Sebastian Park, co-founder of Infinite Canvas and a favorite recurring guest, joins host Aaron Bush to dive into the fascinating topic of building games on new platforms and with new technologies. They discuss how Infinite Canvas rapidly iterates, uses off-the-shelf LLM tech, and what the team has learned lately from building on Discord, Twitch, and Telegram. If you want to better understand how emerging AI-based tools will increase the baseline requirements for competing in game development and what emerging platforms to be most bullish on, make sure to give this episode a listen!

We’d also like to thank nSure.ai for making this episode possible! As a proven industry leader, nSure.ai provides scalable payment fraud prevention that’s not just effective but tailored specifically to your needs. To learn more, visit https://www.nsure.ai/contact
This transcript is machine-generated, and we apologize for any errors.
Aaron: Hi, everyone. I'm your host today, Aaron Bush, and I am joined by one of your favorite recurring guests, Sebastian Park. As you may remember from previous episodes, Sebastian, who is co-founder of Infinite Canvas and a venture partner at Bitcraft, has been building on the bleeding edge. His team was early to build on platforms like Roblox and UEFN, and more recently has been making games on Discord, Twitch, Telegram, using LLMs, and beyond.
It's been six months since we last caught up on the pod, and of course a lot can change in six months, so Sebastian is back to chat about building games. With new technologies and on new platforms, and we'll try our best to squint into the future and gauge some of what comes next. So it should be a super fun conversation and with that, Sebastian, welcome back to the pod.
Sebastian: Thanks for having me, Aaron. It's good to be back. It's always a fun time to talk. I believe today is October 23rd. So late. Beginning of fall, mid fall, leaves are changing colors.
Aaron: There you go. So it should be fun, a good, almost, it's a good Q4 wrap up looking back on, on the previous year and a lot of ways.
And really my first question to you just to, to kick stuff, to, to kick this off is just what's new? Like, well, how would you say? Infinite canvas has evolved over the past six months. So have you changed or not as an entrepreneur, like what's been going on?
Sebastian: Yeah, we talked a bit about this before the start of the show, but I'm a huge believer in the even keel perspective of the world where, you know, we have good days and bad, like there are days where everything's not working and there are days where we find huge breakthroughs and, and it's actually really nice to look back on the company and the team over the last six months, we spent a lot of time doing really core game design, but also a lot of middleware LLM manipulation.
Right. We spent a lot of time with everything from Monet and Claude and Gemini and GPT four, no, trying to get the type of responses that consumers would want out of our products. Right. And that was the name of the game. The last six months, we learned a lot about it. It's funny to say, but I would imagine I'll, I'll talk to a friend about this.
If you were to survey people in the world, uh, I would say we're probably like, like, you know, maybe second or third best. In the world at, at this type of like development and the people who always beat us out are the, are teenagers and people in there, like in college, right? People who have more time, like those guys are crushers, right?
Like, I don't think there's, there's for sure the case that there's some like 16 year old out there. Who has spent like nonstop time working on this problem and probably understands this even better than like I do or anyone on my team does, but outside of him or her, we're, we're, we're really up there in terms of our understanding now.
Aaron: Nice. And with that, I think it would be interesting to just Hit on again, and we, we may have talked about this a little bit in the past with just the infinite canvas approach in terms of, of how you guys work, because it seems like a pretty small yet fast paced team. , you know, even this past year, your team has launched several interesting projects that are all built with different technologies.
They all live on different platforms. Sometimes you have, you know, one game that's been built on different Technology is on different platforms, and we'll talk about the games and a moment and through that some about the emerging platforms. But can you unpack like what the infinite canvas approach is right now?
Like, why build so many ideas so quickly and so many places?
Sebastian: Yeah. So for context in the last four years or so. Infinite canvas has shipped about a hundred games. Wow. Like that's sort of what's good and bad, right? Like the, the real reason is, is our own internal philosophy around what works and what makes sense.
And so we're huge believers in the. And the idea that we don't know what we don't know, the unknown unknowns, the fact that taste change, technology changes, what's big one day is different. The other day, the definition of Ohio has changed the last three years for all. You know, there's so many different things that occur and things move so quickly that the only way to really have a point of view.
When you're targeting these different audiences is to have one, get it out as quickly as possible and figure out what other people think about the point of view. And so that's been our philosophy from day one. So we, you know, really our iteration processes like, Hey, understand platform, find a core loop, a core feature, a core product ties that everyone likes.
And then run from there. And so that I think is, is really the, is really been our bread and butter in terms of how we've operated as a team and our core team, like the core engineering team, uh, product team has been effectively the same for people this entire time, and so that also helps as well.
Aaron: Yeah, for sure. I guess I'm also just curious about the nuts and bolts of that, like, like, how has your team been able to iterate and ship so many ideas so quickly? Like, like, how have you gotten better at that? Or like, do you have core development philosophies that drive your ability to go? You know, pull off so many different iterations and tests.
How does that go?
Sebastian: Yeah, there are two things I think are really helpful for us that are from very different worlds. The first one is the more game jam slash hackathon mentality, which is, Hey, go home. Think of an idea. Let's put it on the board. Let's do this in person. Let's talk through this in person.
We'll have like an offsite or an onsite. Jump into a conference room. I always tell people I want my meetings to be either 30 minutes or four hours. And this is a very much like a four hour style meeting where we're like, just in a room or it's in pizza or salads, depending on how, how much older we get.
We we've gone, gone away from pizzas and more towards salads recently, just to stay up here, but, but we'll like be in a room for like a good amount of time and just shoot ideas with each other the other way, like a writer's room, basically the other thing that we do, honestly. Is we do a lot of yes.
And like from improv, which is there's no such thing as a bad idea, even though there are a lot of bad ideas out there, let's like riff on the idea until we got to the place where we have a lot of different features and a lot of different implementations that we have. So once we have that in the locked and loaded, we just order that sometimes randomly, but oftentimes based on like what we think might work best.
And then we ship right? Like, then we like, cool, ahead of engineering, Brazil will go out and effectively mock up exactly what the different parts are. We, we then divide and conquer. We ship the product. We ship the feature. Usually we do in a few days and then we get out to testing. Just to see like how people think and while they're testing, we're already working on B2 of the feature.
And then really the testing feedback is going to help our B3 as a feature.
Aaron: Makes sense. Cool. Let's shift gears and actually talk about the games themselves. And we can kind of work through how some of this has actually worked in specific, circumstances. And through talking about these games, I think it'll be interesting to our audience to, to just continue to understand what you're seeing.
On these emerging platforms because there's obviously a ton of discussion about like what's the future of discord or you know? Like, you know, there's rumblings about like, oh people are building games on Telegram. But like, yeah, you know the actual lessons and insights from being in the trenches I think is going to be super interesting, so let's let's start with wordcraft , which I have been playing over the past couple days, and there's a couple interesting things to discuss here, but first, , maybe you could just describe the game for people and maybe help people visualize what this is.
Sebastian: Yeah, for sure. So if you guys are on literally any platform on earth, I think, except the PlayStation five right now, you can look up Wordcraft, where wordcraft.gg slash game is where you can play on the HTML five version. It's a game just to take a step back, a game built on. Off the shelf, right? We're doing a lot of cashing, so it doesn't feel super slow, but if you are sufficiently bad, the game or sufficiently exploratory in the game, you'll get to this place where everything is actively pinging an LM to figure out what the responses are.
We don't know. The start and the end of the puzzle. It's a puzzle game where you either combine or split or manipulate words. And like in the lexicon of English words in order to reach a goal word. And so in doing so, like the canonical example we have is if you want an orange tree. Well, how do you get, and we give you the colors red and yellow.
Yeah. Red and yellow makes orange, red and yellow. You can combine those to make orange, combine that with tree to make an orange tree, right? Really funny, uh, puzzle like mechanics in that, in that manner. So that's WordCraft in a nutshell. It's also very fun. You just see experience in a lot of ways because.
You can actually, this is actually based on some like hilarious language theory, which is that you are able to do some like seven, seven degrees of Kevin Bacon, get from any word to any word. It's like fundamentally possible. Like you can do it. It's actually, it's like, there's no way not to do it. You can get back down to the rudimentary bits and build it back up.
If you have split emerge and any number of words, but the fun part is like, you know, having people feel good about the exploration into the end goal.
Aaron: Yeah, I thought it was, it was fun. There were a couple words that really kicked my butt and trying to figure out like, like fossil, like that word, trying to figure out how to get from the, the original word, which I don't remember exactly.
It was like clock or something to fossil. I was like, what the heck? Um, but it was fun. And it's like, it's an interesting, like primitive, , that, , you could see like as a loop being able to build on and an interesting. Yeah. Ways, but , tell me more about the llm side of it I guess first of all you said that you don't know the beginning middle or end of some of these puzzles Like like how does that how does that work where you can create a game but not actually know what is being put forward in front of you Players and be happy about it.
Sebastian: Yeah. Yeah. I mean, so there's a lot of tuning involved, right? There are a lot of times where we're unhappy, right? I think the problem is the core primitive is fairly self explanatory. We're trying to get from rock to Island. How do we get there? Right.
You have, you can implement verbs. We call them split them, merge them, intensify it, opposite, et cetera. The, the, the player experience is the hard part. The actual mechanics are what they are. Realistically what's happening in the backend is we, when you say split rock, we're there, imagine there's a chat GPT call to GPT for being like, Hey, if you want to like split rock in the like lexiconical sense.
or lexical sense, what are the top number of choices that you'll get to, et cetera. You combine that with just a bit of off the shelf machine learning, meaning, Hey, just like iterate Monte Carlo, iterate through all the options, see what pathways there are, give people hints, give people direction. You get to a point where the game sort of plays itself in terms of level.
We then do a lot of tailoring. So like the first like 10 levels you play. Are the ones that we think are most intuitive and then we try to wrap you up more and more, but certainly it's something that everyone can do whenever they want.
Aaron: Got it. And how did working on this project shape your view on how building via LLMs will affect game development?
In the future, like, did your view become any clearer or any more, like you, you just feel more sure about how certain parts of the future will, will work in, in game dev from this project.
Sebastian: So I would say game development and game design is really hard. Like that's like unequivocally true. Right. We are so far away from these LLMs or some of the gen AI tools really showcasing itself.
Well, we are so far away. It feels like early mobile when like the top games were games like sword and sorcery, like a 60 minute RPG, you buy it for five bucks. Right. That is not what we imagine the best use case of a touchscreen or a mobile game is right. Like, I think there's a lot of space here to design on, but it's not easy.
Right. I think there's a lot of focus, especially in venture, but also in the greater world on two opposite sides of the funnel. There's a lot of focus on foundational models, like making the models themselves cost about 80 to a hundred million dollars per pop to just run a test. So billions of dollars of career real model.
So probably, and also you have to be like a top 25 research scientists in the world, right? Like, yeah, I'll be not the best implementation or use of time there. And there's a lot of talk on. Really the, like the stuff that we've talked about before, how do we productize the world? How do we make it easier for QA to operate?
How do we make it easier to do, improve our production flow and design flows? The stuff that hasn't been as explored, and I think people are starting to understand is incredibly important is the middleware part of it. It's really understanding and unpacking that actually you, we can't just give people language and hope that they have a response that they will like, right?
The English language is not a logical language. It's a contextual language. It's a language that has multiple ways to say there, and it means different things. And it's spelled differently and people screw up all the time. And so because it's a contextual language and because it's not logical, a turn complete, the way people interact with it is different, right?
People can get frustrated when they split, let's say a box and they're expecting cardboard, And a square, but they get instead cube and paper, right? Those are both fine implementations and understandings of ways that a LLM, if you assume as like this, like nebulous thing can like split up cardboard box into, but the problem is, is that like, if you, if it, if it attacks and defies your expectation, you might be pretty annoyed.
You might be pretty upset with the game flow itself. And so there's just a lot to be done there. I will say it makes it super powerful though. It is word craft as an example, as a game that at scale, we don't need to have nearly the same number of folks as a candy crush at scale. It can like interact with a variety of different platforms really quickly and tailor very fast because of the nature of the game.
And so that I think has been really fun. It's like building your own engine effectively, like they used to do at blizzard where you're like, Hey, but instead of building an engine, that's. The Maya components and the 3D rigging and whatnot. The engine here is, hey, how we can manipulate an LLM or any type of off the shelf AI to interact with users really well.
Aaron: Yeah. So are you building the middleware AI for just specific games? I like, do you think like that's the sustainable path in the future? Or like, do you see, I mean, I imagine there's going to be so many Tools, built to kind of act as like converters between like these big LLMs that aren't made for games to actually be able to take it and make it much easier to use in different ways for, for game developers.
Like, like, how do you see like actually working with these things? Changing over the next few years.
Sebastian: Yeah. I'm, I'm often bearish on, well, let me take a step back. I don't know. Right. I think one of the beauties of being on the cutting edge is like, you have to have a healthy amounts of just being a fine with the fact that you don't know.
And I have no idea, right. It changes all the time. I will tell you the things that haven't worked. Those are really easy. I haven't had that much success. With the litany of AI startups that say that they are a better version of talking to these systems. I haven't had any success there in part, because at the end of the day, it's less about language and more about intent.
Like I want you to understand my intent as a creative or founder or, or engineer, not necessarily what you believe the intent ought to be. Right? Right. The beauty of programming languages that's often misunderstood is that the reason why it's so logical and oftentimes code like is because it can be read by every machine the exact same way.
Right? Like that type of understanding is pretty interesting. What I will say is that we've seen some implementations that are better than others, but the one that works the best for us is really trying to give as many contextual clues to people such that they understand how it works for them and in their game.
Here's an example from us. We have a lot of art assets, right? Our, our, our art assets in the games are often unilaterally with the exception of a couple of characters generated by some combination of mid journey or Gemini or whatnot, it's made. In real time, it happens as you play the game. If you go down a weird path where you're like trying to do something new, it makes the art itself.
Right. What's interesting is that there are three ways that art designers typically communicate with each other visually. They like show you the mood board. They show you the storyboards orally. They talk to you through like what their intent is. And then like notes, like they like write out on the bottom with words, like, Hey, here's what we have.
And what we've basically been doing has been like less the, like span, like fancy, we're trying to train our own LLM. That's nothing we'd ever do. It's far more actually in this regard, which is. Hey, we want this art asset to have this style and feel show us exactly what the words were used to generate it as well as the art itself that came out of it.
And then over the period of a few weeks or a few months, you tune those things such that the words and the image match the taste that you're trying to generate. Right. And so that I think has been a really cool process of this. And in that way, yeah, the stuff we're working on is fairly bespoke per game.
Like there is no overarching middleware thing that's working for everything in part, because we're tuning it towards our audience. We're tuning it towards the people we're talking to.
Aaron: Yeah, makes a ton of sense. I want to talk about a couple other games, but before that, I do have a couple other questions, more even specifically about WordCraft.
One just has to do, I mean, all the LLM stuff is interesting, but even beyond that part of the technology, like, this is a game that was built differently and tailored for, Platforms like, um, and your, your notes to me, you told me that, that WordCraft was built on Godot for mobile HTML5 and JavaScript for the web.
And it's even a discord activity. And I'm just curious, like. Like, do you think that this approach of tailoring games to more platforms is going to be more common in the future? Or are you just trying, you know, to build these games in different ways to experiment and see what sticks like, just tell me more about like, why you're taking this approach technologically.
Sebastian: Yeah. I mean, so first of all, like the, the coolest part about. You, the type of ways that we've been building these games and the style of games you build is that it does work on multiple platforms and that you can tailor them and you should be tailoring them, especially with the tools we have today.
Like it is possible to tailor a more mobile experience for people who use mobile. Right. And that was great for Godot because we were able to get people to feel effectively the type of progression you will see and stuff like Candy Crush as an example. For HTML5 JavaScript, you know, really, you want to give people a flow that's similar to a wordle, right?
Like you want them to be able to come back and engage differently. This core activities, another fun platform to build on. So I'll say there are three main things here. One, our team is capable. Like you do need some senior engineering to be able to context switch between platforms. Once you have it, it's nowhere near as difficult as it used to be.
Like it's something that is possible to do. And two. You want to tailor it to platform to see if there are versions of the game that make more sense for that genre of folks, you're effectively having a, you're shooting an arrow down range. You want to make sure there's enough targets there for you to hit.
I'll say the final thing there is, is this a, you ask, like, is this an approach people can take? And I think people should take it. I think people should be tailoring more. Towards different products, if they have the opportunity to, there is a chance that will work better one way or another. And I think that's something that like we don't see as often in part because there's been a large amount of technical overhead to do so.
But as that bar comes down, especially with the integration of sort of this like runtime AI bits. You don't run into the same amount of headache moving one thing to another.
Aaron: Interesting. And one last question here. The, in our previous conversation, six months ago, we talked a little bit about how maybe one day some of these LLM platforms, like chat GPT could actually become distribution mechanisms, for games and If you think about it, like it's probably like these types of platforms are the next big consumer internet platforms that have the potential to to reach billions of people, which is a pretty powerful thing to build on.
How is your thinking changed there? Are you still bullish or at least curious about that?
Sebastian: Oh, we're so bullish and still super curious. The number of players is starting to shape out. I think that's something that we didn't expect as quickly as it's been happening. Like the number of players has been shrinking down to your effective paying companies.
But we are starting to see what that looks like for Gemini, right? Like, we're starting to see what Google looks like. Open AI, I think, has been leading the way in terms of integrations there. I'd imagine Perplexity is going to go in that direction as well soon. So, these are things that they need to solve for themselves, but once they have it, they clearly have the eyeballs.
And I, myself, find myself asking an LLM, Questions more often than Google most days nowadays. So it's moving in that direction for sure.
Aaron: Okay, cool. Well, let's go ahead and shift gears and talk about creature craft , you put on discord. So I kind of want to talk about discord through the lens of this game too.
But I guess, first of all, again, like. Just tell us about the game. Like, like, what is it? Why did you build that game? Why put it on discord? What was your thesis behind?
Sebastian: Love creature craft. Creature craft is, it was initially a bot we're turning into an activities as well, but it leverages runtime AI to generate new levels for you to play indefinitely.
We wanted it to be effectively something where between an idle RPG. And an interactive game that you would play in a chat functionality when we built a game about a year ago, the visualization parts of AI, the Jenny I visualization wasn't as robust as a text part. And so it made sense for us to pick this court initially, in part, because of 2 reasons.
1. Texas, a huge part of the ecosystem. There are bots available. We could totally leverage that, but two. And I think also interestingly, this court is like the second or third largest social media network in the world. And people just like, don't think of it that way. Like this score in 2024 is probably my favorite platform.
It's a platform. Why interact with people far more than anything other than memes on Instagram. Right? Like this is really, really that type of next generation social network, right? And they're offering voice over internet protocol in these communities really as a top of funnel, just like how facebook offered status updates and photo uploading as a top of funnel to allow you to do a lot of different things.
We are so positive, bullish about this court in part because. There are so many eyeballs there. There are a lot of folks that are really interested in being effective, engaging there. And we've, we've quite liked working with the discord team as well.
Aaron: Got it. Well, tell me more about your experience building, on discord, I guess, probably early iterations were robots as you were alluding to, but more recently discord, you know, release there. Activities, we can maybe talk about like, like what this is, but like, what has been your experience or early lessons learned so far about like building these activities on discord?
Sebastian: Yeah, the notification loops are probably the place where we're spending a lot of time on right now trying to understand how people use discord and interact with discord.
You know, some high percentage of discord users are mobile only randomly enough. Like it's like a hard thing I think for older millennials to understand, but most Gen Z probably only have discord on their phone. Some, I remember it being like 20 to 30 percent of all discord users are mobile only. Right.
Like. That I think is a really important part of the analysis in terms of how we're thinking about it and how we're approaching this court. The second part is, you know, this court activities right now, they just did the public launch. We're starting to see some new games come out. These games won't be the same games that are on the platform in six months, right?
Like there is a bit of a learning curve and growth to it. This court activity that we've seen. Is often communal based in that you see groups of people coming together and hanging out there. It's effectively a digital second home or, or what's that, what's that term again? Second space. Is that right?
Aaron: But it's like, yeah, your third space, your like homework and a third space.
Sebastian: This court is often becoming honestly. A digital third space for a lot of people. And it's reminiscent of like cheers or friends or how I met your mother. Like these types of like three, three camera sitcoms where they're like all going to the bar, that's really where I I'm seeing a lot of activity on discord for people come home.
They like have their group of friends that they hang out with and they just like jump into the channel and they're hanging out there just like actively lounging. It's, it's also reminiscent to use a more. I guess more modern ones is people have now seen sitcoms, but like, well, like the, like what the different capitals of world of warcraft were for that game back in the day where it's like, look, is this where everyone's, everyone's around the auction house.
They're all shooting the shit. Like it happens, right? Like that, I think that digital dirt space feel is really interesting. So the games that support that are super interesting. They ought be more asynchronous. I'd imagine like it's, or incredibly synchronous. We've seen a lot of like party synchronicity, like.
Party games, Jack box style games that could make sense, but I think there's also room for some asynchronous back and forth, like your mail order chess as an example. And so I think that's been a lot of fun to explore the rapper itself is pretty intuitive, right? Like if you know, HTML five at all, you know, it doesn't use, it doesn't use frames as much as it uses like no JS as a rapper.
It's a, it's doable and it feels a bit like web with some other constraints on it. And then in that purchase ecosystem, there really shows this court's desire to perhaps move to, you know, six facebook, which I think would be the most exciting world to be in.
Aaron: Yeah. Well, how big of a moment do you think this is for, for discord?
Because, you know, as I see activities, I mean, like what's most interesting is, being able to build games primarily that tap into just like the inherent multiplayer nature of a social platform and I don't know much about like the monetization capabilities other than I guess it's, it's possible now.
And I, I, yeah. Assume they'll make it more robust, but, maybe it would just be interesting for people who are more watching from afar or haven't attempted to build anything on, on discord yet. Like. What is most interesting about this moment? And like, what does it say about like where this is going to go for the platform, if you have thoughts?
Sebastian: Yeah. So I will say the most interesting thing about this moment is the, I don't think it got nearly enough fanfare for what it is, right? This is a, like I said, in arguably one of the top social networks in the world. Launching effectively a games and product platform. That is not a small feat. That's not something you should just ignore.
That's something that there's a lot of potential value there. Right. Those succeed, right? Like this is not an estimation on any individual person succeeding or the platform succeeding, but this is an indication that, Hey. Here's another place where they want user context. They want productized bits to interact with communities.
They love that part of the ecosystem. I think Discord is moving more and more towards that direction. They're becoming more and more a third home productized area. I think that's like really bodes well for people in gaming, given that they're. Initial tranche of things was gaming. So I think that makes sense.
Aaron: Do you think like in the same way that like on a platform like Roblox or, , I guess maybe even UEFN, like kind of like the initial builders were more individuals, but at a point it became more professionalized and we see like venture backed teams building all sorts of stuff on, on, on Roblox. Do you think like a platform like discord would, will have kind of the same type Trajectory or like is what is being built and therefore who is building on it just like something inherently different.
Sebastian: So we've already seen this even before and that's sort of the beauty of what this court is, is, is about, right? Which is the best e sports to use a random different example. The best e sports in the world historically have been ones that sprung up organically. They're the best speed runners in the world play games organically in that it's basically someone loves this so much they start building or they start building a community.
Next thing you know, you have something interesting there, even before this court supported anything, there was an entire ecosystem of individuals and small groups of people being like, it'd be really funny to have such and such game, like a drop, like a bot dropping game in our ecosystem. Let's create a drop bot, right?
Okay. You had this entire space of builders who are like, Oh, there's a lot of people here. I want to make something cool. I'll figure out and hack away around to do it. And so in a lot of ways, discord already has that community of folks who want to build for their friends and family and larger communities.
So given that, I think there's a lot to be said there. I will say the biggest difference between 2024 and like 2006, though, is that in the last 18 years, like the idea of a professional game developer has, and especially professional indie game developer has exploded, meaning that there are a lot of teams that, you know, make enough revenue to deploy capital and resources towards new products and platforms.
Some are venture backed, some aren't, right? A lot aren't, honestly. And so I think that's where we're going to start seeing a lot more movement is actually in that tranche of, Hey, these like three to five person teams who have been making things for their platforms. They like this score. They're already supporting this score.
Let's go there. I wouldn't be surprised. I think it's very intuitive for more board games, like physical board games to have like digital companions inside of discord. That makes a lot of sense to me. And so I think we can see a lot of movement in those worlds.
Aaron: Interesting. One more game quickly, your team released AI Derby, which was, uh, built on Twitch as like a live streaming, , experiment, and I know you guys killed this game, so maybe it didn't get the outcome you were hoping for.
But I'm, I'm still really curious what you learned from building and operating a game on Twitch that maybe you didn't appreciate before jumping into it.
Sebastian: Yeah. So AI Derby is a horse racing game where all the horses are generated, , using Gen AI. It's really fun. It's all runtime. People type in like, Hey, I want this horse to race.
They, if you ever been to a horse race, you haven't like, and, and like, you know, depending on where you fall, like it might be an interesting thing for you to check out, but one of the famous things about horse racing is that the names of horses are ridiculous. And so the idea came out, Hey, horse names are ridiculous.
Gen AI, like runtime generation of ideas and images is ridiculous. We will combine the two. We had a lot of fun with that game. I actually do think the game has some legs. It was backburnered a bit in part because we were working on WordCraft and some of the other fun things. The theory behind it, I think is incredibly sound in that the highest cost of Twitch streaming or content in general are the individuals.
If you can generate a style of game, like Twitch plays Pokemon, if you guys remember that from a decade ago, where effectively a computer, an AI, some other type of mechanism, a virtual announcer is operating the game without you, there are some like really low cogs to overcome. , we can talk about this a little bit with regards to like Wukong black myth and some of the other live ops games.
But if you don't have to hurdle these large costs in order to make money, it makes for some really interesting game design and really different styles of game design. And so that was a ton of fun. It was really cool to have announcers that were just like playing through and having people come in and like effectively betting on different horses and all that jazz.
I think there is something to be said about live streaming platforms, which you don't need to have. 10, 000 or 10 million monthly active users to be successful in the same way that you do right, which they're the way you generate value on those platforms is the viewable hour, right? And so given that you can generate such a high number, high volume and honestly, high like velocity of viewable hours, I do think that there it's a really fun platform and.
It's something that like, we had thought about both Tom, my co founder and I had thought about years ago when we were working in e sports and we couldn't quite make work in part because it's hard to convince people to stay up from like one to 6am every day for their entire lives, right? Like you don't have as much coverage there.
Aaron: Yeah, yeah, makes sense. And I've been interested in like the idea of these like massively interactive live events, the miles for a long time, which, , I guess, I mean, it really started with that Pokemon experiment from a long time ago. And, , I guess Jinvid specifically has Taken it more in a different direction to being more like, like in apps, like people will vote and do things that like affects story beats, but it's not necessarily like being on 24 seven.
And so, yeah, it's still a very experimental place to be, but sounds fascinating as like, again, like creating new primitives. learning how different platforms just like inherently have different business model dynamics, , and, and different like game mechanics that you can, bring to the front, the front because of the emergent.
Properties of these platforms. Like it's so fascinating. , , like a follow up to this is , we've talked about a bunch of platforms. Your team is built on a bunch of platforms, you know, beyond like what we've mostly talked about today, like discord, chat, GPT, Twitch. , like since we last talked, like, are there.
Any platforms that you have become more or less excited about or bullish about really mattering in the future?
Sebastian: Yeah, so I mean, I'm more bullish on Telegram. I'm more bullish on Discord. I think those are probably the top two that I'm pretty bullish on in terms of you just have large communities of people, you have access to them.
It's a really cool environment to build things on. I think chat apps as a result, like WhatsApp, I'm actually fairly bullish on, , which is not a, , environment anyone talks about anymore or ever. And so I do think that those are probably the top of my list. I think, I'd imagine there also is something to be said about, I am still fairly bullish about all the other platforms as well, right?
I think it's just that as. , I explain this to folks all the time. It's like, as things become more red ocean, , it's not that the economics change is that the rate of success changes. I still think those ecosystems are pretty a ton of fun.
Aaron: Yeah, that's exciting. Give me your like 30 second updated thoughts on Telegram.
Sebastian: Yeah. Telegram super awesome. Really fun. Lots of users. , I can probably talk more than that. , I'll say the biggest issue of Telegram is honestly the crypto aspect of Telegram. Yeah, what people don't recognize is just the legal headache involved in dealing with crypto if you're not in crypto, right?
And so the best payment accepting mechanisms on telegram are still ton coin fine Like like it's their platform They do what they want But the downside is as a game developer as like a as a US based company that doesn't accept crypto for anything else we do Like the amount of headache it'll cost us in order to accept crypto.
Pretty interesting. I also think that the press is a bit overblown, right? Like there's been a lot of pieces about how telegram is just like awful, evil place. And it might be in that if, if you have the world view that all social media is bad, then yes, it is a social media platform. , but if you have the view, that's another place for people to talk to each other, I think it's fine.
Aaron: Okay, well, we could talk about any of these platforms a full episode, but I do want to keep it moving. So we were both recently at the bit crafts annual event in San Francisco a couple of weeks back at this point, and you were on a panel that discussed the intersection of AI and mobile games. And I thought it was really interesting.
And one thing you said was that while Early adopters of AI tech may have certain advantages in the very near term over time. And as how all technology has always worked really, just give it enough time. It just creates a new baseline in which everyone is competing on level footing. Again, it's just the, the rules of the game on the field have changed and feel free to expand on on that point more. But really, my question is like, like, how do you think that baseline will expand over the next? Um, you know, you're working with a lot of this technology in a lot of different places, and I'm just curious, like, like, what do you actually see being that new normal, in the next two, three years or so?
Sebastian: To give some historical context, the analogy that always resonates is that anyone who's a top 100 baseball player in 2024 would be MVP. To 1970, right? Like it's just like anyone who is a top 50, a hundred basketball player in today's NBA would be an MDP 1970, right? Like the advancement of humans as a ecosystem, as, as a be a sport or tech or otherwise, is that strong?
If you think about what constituted like high level physics understanding in 1910, if you look at 2024, like that's probably undergrad physics at this point, right? Like it is really that different in terms of that evolution. We see this also in game development, right? Like what's considered the interesting, good game loop in 1994 is considered table stakes in 2024.
We see this in technology, having social reach out and having abilities to like Add your friends to your game in a mobile game is a no brainer requirement for making any type of social mobile game in 2024 and would have made you billions of dollars in 2009, 2010. Those things are what I mean by table stakes and what I mean by the creeping of technology.
What's cool about technology is that it gives you this window. But it has to be accepted that it's a window of time, as opposed to an indefinite period of time to push your edges in ways that other people are not yet comfortable with or don't quite understand. But once people understand it, it becomes just the norm.
Right now, we expect indie game developers to be able to make not only for PC and Xbox and PS5 and Switch and everything else. I can tell you when I started. Making for any new platform required entire new hardware stack, like you just need to go and like ask Nintendo for a developer license. You have to do so much to get stuff done, and so the table stakes today are just much easier to make for different platforms.
AI is changing a lot in terms of table sites. Number of humans. You can get so much more done with fewer people again. It's been weird in that gaming has gotten more and more complicated over the last half dozen, dozen years, such that these teams that continue to grow larger and larger. And what's been really interesting is that this new tech allows you to, as an indie developer or a triple I or even like a A or AA developer. Get two levels of fidelity and get two levels of iteration speeds that we would have only seen, , at the highest high of places prior.
So that's been really cool. That's almost certainly going to be just become the norm that you're going to be able to get our direction articulation done in 24 hours that you're going to be able to, like, people can show you the images can show you the prompts that come up with the images. They're going to show you the people involved.
They're going to get you to get, figure out your taste much quicker. And that's going to be the expectation there. There's a company called Latent Technologies that does a bunch of 3D rigging. That's completely AI generated. If their AI physics engine is the ubiquitous thing that you can just plug and play.
Every 3D anime game ought to have some version of a physics engine that's bespoke and tailored to whatever they're doing. That just becomes a table stake of how you Build out these games and so in a lot of ways, we're at this really fun time where we got to experiment and find these edges and then use them to build advantages and try to overtake the incumbents.
But if you wait and if you let the incumbents catch up. They still have the incumbency advantage, and then you are back to being normal. , my, my favorite thing also is, uh, the other one would be poker, right? Like poker. If you look at how poker was played 40 years ago, or even 20 years ago versus today, the people today are would be like super crushers 20 years ago.
Right. And so that I can't emphasize enough is the ebb and flow of technology. And so the idea that we're going to emerge from AI and like, Oh, you're going to have an advantage needs to be couched in the idea that the advantage disappears over time.
Aaron: Yeah. Which. Is perfectly put and makes a ton of sense to me.
I think it's called the Flynn effect, which, you know, has measured how like humanity, just like across all dimensions, sports, et cetera. You know, just like gradually levels itself up and then the baseline increases. And yeah, I remember that concept blowing my mind. , but maybe, and I think you described it really well for the games industry, maybe you could, , help ground it in more, , examples in terms of like, you know, in terms of like the, you know, both as like a founder, that's building games, but also as a venture partner, at, at bit crafts, like.
Any early winners in terms of like companies or tools that are like the facilitators that like you're pretty excited about of What is going to increase this baseline for everyone?
Sebastian: Yeah. So, first and foremost, I will say there is an awesome amount of people shipping things. And so one of the big table stake changes I've seen even the last five years is that, you know, a lot of people came to the market to raise money, to build a vertical slice, to show off their game.
Given these new tools, given the fact that things are easier to make, there is no risk in theory of you being able to make something like that shouldn't be the risk here, right? Like that, there is no reason to say that you're able to raise a few million bucks. And then you come up with a demo that you should have metrics in hand.
You should understand what your audience is. You should have communities, you have things built into it beyond just, Oh, I made the game because you making the game is just table stakes at this point. Like we assume that. When someone gets money, you'll be able to make the game in terms. So that's, that's in terms of like how it's changing the world in that world, in terms of how it feels like in the world of like tools and whatnot.
Like, it's weird to say, but I still find the most amount of fun and the most amount of values just from using the same stuff we've always used. In that, like, it's still GPT four, it's still plot. It's still Gemini. It's still mid journey. It's still these places. Those places are just improving their tools.
So tremendously. Yeah. Like all the time, like it has been hard to justify for us to like, be like, Oh, like, why should we work on this new up and coming platform and product? The only times I don't see that are people who aren't doing middleware. Right. So this is the people who are doing fundamental research.
That's where I see some really cool advances. I mentioned late to technology. I think, I don't know if you had a chance to check out their demo at the bit craft event. I was a huge fan. I thought they, they were doing some things that were like. Really cool. And changing what could be made. And I think that's the thing that I look for the most is not, Hey, will this improve my process of making stuff by 10%?
I'm looking for technology that like changes how we think about how we make games and how we make products.
Aaron: Yeah.
Sebastian: The, another example is something that we've built internally, but like, that we use all the time, which is. Just making sure the levels are geared towards you at a certain point, Aaron, as you play WordCraft, for example, depending on how good or bad you are, the we'll start adjusting the difficulty of the game towards you, which is something that we've always wanted to do in gaming, right?
Like we've always wanted to have people feel challenged, but not too challenged. Right. And so you don't want, I want to be the boss sheet as like the entry point for a game, but you also don't need something a four year olds can be. Right. And so that type of customization, that type of tailoring, I don't think it's just going to be for products and platforms.
It's also going to be towards individuals because the cost has come really far down there.
Aaron: Super fascinating. Well, we're coming up close to time. And so I have a, maybe a couple final questions for you. , and the first one is just. What's next? Like, you've been building so many things in so many different places, , Like, you know, as you look towards the next year or two, is there any areas you want to double down on and focus on more?
Are there any other, like, new technologies or platforms that, like, you're excited to test out for the first time? Like, how are you thinking about what's next for you and Infinite Canvas here?
Sebastian: Yeah, we're going to continue to work on the intersection of AI and consumer. I mean, that's someplace that's really awesome.
We also think there are some like really interesting non pure game plays here too, right? So gamification, , it became a dirty word for a decade because it was so poorly in 2013, but there is room for it and there's a lot of value there. So we want to spend more time on, on those parts as well. But right now we're heads on WordCraft.
WordCraft is our, our flagship game that we're trying on lots of places. So we're super excited to see what comes out of that in the next few months.
Aaron: Well, what's next for, for WordCraft, anything you want to share?
Sebastian: Yeah. Yeah. So we're doing some really interesting customization work, but we can't exactly share exactly who we're doing that customization work for, but because of the way of the nature of the game, it can ingest content from people and like, for example, if we could, we can spin up a custom Novick one, even right where we just like all of Novick's content and it'll make a bunch of puzzles and the link to the Novick, puzzles, , articles themselves based on the puzzles.
So it's pretty fun because we've been doing a lot of stuff like that recently.
Aaron: Sweet. And then, lastly, just one, one thing that came to my head while I was catching up on all you guys were working on was, like, if I remember correctly, um, you guys in Infinite Canvas really started building with games on, on Roblox.
You did some testing with UEFN, but, you know, as. Our listeners can probably tell from this conversation like where your focus has been on lately has been like all of these other platforms and so I'm curious, like, is that a result of there being something about the major UGC platforms that everyone's talking about today that is just not as interesting to you or to Red Ocean now, or is there, , just something unique about these other areas that you just like, A lot more and feel like you can unlock something bigger than on these other platforms.
Sebastian: For sure. We're, we're still fans of Fortnite and Roblox, and we still have a library of games on both. What I will say is I think a lot about my edge, my team's edge, like what are we good at, sufficiently good at, and What table is already picking where does our technological or our abilities actually benefit folks?
I will say when it comes to Roblox and fortnight the headwinds there have always been that We're a group of 30 somethings Living in you know, New York, right like or New York and all over the u. s We have been in games and been in tech and work for different tech companies for like, you know, a dozen, a dozen years.
What edge do we have often against 20, 000 Roblox, 15 year olds for making games are towards their tastes. Right. For the first couple of years, we had a huge edge. We had a huge edge because like, we just understood how games Work. We understood how it's like rip out live ops. We're doing a lot more publishing.
We're like working with game developers all the time that edge started to diminish. And then a lot of people went towards more sponsorship based revenue generations. They went towards the idea of like partnering with different brands and whatnot, and I can tell you from experience, having done this in these boards, that that's not a venture case, right?
It's not a venture case that is going to generate venture returns. Is it a good business for sure? It's a good business. I don't doubt it's a good business, but it's something that, you know, we as a team have always been heads down on, Hey, what's our goal. We've raised venture capital because we want to be a venture case of a game studio publisher.
And so that's why we make the decisions we do. This reminds me a lot of a conversation we were having earlier about the difference between. Wukong black myth and mobile games. Wukong black myth is a great game, but the number of people I see used as an example of how AAA game development is changing, miss the forest for the trees, which is to say that Wukong is a 40 to 60, depending on region dollar box game.
It's sold its copies. If it never sells another copy, it's still a successful game. If any of our mobile games. Launched and a million people play the day one, but no one played a day too. We're a failing game company. Right. And so it's weird to say that as we have all this, like stratification and games are starting to split into different silos that we still compare things like they're apples to apples, when in reality, these businesses are just completely different, a business that's.
Funded and the revenue growth comes purely from, you know, your sponsorships and your brand deals is a very different business than a business that's built on a game that has, 10 million monthly actives that aren't churning at all. And so I think that's really where I see a lot of these things go.
And it's something that I see pretty often and we saw a ton in e sports. And so that's something that like I always have top of mind.
Aaron: Well, final question for you. There's only a couple of months left in the year, but Q4 always has a bunch of big game releases. Holidays are coming, which gives us some time to hopefully play more games.
Uh, what are you looking forward to playing most for the rest of the year?
Sebastian: Oh man, I I'm hoping to, so it's, it's a fun question because. I've been playing a bunch of deadlock recently, the new game, which is completely crazy. I have no idea how it was made. Like who would make a game this way? Cause of course valve did.
So I've been having a ton of fun of deadlock. I actually have a meeting to pick up the Pokemon trading card game. That's on my list of things. I've heard they've made some really interesting adjustments to how that game is played. So I'm curious to see what's happening there. And then I have like a back catalog of just, you know, I'm going to try out.
So, library of Runea is something that I've heard good things about out of Korea that I've been meaning to check out. , I was just playing, , Backpack Hero for first time since it launched to see how the game had been updated. , I'm excited to play some more. , what's it called? , Oh, I call it crack simulator, but, Bellatro on mobile at some point. Yeah, I mean, Between that and some poker and spending time with family, I think I have a nice full library to play.
Aaron: That's awesome. Well, let's go ahead and wrap up here. Seb, it's always super fun to talk to you. I always learn a bunch. I'm sure our audience will as well. So thank you as always for hopping on, man.
Sebastian: Thanks for having me, man. We'll talk soon.
Aaron: Sounds good. And to all of our listeners, thank you as always, for spending your time with us and we'll catch you next time.
If you enjoyed today's episode, whether on YouTube or your favorite podcast app, make sure to like, subscribe, comment, or give a five-star review. And if you wanna reach out or provide feedback, shoot us a note at [email protected] or find us on Twitter and LinkedIn. Plus, if you wanna learn more about what Naavik has to offer, make sure to check out our website www.naavik.co there. You can sign up for the number one games industry newsletter, Naavik Digest, or contact us to learn about our wide-ranging consulting and advisory services.
Again, that is www.naavik.co. Thanks for listening and we'll catch you in the next episode.