much quicker than than other people. So it's a very highly talent dense team. Um, and you need to [00:34:00] make sure they have the right traits because it becomes less about knowledge. You know, I think in the past we hired a lot of people for their knowledge, and more and more the knowledge.
There's some specialized knowledge AI can't capture, but, you know, AI's got a lot of knowledge. Like there are many areas. AI is much more knowledgeable than I am. Um, and so what matters is like. Really great product judgment and taste. Really great agency. This obsession to drive things forward, right?
Where you know, you're like, you're so obsessed on building a great product that you're just like. You stay up till you know, you, you, you, you go from seven to 10. Not that we expect that. I don't, I don't expect people to like do in crazy hours, but I, I, and like several people do it naturally just 'cause it's so fun.
Right. It's so fun to just like, build compelling products. It's its own game that becomes very compelling. Yeah. Interesting. So, because you're so deep into the products and you've been so on the forefront of the thinking, you've kind of become the, the go-to guy, everybody calls, right? Yeah, I think I, I, I mean, I.
I like, I think there are a lot of smart people in this space and I think a lot of like really great people doing really great [00:35:00] work. I think our advantage is just like we were so early, right? Like I think we, we were so early and that gave us a couple things. Like we've gotten the chance to like, you know, think about these problems deeper and longer than people have had time to yet.
So I think we've been able to build a lot of deep mental models on this space and learn a ton with AI dungeon that's been operating at a very high scale. Um, and I know you take a lot of pride in just being helpful to other people and helping them along and pointing out where they could do better, because as, as, as the whole sector emerges, uh, it benefits everyone if Yeah.
More, more and more people are doing better. Yeah. I'm very much just excited, like, I got into this, not because I wanted to be like. You know, a super successful startup founder, but just because I was a nerd who had a dream, right? I like, I saw this vision of the type of experience I wanted to play, and I was like, oh my gosh, this would be so cool to create.
And so I just got sort of like sucked into that. And so I'm super just excited as a gamer, as a, you know, tech enthusiast to see all the amazing things people are gonna make in the AI game space. And I think there are gonna be so many amazing things. You know, different than [00:36:00] what we're making that will bring a ton of value and I'm, I'm stoked to see that and be helpful, you know, however I can on that.
Yeah. And so, so you and I are hosting this, uh, with Silicon Valley Bank. We're hosting this, uh, gaming, uh, AI gaming Summit on, uh, March 8th Yeah. Of 2026. And, uh. You know, we've been very selective about who we're bringing. But it's, it's gonna be a really fun group and, uh, you're gonna be giving a talk and gonna be, uh, you know, running some, uh, user generated conference type things and it's gonna be a great day.
Yeah. It's gonna be a great day coming up. Um, and so you keep talking about this magical experience. So what are some of the features that, that we can expect to get in Voyage and AI dungeon as they go forward that are feeling really AI native? Yeah, so I think one of the things that like. Has been the most impactful is just like the connection to characters that you can have.
So I've had several experiences playing Voyage. Um, one where I, like, I woke up at like 8:00 AM on a Saturday and I just like got so hooked. I played till [00:37:00] like 9:00 PM that night and it was just like this whole arc of a story that was unique to me where. I started off like on the run from this like void queen, and she was like oppressing all the people.
Everyone was scared of her. So I, I, I originally was like gonna get arrested and managed to escape, you know? And like, I was a fugitive on the run. And everywhere I went, all these people were like, you know, I'm like, I'm like, I wanna fight her. And they're like, I, I won't touch them. And eventually found like these rebels.
And even the rebels were like, ah, nah man, we like, we don't, we don't fight her. We just sort of like. Disobey, you know, along the way. But, and so I, I was like this very lonely experience and then I met these like two companions along the way. So in one, I was like in this archist workshop that was like abandoned for 200 years and there's this like robot named Sentinel Prime, like tries to kill me and I managed to like override his like.
Defense mechanism. And then he like switches out of like, you know, kill mode. And then he starts talking to me and like, but in a very robot way. And like we start building this connection and then we also meet this, like this void crow there too. And so we go on this adventurous central prime and we're like, [00:38:00] you know, we're on the run from imperial patrols.
We like are killing them. We're like having these conversations along the way. We have these like touchy moments where it's like I hug him and he's like not sure what to do 'cause he is a robot and has this like, very funny but like very heartwarming like. You know, robot dialogue around that. Um, and, and like it was just this experience where like, no one else gets me.
Everyone is against me. And I've got this one friend of Central Prime and, and so like eventually we to fight the empire, we're like, we need to leave the empire to find allies. So we go to the north and there's like these northern ice tribes and they're like these Raiders on horses that are like fighting the empire and these raids, it's very like dune, like Remen p tradies.
And so we're like, they like, are super distrustful. They will not trust me at all. And that, and that's one of the fun things about, is like social persuasion becomes such an interesting mechanic because you're like, how do I convince this person to like be on my team? And I had some missteps where they were like, I tried some things that like went very poorly and eventually I was like, let me just fight with you.
And they're like, okay, you can fight with us. So we were like in this epic imperial [00:39:00] Patrol and we like managed to escape and you know, and, and, and then we got hit by this ambush. And in that ambush, central prime got destroyed. And so there's this moment where the AI like. Recognizes a sad moment and plays sad music and like I try to put Central Prime's core back together and my crow like flies over and sits on my shoulder and it's just like.
I, I, he can't come back. Like he's, he's gone. And I like, I like started like almost crying, like, well, you're tearing up now. Yeah. Like my, my room, I'm, I'm tearing up now. My roommate came back and he's like, are you okay? And I'm like, Al Prime died. Like, it was this very emotional experience because he was the only one that believed in you.
He was the only one that believed in me, and he felt real, like, he felt real. He didn't feel like a, like a, a scripted dialogue tree. You know what I mean? Yeah. Um, and so eventually, like, you know, I, I have this other companion, um, who, she's the leader of the, of the ice tribes, her name. She's like Astrid, I think was her name.
And I eventually like get her on my team and she gets like a few elite warriors and we make this whole like, heist plan. We're gonna sneak it under the [00:40:00] palace. We're gonna like, disable her, like, you know, defense mechanism, blow them up and then we're gonna like go to the palace and go fight her. And we go to this whole plan and it's like, it's super epic.
And we like have this final bite with fight with the, um, the void Queen. And like, we almost all die. And like, I just barely survive with a few HP and like Astrid's knocked out and like three of my companions that the, the other allies are dead and the crow like hops over to me and is like. Nudges me. He is like, wake up little warrior.
Like, you know, your friend, your friend needs you. Like it needed me to save it. So I like crawl over. I like save Astrid. We have a funeral for one of our companions that died and we're just like, we're all like sort of like have this like beautiful touching moment and like it was just this experience that was like, I haven't had that inexperience and it was mine, right?
Like. There's something so powerful about a story that's yours. Mm. It's, it's like how, I don't know if you've ever like run, done a chat g bt query and like you're really interested in the output and you share it with someone else and they're like, yeah, sure. Right. There's something about when you are like the [00:41:00] initiator, the driver, right?
Like you, it, it becomes so meaningful to you. And so that was like such a meaningful story and I've had a ton of others. I've had other stories where it's like, when was this? This was pretty early on in voyages development even that was probably like a year ago. A year ago. And it's gotten a lot better since then.
Got it. So that was a year ago. And you can still remember all the details of this? Yeah, you can, you can tell the story incredibly quickly. I'm tearing up. You're tearing up. Yeah. And that was a year ago. Yeah. And when are you planning to launch Voyage? Um, in the next few months. Next few months sometime.
Yeah. So you've been under development for about a year and a half and Yeah. And it should come out then. And so that kind of an experience is, is what people can expect from this? Yeah. Could you pick up that story? Yeah, yeah. After, after your 14, I could keep playing 14 hour binge you. Yeah. Yeah. I could keep playing.
And I think in that story, eventually like. You know, took on the crown of the void, the void queen, and became like, you know, had this new story arc and there's this corruption hitting the kingdom and like this new, you know, plot arc that started. And I could have kept playing that. Right. And I think I was like, I was ready for like a new story and a new space.
And [00:42:00] that's what's interesting is you can, like, you can always start a new story. It's always fresh. Right? I, I, I suspect Nick, that you just didn't wanna be corrupted by the power of the boy queen. Yeah. It's like, then it turns from. Like coad, we said no to the ring. Yeah. Now we go from dune one to dune two, and I go from the hero arc to the anti-hero arc.
And the, that, that's a cool story. But like, I, I didn't wanna solely the memory of, you know, sentinel pride. We gotta That's true. Sentinel pride. Yeah. We gotta keep their memory pure. That's awesome. That is awesome. Another thing I just wanted to understand, uh, before we wrap up here is. There has been in the gaming community, if so to speak, a lot of talk about sort of hostility to ai.
What types of hostility are there and where's it coming from? So I think there are a couple things in that. Um, I think one is, I think people are. Especially internally at game companies are like nervous around how their jobs will change and what this will mean. And this [00:43:00] is a cycle that's happened before it's happened in the gaming space before as, as things went from 2D art to 3D art as things went from like animation to mo motion capture.
Like it, it's happened many times. Um, but this is an especially dramatic one. So I think there's a lot of like, sort of like game industry people who are very like kind of. Resistant to the way things will change. Sure. Because they're valued and loved and, and, and, and, and paid because of their skills and knowledge.
Yeah. And when that skill, those skills and knowledge are less valuable, it's scared. Yeah. Yeah. And this makes total sense. And I think, and I think it's hard 'cause it's like, you know, I think they wish for a world where AI doesn't happen, but the reality is like it's just gonna accelerate and there there's no stopping it.
So I think the best move for them is to. Lean in, right. Figure out how to use the technology to deliver value. And some of them are and, and are building really cool things. So I think that's number one. I think the other thing on, on the user side is I think, I think some users are rightly like skeptical and rightly resistant to some of the AI efforts that game companies have done.
'cause [00:44:00] I think some of the game company efforts have have broken the rule of like. Great game development, which is like always deliver more value to your users. And I think, you know, where AI is used to like deliver the same product, but just like faster but worse, like that doesn't really help users.
Because I think one of the things is like users, like there's lots of games out there that are low quality. And so like if it's just the same thing, but slightly worse, it's like it doesn't deliver value value to them. And I think there's a perceived lack of authenticity, right? Where it's like you've built up this whole.
You know, company ethos of like being creative with your art and your story, and it's like, and then shifting that feels sort of inauthentic. Um, and the user can feel it when they play the game. Yeah. And, and so it feels inauthentic. So the people working in the gaming and company and then it feels inauthentic Yeah.
To the users. Yeah. I think, I think that's where some of the reaction is coming from. Um, I think, you know, for us it's been very different because like we've always, like we've, we've obsessively focused on how does AI [00:45:00] deliver more value to players. And I think, I think at the end of the day, that's what they're gonna care about is like if they, if it delivers a drastically better experience.
Because it has ai. I think players will love that. Yeah. Um, your community is built on a different premise. They've come to you expecting you to give them some amazing AI experience, and that's what you're focused on, as opposed to people coming to the incumbents saying, give me this great scripted experience.
Yeah, yeah. Which think gets a little bit corrupted with ai, particularly last year when it wasn't as good. Yeah. And so that, that creates a, a little pushback right now. Yeah, yeah. Yeah. That we'll probably get through in the next two, three years, four years. Yeah. I think it'll shift over time, especially as players start to.
Get value out of the experience because I think, I think part of the problem is just like, you know, for developer companies. If they're not doing truly innovative AI gameplay stuff like we are, it's just speeding up their pipeline. And players don't really feel the value of that. And so like, there's not really, like, they're not excited about it.
So it's like, I think players have to feel the value of something to be excited about it. And so at the end of the day, it's like we've just gotta ship [00:46:00] something. We've gotta ship AI powered game experiences that are really fun and players will. Appreciate and, and get on with that. And we've seen some, we've seen other, you know, both AI dungeon and, and there's been other experiences that have started to do well, um, leveraging AI and, and I think players have have appreciated that.
Interesting. Well, Nick, as always, man, I always learn a ton talking to you. Yeah, thanks. Great to be here.