Why Most Game Engines Die, and How AI Builds "Living Worlds"

Andrew, you've spent 15 years at Havok, a decade at Unity. Havok solved physics, Unity solved accessibility. What's the unsolved problem in game development that's fundamental as those two were in their era?
Great question. Well, um lots of ways to look at that. You know, I think one of the big observations the last, if you put those together, last 25 years is we were riding a a one of the ways we were riding was hardware, right? So, you know, Havok was all about uh big opportunity that opened up to do physics and make things feel real, make things interactive um using the CPU power that was available and console power and so on. In many ways, the game engine space, Unity and and others, were riding the wave of accessible, you know, GPU hardware, accessible mobile hardware in the case of Unity, of course. Um and I think for for those are the next

15 years, there's an entire realm, a paradigm that's opened up that is um you know, it's both, you know, combination of a new form of compute and and, you know, what we see with machine learning plus, you know, the arrival soon of hardware that's going to be built for that very purpose. It's already built for that purpose. You know, obviously GPUs are are today perfect for those those kind of workloads. So, I think, you know, when you look at sort of what we solved in the last 25 years, I mean, it was I would say hey, we solved, I mean, obviously, you can look at different different sectors of the market, but, you know, pretty pixels was a big one, right? We want to make things look gorgeous. We want games to be, maybe it's photoreal, we want them to be beautifully stylistic, we want them to feel real. Um but, you know, in many ways, something that hasn't changed in gaming is the fact that many of those games are still pre-scripted. You know, you ship that

game like the core loops, the the the world, the stage, the characters are largely predetermined as they go out the door, right? Um so every play through of that game, of course, you know, it'll it'll differ and of course multi-use multiplayer games, you know, it's sort of you know, completely diverse every time you play, but I think the next shift, next big shift in in gaming and and engines and technology and what this is all opening up is this sort of truly dynamic, emergent, personalized experience um that is going to be really enabled by a bunch of new tech um a bunch of new game studios. And I you know, frankly, I think this is sort of opening up things. So I really see this is sort of an additive to the industry. People love the games they of today. You look at the top 10, right? Same games that have been in there for, you know, the last 10, 15 years.

Counter-Strike, Minecraft, and all the classics. Probably not going to change, but I do think we there's a real opportunity for us as an industry to start to think of new types of games. Like what could I do if I had this capability at at runtime, at the you know, in the player experience. Um and I think that's a big that's that's super exciting. I think it's sort of a whole new dawn of um you know, I I love the mission of Unity, the democratized game development. Um I think they definitely achieved that mission and I think now in many ways you could sort of look at what's available in tooling for with with coding, potentially with content creation, a sort of democratization of of building, you know, at that level. Um so I think yeah, it's it's a really exciting time.
kind of leads to your startup. Iconic is building an engine to craft intelligent living worlds. Uh, the history of new game engines with

experienced people though is mostly a graveyard. Uh, when we talk about Stadia, you know, build a custom tech with Google, infinite resources, died a sudden death. Amazon spent at least a billion on Lumberyard with quite abysmal of results. Improbable, I just checked it today, they raised $855 million for metaverse engine, and we haven't heard much about them ever since. And let's not forget about Crytek's collapse or Godot that kind of never really got beyond the starting line. So, graveyard is large, and I think my question is, what makes you think Iconic avoids that list?
Yeah, so a couple of things there. I mean, I think well, first of all, one in say, I mean, hey, someone we nobody gave us a billion dollars, right? That
Yet.
Yeah.
We're a startup, okay? Um, yes, exactly. Um, so, you know, I do think it you know,

it's quite an interesting challenge, I think, back in the day where you had these big incumbents like Unity and Unreal, um, and just someone say, right, we're going to we're going to join that that sort of like for like when you just think of the the adoption that those engines had already, means a massive undertaking. Um, you know, very So, it was ambitious, for sure, right? I think what, you know, what's different about Iconic is a couple of things, right? First of all, I mean, we are absolutely charting, um, just new new seas. This is We're we're not trying to jump into a space that has at least, you know, we know there's a few others in our, you know, smaller companies, startups essentially, trying to do what we're doing um which is you know essentially create like you say a both a tool set and a runtime for intelligent living worlds like so to how to how can I make the like basically bring real-time AI to the

player experience so that mini-game and I sort of opens up a whole new realm of of kind of really interesting sorts of gameplay. We'll get we'll get into that. So I think um one of our one of our sort of you know starting points for all this is it has to it's all about the experience, right? So Iconic actually started as a game studio. Um basically the idea back then we basically sat down and said right well with this is sort of the end of 2023 like with all this amazing tooling that's coming out sort of you know content generation and cloud code and all these cool tools like surely we can build a big game in a lot less time. And actually you know to be fair when you look at the headwinds in the industry like you know content creation of content and giant games is all linear. You just need more people more time to make bigger games. So the sort of question was can we make a big game but in less time with less essentially with a with a small team.

Now you know definitely that was still really challenging to be honest um but the team was was you know really drawn to a couple of like areas that opened up for them. One was one was voice. So they hit on like the ability to be able to to be able to basically be able to talk to the game, right? Um so you're interacting the game through that modality which was new. And then the other thing is well what if the game can talk back to you? So right so you now have like something quite unique in there. And that sort of comes back to when you ask like what about this graveyard like we are we're doing that that's we're doing something new here. So we're we're not trying to come in and basically directly compete with the the big engines here. Um if anything, we're going to stand on their shoulders and, you know, basically provide folks with this case the capabilities that we'll talk about that will integrate into those engines, right? Um look, I've I've had just to

say I've had a first you know, front row seat into what it takes to build engines like that, like the many, many millions of lines of code. It's a it's big. It's such a huge undertaking. And, you know, I think um so, you know, hence why I think we are we are in this place where there's there are there is a graveyard cuz this is super hard. It's just a hard thing to do.
The pitch for Iconic says you create worlds with emotion, agency, and intent. And, I'll be honest, every AI startup in gaming says something similar. So, if we if we strip away the marketing language, the real question that I have is what Iconic actually does today that uh Unreal and Unity cannot. For 15 years, mobile game studios have handed Apple and Google 30 cents for every dollar and call it cost of doing business. AppCharge calls it a choice. Over a billion dollars processed. This isn't a payment plugin bolted onto your live ops. It's a premium web shop your

players actually want to use. And, a direct line between you and the people funding your game. Own the checkout, own the player. Go to appcharge.com.
Yeah. So, just say one thing again cuz I, you know, I know both these technologies super well, they're they're open, right? They're open ecosystem. What Iconic is doing can be done in both of those engines, just to to state that. And, if anything, like we see folks that are are doing similar things to us as validation that, "Hey, we're on to we're on to something here." You know, as soon as you have competition, it's generally a good sign that, you know, you're you're you're kind of striking at something that's that that's that's really interesting. So, um but what, you know, if you strip it all back, right? Um when I was talking about the experience, like as the because the team is essentially made up of game devs, so creative director, you know,

art director, gameplay engineers, um narrative writers, they all know what great looks like from a just the point of view of like what's a great game. So, what we're trying to do is create these intelligent living worlds as I've described. And the best way to best way to sort of visualize this is actually think of a play like theater. So, you typically have you've got your cast, you've got your actors, um you've screenplay or, you know, script, you've got your set, your environment, you've got your story, you've got like what happens in this in this in this narrative in this world. And we're taking those ideas and bringing those into a game, but rather than those be entirely pre-scripted characters, for example, they're more like improv actors. So, I mentioned being able to talk to the game. So, what if you just drop into your favorite game, and instead of like the classic, you know, you walk up to the person behind the bar and they's two

they give you a line and then there's like press A or B, um you can just talk back to them. And those characters, you know, will essentially respond back to you. They have memory. They have a persona. They are fully immersed within the lore of the game. And of course, you know, you're on some quest, some mission to achieve some objective. You still have to do that. But you get to interact with the world in that way. Um so, like how you play the game, how I would play the game, quite different. You know, you might go in you're like, "Hey, come on. Tell me what we've got to do." I might go in and be like, "Hey, um you know, can you tell me more about yourself?" Like what trying to like maybe pull some bit of information out. And that that's what we're when we talk about sort of this idea of intelligent living worlds, want the worlds to feel alive, want you as a player to feel like you have you have this agency to kind of figure out what to do next by by interacting with the world. Um and I

want to say like from my perspective, we're starting with characters. Like we think that is like the the first big kind of big problem set to tackle. And there's a bunch of actually really We'll get onto it. There's a bunch of interesting like efficiencies and production related kind of things that come from that by by basically introducing this kind of technology into into how games basically manage dialogue and how they manage voice and all that kind of stuff as well. So, um so that's that's essentially when you sort of dial it all back, like we're not building a renderer, right? We're not building a physics engine. Um there's already amazing game engines out there do these things. What we're trying to add is this component which brings basically makes makes the games come to life. Um and like I say, the bar there has to be quality. Like the the the interactions, the the gameplay that the outcome of all

that has to be like at a bar that enables people and makes people like them get more and more immersed into the game. Immersion, engagement, like they're the metrics that that we're shooting for. You know, when we when we build our our technology, when we build our demos, and and push things along.
types of gameplay that is very immersive and generative, if you will.
Yeah, definitely. I mean, that word is kind of a bit of a can trigger, you know, generate generative this, that, and the other. But, the um we like to think of it as sort of emergent and dynamic. So, like as as you're basically playing the game, things are progressing and evolving based on exactly how you have the especially with with regard to how you've been interacting with the cast of the game, essentially.
So, story-driven games probably very much so narrative-driven games. I'm almost imagining like PlayStation's portfolio being suited for this quite

nicely.
Absolutely. Yeah, completely. I mean, that's that's where we started. The the title that um that we put out last year, which was the demo of of game that's in production in our studio called the Oversight Bureau, is exactly that. It's sort of a sort of a, you know, puzzle-based voice interaction um title where, you know, you're progressing through kind of these different uh these essentially psychological conditioning uh facility. Don't really know why you're there. You're trying to figure things out. They're trying to figure you out. Um and that, you know, that that kind of game absolutely lends itself. I think RPGs, as well, would be, you know, a a really good candidate for this. You know, a game where you've got big lore and you've got a huge cast um and everyone in this world could have a backstory. Everybody could have a memory of of you as you interact with them and so on and so forth.
Actually, I can I can show some of the other trailers from Oversight Bureau.

So,
That'd be awesome. Yeah.
So,
[clears throat]
very much the sort of a player-led game. So, for those watching the video, Oversight Bureau, they can they can check it out. But, I'm I'm definitely interested in in one part because I'm showing the game that is on Steam from Iconic. And the question kind of leads to like, are you building an engine that you license or are you building games that prove the engine and then license? I know you're pre-revenue as a after the other pivot from game company, but like have you thought about this part?
Heroic Labs has spent over a decade quietly building the back end that powers some of the largest games on the planet right now across every single platform. Their products, Nakama for multiplayer and social, Hero for meta game systems, Satori for live ops and experimentation, and Heroic Cloud to scale the whole thing. The pitch is simple. Stop burning senior engineers on infrastructure problems that were solved years ago.

Build your game, let Heroic handle the rest. Go to heroiclabs.com.
Yeah, absolutely. So, you know, big a big part of our strategy now is like one of the learnings from my time at Huuuge and Unity is like dog fooding is king. You know, if you are your first customer of what you're building, especially in the that customer being a game team and like the the team behind the Oversight Bureau, your results when you're going to ship that technology out to the world are going to be way better. Um you know, if before it leaves the door, people know that this thing can really achieve really great outcomes. Um that's that's key, right? So, the So, in this case um you know, the technology built for the Oversight Bureau has evolved already. I mean, talk about like I will say working in the AI space is quite insane. Like literally every week, you know, there's

we have obviously internal slack and someone's dropping in a paper or something that basically changes our thinking again, right? So, um So, we are absolutely building a you know, essentially a product that we will be able to ship commercial license commercially to other game developers that enables these kind of experiences. So, it's a tool set. So, there'll be an authoring platform for um for writers especially to come in and it's obviously human-driven, it's human-centered, creator-centric. Um there's a runtime component that sits in the game that does all the fact fact the heavy lifting. Um and the way the way to think about it is like one of our thesis in all of this is that on device is king, right? So, getting the full stack running without need any need of a cloud connection is like is the the real goal here. Um

and you know, if you think about that like why? Well, I mean, you know, we all know token the token economy, it's it's crazy, right? So, think of your game becomes super popular and you have an LLM running in the cloud doing some work for your game. Like what if suddenly you've got a million CCU? Um you know, it's uh it's huge. Plus obviously there's privacy, security questions about, you know, data like that going over the wire. Prices, there's also just like drift as models change, you know, if your behavior is pegged to a particular model and that model evolves somewhat, it it you know, you will know like people talk about they notice week to week, right? The evolution of the LLMs and the responses they get back from them. So, um so, our goal there is to get this tech stack running on device. And in the case of the Oversight Bureau, we achieved that. So, the min spec for that is essentially a Steam Deck if you have had one. Steam Deck somewhere in the office here. Um and to do that we had to obviously make some,

yeah, that's constraint, right? So, we had to take some shortcuts there to get it running. Um and what we're working on right now, maybe actually this is something we can interleave is uh something we're we're going to be pushing out soon as a public demo which shows like a much deeper tech stock and much more like essentially a full kind of LLM based runtime for the game running where you get really quite you know right natural responses from from the play you know from the characters in the game. Um the way that we approached the Oversight Bureau was more like at some point we needed to essentially say right this is the conversation space you can be in huge so it feels really natural you can almost say anything to the game but it's still fully deterministic so every voice line was essentially pre-baked into that game as it went out um and that's actually that in some

cases is what devs want right because they just want say look we want absolutely zero ability for anybody to have this game say anything it shouldn't say right. Um cuz that's you know that is one of the the questions that people have when they talk to us is like well you know is it possible that the game would say something and you know we're like that that is just you know that's the number one requirement that we have for everything we're building. Um and it's um yeah I mean it's actually doable and something that we're you know it's one of our sort of key USBs really um around what we're building.
And what you're building like was there a specific market gap that you were exploiting or are you betting that the gap doesn't exist yet?
Yeah I think you know I think you know if you step back and look at sort of this the arrival of this kind of technology there's there's kind of the the pessimists who think people's jobs will go and that you know

a lot of things will be automated away and then I think there's sort of the optimists who think well actually this is going to make your existing job just better. It's going to actually lead to new possibilities. And I think the way we see the tech as it rises to gaming is it's going to be new categories of games. Like like I mentioned earlier, I don't think we're going to you know, you might see hey and I you know, this has been done already, right? People have integrated voice into existing games just as a sort of as an add-on. What we're finding at least is that doesn't work as well as when you really think about the game design from the ground up and think well, if the characters in the game have the ability to remember me, that they can talk back to me, that the the the narrative can evolve as the game as the game plays, what what is you know, how do we actually design the game? So, we think this is think of this really as a new category of game. Kind of like, you know, narrative um

intelligent uh narrative games, basically. Um that that people will will uh will will come to market.
I wanted to kind of go back and you know, go back to your time at Unity, but most importantly your time as as somebody who led AI strategy at Unity. What did What do you think Unity got right about AI that nobody gives them credit for?
Your competitors just tripled their meta spend, launched 12 new creatives in a day, shifted budget from TikTok to YouTube overnight. Would you know? Well, with Pathmatics from Sensor Tower, you can get access to ad spend and impression estimates across all the key platforms from Facebook to Instagram to YouTube to Snapchat to Pinterest to Reddit, everywhere your rivals are buying attention. You see their creatives, their seasonality, their share of voice. That makes you stop guessing what's working in your category and starts planning against what actually is. Pathmatics by Sensor Tower.

Go to sensortower.com and sign up.
Yeah, well, actually it's it's pretty timely. I mean, obviously I'm not I'm not at Unity now. I'm more customer of Unity's. Um and they've just shipped their beta of their Unity AI product. And I think what they got right was an instinct that So, obviously Unity is a is a creation platform, right? First and foremost, it's you know, the the democratization, make it easy to make for anyone to make a game. So, they really lent into that with AI. So, there's, you know, two parts to that. One was just, "Hey, classic kind of assistant, like I'm trying to do X, help me." Um and it has full I think what's really cool about that is it kind of full context into your scene, right? Into your project. So, you know, you can ask it really specific things about this texture or, you know, this game object or why is this not working with that? And that's a huge, um advantage over say

standalone tools that sort of have to maybe inspect in and work a bit different. Um and the other thing that that, you know, was something that was being worked on but is now sort of bearing fruit, um at least as I say, as I as I see it as a as a customer, is the agentic side. I mean, obviously, you know, agentic, this, that, and the other, it's like really been the thing this year, 26, and that's something that I've seen Unity lead into lean into, which is I think really smart. So, you know, if you could imagine having like sort of an expert, you know, uh rigger on on your team that you can just talk to in Unity and say, "Hey, can you help me rig this character? Needs to, you know, has these kind of odd [snorts] properties or I want this or that to happen." I think that's that's uh super a super way to go. Um to really kind of give you the, you know, the fact that you're you become more like an orchestrator, right? A conductor of of lots of game specialties through through

that interface. I think that's really smart as well. Um So, very much but it is very much on the creation side, right? Um whereas, you know, just to contrast it like what we're looking at is really at the run time. It's like, what cool stuff can I do? Can I bring to the player run time?
Looking back, was there something that you feel that you've missed when you left the AI strategy? As you said, things change really fast.
Obviously, I'm biased now. I think there's a ton of really cool, interesting emergent ideas that can come from the run time side, right? I think they're hard problems like Unity used to always pride itself about solving hard prob- hard problems. Um, you know, they're the ones that people are happy to outsource, right? They're the ones that as a game dev, thank goodness someone has that thing working. I don't have to build that myself, you know? Um So, I think I actually I you know, I I do come back to the run time side. Um I think that's something which which is

just we're just at the very tip of. Um and there's just going to be, you know, as the years go by now, more and more really interesting new games that will emerge from kind of putting together what can be done now at at run time with with AI technology.
And when we talk about AI tooling, like most of the AI tooling in game development right now are different sorts of asset generations. You can get faster textures, you can have faster models, you can have faster dialogue, and everything is kind of aimed towards productivity tooling. And that's hardly a paradigm shift that is being sold in the in the media. So, where do you think is the actual paradigm shift if if there is one at the moment?
So, you know, a lot of this boils down to how do I make a really interesting fun gameplay experience that is different every time

[clears throat]
that is personalized around how I want to play the game but is in my favorite world my favorite lore with my favorite cast of of characters right you know I think today you can you can get close to that I mean I mean people love people do a lot of games have replayability and people will go in and try and achieve every single you know trophy etc. but I I think you know personally to really I like I love immersion that's sort of one of the things that got me into the industry in the first place was was Havoc and the fact that you could make this world feel real cuz I could knock over a bunch of crates or you know bump up against a a ragdoll or character and was it would it would react back to me. I think what I'm talking about now is is that kind of next level of depth. Right? So one of the one of the analogies right that we we use is The Truman Show. If you know the movie um and Jim Carrey movie it's and it you know in

particular like that first part of the movie where he doesn't know he's he's going about his life he's making his decisions and little does he know but there's a you know there's an off-screen director who's like okay you know crowd quick run across the road now block his block his path so he like moves he he doesn't take you know he changes direction which leads him to the coffee shop which is where he meets the the next person and he doesn't know this is happening. And I think like that kind of idea where where you you're in a game and you're you're on a quest and you have a mission and you know you basically just have to figure out how to you know obviously how to achieve that doesn't feel like it's on the rails. And it, you know, it really truly feels like I could kind of do anything. I'm going to try this and you know, let's see what happens next. I think that kind of level of immersion um where you're being, you know, when

you're you're talking to the characters around you, you're listing information and in many ways like, you know, our long-term vision is that, you know, I've talked a lot about characters and how they'll react to you. But what if the, you know, the world can evolve around you as well? So, you know, some of the game itself will will adapt based on um the, you know, the environment will adapt based on how you're playing the game as well. So, I think that's that's a whole new you know, paradigm of of possibilities for people, basically, for for for gamers. Both players and open world creators.
Truly open world and what's it, the term we use is um like directed open world
experiences.
The directing being like a really key part because sort of there's a good bit of evidence to say that if I just drop you as a gamer into an open world where you could just wander around and do kind of anything

without a purpose or without some direction actually gets pretty boring pretty quickly. Um so you do need, you need this sense of like purpose. You need, you know, which is games are already good at that, right? So then it's how do I achieve that with with this uh you know, in a game world today? Um and that comes down to like the characters and how they how they react to you and how the game world reacts to you and how it steers you essentially without you really knowing. And that that's kind of that analogy to the the Truman Show.
Coupling that with with your pitch that says intelligent systems that listen, react, and evolve alongside the player. I have to ask, like, are you describing something that ships in 2026 or this is something that ships in you know, 2036? Because as we've seen with a bunch of demos in AI gaming, the actual gap between that demo and the production capable AI is

enormous. Loyalty is the rare user acquisition channel where the audience is actually high quality. But the platform you pick matters more than the channel itself. Vybs is built by game industry veterans. It runs on AI-powered targeting and optimizes for ROAS instead of just impression or specific retention number. It's the fastest-growing loyalty platform in mobile games for a reason. Go to vybs.co and get yourself the best loyalty platform in mobile gaming.
Yeah. No, I agree. And so, look, a couple of things there. Like, one is so, we have to prove it works first. That's why we're building the Oversight Fiero. That's why we're building like the internal um games. So, you know, if it doesn't work for us, it's not going to work for anyone else. It's Yeah, to your point, it is just pure hype at that point. Um there is definitely a hardware thing,

like a hardware kind of tier that um that we're going to have to deal with. So, think if you look on, you know, Steam stats today, um and sort of the the breadth of folks that have a say at 4080, which is a really capable GPU, it can absolutely do what we need what we're asking here. Um it's about 30%. So, you know, there is an addressable market there for this. And it cuz you need the hardware for sure. Um then the other thing that you have to do is you have to have a lively dinner debate with your your art team. Cuz they all they want all the GPU, as they always do, for to make the game look beautiful and and great. So, you know, today if you're doing any kind of AI payload, especially what we're doing, it uses the GPU. So, there is there is sort of a, you know, uh like a a handshake that needs to be made there. But, that's fine. You know, that's definitely achievable based on again, like you got to come at this from like what game are we building? Like what

what do we what experience do we really want players to to have here? Um I think the third thing that needs to be real is the size of model. So, I mean, obviously, you know, most foundation models are getting bigger and bigger. So, getting more capable. Um we want we actually want them to get smaller and more capable. So, we're working with, you know, Google and Nvidia and others um using their small models. And we're seeing actually really impressive results from that. And that's literally something that's changing every month. Um you know, it does require a good bit of work on our side to kind of, you know, essentially fine-tune and in some cases post-train some of those models, but the the model has to be, you know, there's a trade-off. It needs to be small, but it needs to be really good at what we want it to do, which is essentially provide really interesting in-character responses based on what, you know, you as a player

ask of it. Um and that is something which if you look at 26 by sure at the end of 26, this will be able to run our tech stack and it will have a kind of a reach like I mentioned. Um you know, and that's so that is that is in the the land of of absolute possibility here. Um you know, I think my own background, like certainly at Unity, you know, you have a lot of the big selling points of Unity was platform reach. So personally, I'm very interested in how can we get like real reach? So, you know, all devices, XR, mobile, console, PC, obviously web. Um and we're also looking at kind of hybrid solutions there. You know, I think obviously if you just push everything into the cloud, then you don't need a powerful device cuz it just runs in the cloud, but then it has the trade-off there is cost. Cost you as as

a studio to run that service. But I think there is there's a balance there. I think there's a way to just structure this that um we we're looking at that that could be quite interesting give us the breadth that we want. Um and that's, you know, because I know I kind of know the first question from most customers will be, "Great, what what does it run on?" You know, what does it need? So, you know, like I say, there's sort of that spectrum for full on device, you do need it pretty capable machine. And then I think we're going to have, you know, we'll be able to tell a story that goes down to well, it actually can uh you can split that workload um in interesting ways.
that that would use your engine, they need tokens to run?
Well, I mean, somebody has to be running the model exactly somewhere.
does that kind of push them into the free-to-play category because there has to be some sort of a microtransaction in order to be able to run these games?
Yeah, I mean, I think if if it's if you

went fully free-to-play as a business model for your game, for sure, it would you know, it's that does change the economics a little bit because who is paying for the tokens at the end of the day? Um where where does that revenue come from? But yeah, it doesn't rule it out if if maybe it's that model.
was more the other side like the game couldn't probably be a paid game because let's say you sell a game for $29 and there's no microtransactions and you know, you you love the game you got to play it for years you end up being cost center versus yes a person that actually you know brought revenue to the company.
Yeah absolutely. Now that's that's for sure a big concern. I mean look at multiplayer games today it's essentially have this sort of
Yeah they're free to play games games as services that's how they're being called so
Games as services exactly yeah so it definitely you know it it can slide into that category pretty well but as I say like I think the real prize

here is getting it running on device this particular type of cuz one of the things that we care a lot about is you know feeling that you know your data stays with you it stays on device it doesn't leave especially if you're talking to the game right you know it's you you certainly don't want to think that that data might be made available to
you had a bad day and now the game's going to hear all about it.
[laughter]
I mean exactly so yeah so we can we can you know definitely you know I think that's that's an important uh
100% looking at from players perspective you know the best selling games right now they're quite hand authored experiences. So is AI native gaming or game design actually something that players want or is this a bit of like engineers fantasy of what games should be?
Yeah I mean look I think it definitely should be player first you know experience first and like that

is it sort of comes back again to kind of our take on this so the there's for sure I mean you talked about kind of the graveyard and there is a ton of companies who are just thinking purely tech technology first like hey we've got a great you know voice model or we've got a great kind of AI capability surely everybody's want going to want to come going to want this. The classic build and they will come kind of fallacy. Um and you know, we're totally committed to the opposite approach. So, first show that there's a real experience here that is new, unique, novel, engrossing, immersive, and that's what I want to, you know, that's why I want to pick up this game. And to some degree, of course, we got a responsibility disclose that there's AI behind that, but that's not the selling point. Selling point is I built this amazing, you know, narrative murder mystery game or, you know, really quirky, whatever it

is. And it's just so something like I've never I've never had this experience before. I want to play that. My friends told me about it. You know, that's that's sort of our, you know, kind of gold standard there. Um but it's not to say that you're going to see for sure technology being pushed out for technology's sake. Um it's just not what we're doing, basically. And
I think that's an important point because it feels like over the just a couple of decades, we've gone from a place where technology is built to solve challenges like we are old enough or mature enough where we remember like CD players and it was hard to do sports with CD players cuz they would skip and you couldn't listen and then the technology started evolving to MP3 players that were like it was the mini discs before that and the MP3.
Yes.
the technology was solving against and then finally they're like, you know what? Actually, you can just put it on this device that has thousand songs and

then now we have the phones and so forth. So, we're solving against a problem that exist. And then at and then at some point it kind of switched to we are just bullying people to to use technology that we come up with like web three. What? You're not using crypto? What kind of a person do you think you are if you're not, like you're missing out on all this stuff. And and it feels like there's this sort of a thing partly happening with AI as well, where new things come up and you have to ask, is this a problem that somebody has? Or is this just a evaluation play from a venture capital that tries to get, you know, 1,000 X on a on a on their investments?
Look, I think, you know, you've been around long enough to see the patterns and it, you know, you can you can see them, right? And I think I think two things, right? One is like the gamers of all people will vote with their feet. If they don't
[laughter]
If they don't like a thing, they're not going to play it. And also, if they If something's added to their favorite game that does the thing they don't want it,

they will you will know about it within, you know, seconds on on, you know, Twitter and elsewhere. So, I think, you know, that's like the litmus test for us is hey, we put this experience out and actually, you know, we got So, with the with the Oversight Bureau, you know, the the sentiment there was really quite positive, right? Around what we're offering. Um we had to disclose, as you do on Steam, that it was an AI AI was a part of the tech stack. Um but, you know, we were honestly a little worried about that because you could just get review bombed. You say, oh, you know, we not even going to play it, like just immediately, we're going to like down vote that. But, we're kind of glad that people did try it, played it, and realized, hey, this is actually like this is this is cool, like and you know, you can read some of the reviews and whatnot. It's it's different, it's novel, it's a new way to solve a problem. Um and I, you know, to be honest, like way back when when we were bringing physics into the world,

there were similar concerns, right? Um you know, it was there were people would, you know, well, why do I need this? Like what's it really going to bring? This is just technology, you know? And you know, the over time it was really to be honest, Half-Life 2 that that did it where they, you know, did such an amazing job of integrating physics into the title in a way that just made it so much fun, you know, the physics prop puzzles that you had to solve. The people like, oh, okay. This thing is not so bad. And then, you know, people started uh like want that in their game for more than just the very basic stuff. Um you know, I think that's what it's going to take here. Um you know, we're going to we're going to be put out our title, for sure. But there will be others. Um and I think, you know, I genuinely believe there are going to be new and fun experiences as a result. And that's what's That's proof is in the eating at the end of the day, right?

So, um and look, if we're here in a year and you were both looking back and going, wow, man. You might be saying I told you so. I mean, that's just, you know, it it's going to come down to like how as a as an industry we have an opportunity here to do something that hasn't been done before. It games industry loves innovation. It does love trying to push the envelope and bring, you know, kind of merge the creative with the tech, you know, with with bleeding edge tech. Um so, I'm really excited to see what what the industry can do with this.
I'm I'm not looking to say I told you so. Like it would be It would be pretty sad because we know that both most of the startups most of the startups fail. So, it's like it's it's most of the new games fail. It's it's pretty pretty simplistic to say I told you so. I'm I'm more trying to like extract of like what is the actual demand and what is the what is the narrative of the demand? Because like as you mentioned, there's this cautiousness from players and even negative sentiments on different playing outlet against AI. And then we bring in

a whole new engine that is purely interaction with AI. Little bit dystopian, uh you know,
[laughter]
it's like it's so immersive that you kind of lose yourself. Those are positive things, but also you can use it from a different perspective to look at like do we really need people to be even more immersed to gaming experience? Again, philosophical conversation, not part of this, but but overall it's it's I'm trying to understand like what is the real demand from the audience and sometimes we don't know. Sometimes we introduce a product and it creates the demand. But quite often, even when talking about iPods, there was a clear need that that product solved. And you could say it was a totally new product, but that need was unsolved and that's why this new product hit so well.
Mobile, web, CTV, PC, a dozen channels, 100 creatives a week. Now that AI writes them for you and somewhere in that mess, the signal that tells you where to double down and where to cut. Good luck

finding it in a spreadsheet. AppsFlyer gives you growth teams one source of truth across every platform and every channel. So you stop debating whose dashboard is right and start making the call. In an AI-first user acquisition world, the winners aren't the ones with more data. They're the ones who trust theirs. appsflyer.com for the single source of truth.
Yeah. Yeah, and you know, what's interesting is like there's quite a lot of despite maybe to your point the sort of an
air of like nervousness or or kind of general skepticism about AI, the industry is using quite a lot of AI in its production today, right?
For sure with development with software development you can be guaranteed most studios are using some form of code, etc. Um when it comes to you know, the content generation for you know, that's that's you know, you're straying a little more into areas where yeah, you

need to be mindful, um, especially honestly it also comes down to the quality of the output there. Like, you know, it still doesn't quite live up to what you can do, obviously. Um, if you look at monetization, like mobile monetization, like the amount of AI that's running behind the scenes there is insane. Um, so like there's a ton of adoption today already of AI. Where where the gap we sort of maybe we're sort of getting to it here is that, you know, at runtime at in the player experience there just hasn't been as much adoption. Um, you know, sure there's some great navigation AI, like pathfinding. There's been some pretty cool like player behavior kind of, you know, NPC behavior AI. And there's been some great games over the years that have done new things there. But it's that's pretty much where it's where it's stopped. And now there's there's this new capability that is untapped for the

player experience. So so it's it is it's like honestly it's we're breaking new ground and we're we're trying to get into uh, you know, like into the totally into the realm of show don't tell. Like, you play the demo. We're uh, you know, you know, that if you like it then, you know, tell us and that's kind of where, like, you know, we'll build on that. We'll build on what you like and we'll learn from what you don't. And then, you know, we'll we'll make the tech better, um, as a result.
Well, I wanted to finish this up a question about your style and your leadership, maybe your experience because you've scaled a team from from the zero to 50 inside of a rocket ship, like Unity. Um, and now you're
[snorts]
pretty much almost at zero building Iconic. Not quite at zero, but still it's it's the first steps of this company. So, what do you know about building teams now that you were completely wrong about in your previous

position?
[sighs and gasps]
Yeah, it's a good question. I think um the way I'd look at that today is just that things have changed for sure, right? Um you know, I was the first product hire into Unity and like it was definitely at a time where product management was still sort of traditional discipline, you know, go out, understand the market, understand your customers, synthesize that, provide kind of a strategy to to the engineering teams. And, you know, today, like honestly, you can do a lot of that work. A lot of that can be automated. A lot of it can be can be driven. Um you know, and I think as Unity evolved, at least my perception of it was, you know, there were some quite traditional roles that scaled into today, though. It's kind of like, you know, everyone's a builder, basically. You know, everyone can be equipped with kind of all the tools they need to sort of skip the bit about like, "What is product management? Can I be a product manager?" Like everyone can pretty much wear part of that hat. Um

and that's amazing. So, like we have some, for example, in in Iconic, like I've been like really happy to see that our engineers kind of I'd see them almost as product engineers. Like they they have enough tools and data um because we we try and make all the data in the company accessible to everybody at all times to to sort of be able to mine that and make decisions based on obviously data, but without the need of like, "Okay, now I have to go through some product hierarchy, you know, to find a product manager to give me the answer." I can just get that. Um so, I think it's less about maybe what we got wrong back then, it's just more how we're adapting today to like the sort of the new landscape. Um so, things are super flat, obviously. That's one thing advantage we we we can play. Um we have, I think, quite clear specialties, though. Like, we do have I talked a bit about game team. Um, we also have an AI research team cuz

I do believe a lot in still in like this kind of live cycle where our research team are really trying to figure out like kind of what's possible. They're living in the world of of kind of papers and very much the the cutting-edge research. And the game team are like they'll see what they do and go, "Okay, we could probably use that." And this is how we use it like in some game player experience. So, there's a a really nice cycle and obviously there's a tech team in the middle of that who kind of gets it running on device and gets it running in a performant way. So, outside of those kind of three three you know, groups of of verticals, everyone's basically, you know, it's it's really just a big team of builders. Um, so I think that's kind of the probably the the bigger change uh in you know, that I'm making today is when someone joins, you know, we're we're sort of equipping them with all the tools, aligning them broadly with one of those like outcomes. Say, am I building

the game or am I doing research or am am I am I building tech? And, you know, it's you know, there's very little um structure outside of that, right? So, you just jump in, you get get your MacBook, tell us when you got your first PR ready to go and uh you know, uh away we go. So, it's um it's great, actually. It's uh you know, the the onboarding is super quick these days. You know, when someone joins, they can just with all the data that's available and, you know, be able to just ask, essentially, "How does this, you know, how do they do this? How do they do that?" It's all there at their fingertips. It's pretty it's pretty impressive, actually, how how quickly people can spin up and and become part of the team.
that part has has definitely improved and And assume your company is remote, right?
We're We're remote and we do have um a sort of hybrid, so we have an office in London where we're based um hiring, by the way, at the moment. Um and uh

we So, the team goes in, you know, at least 3 days a week to be together and spend, you know, do all the human stuff, the brainstorming around the whiteboards and all that. Um and a couple of days at home, uh you know, and and that works That still works really well. You know, I think like for culture and for, you know, I don't know, but but you but certainly, you know, you kind of have to be together to really build culture, you know,
Yeah.
physically together occasionally just to really form the bonds and, you know, have lunch break break bread together, as they say.
either remote or on-site cultures work. I find hybrid the weakest. Cuz it's like neither.
Yeah, I mean, this is hybrid It's I would agree if you can't get people to get like we get when I say hybrid, folks are a two-hour ride away from the office. It's more just that they have the flexibility to be, you know, to operate uh to do whatever they need to do. Um

but um you know, I So, I would say we're biasing more towards, you know, the the in-person um just to get, you know, like just to build that that young company culture.
significantly with being fully remote and being very successful, but they've been fully remote from the beginning and quite often uh they're ho- homogeneous in terms of their culture. So, let's say they would be an engineering led organization where everybody's pretty much an engineer and people just really want remote culture and they make it work and the company has become successful. But where it doesn't work is where the leadership is kind of wants everybody on-site or kind of wants different like there's a different needs and there's indecisiveness and they choose the hybrid and it's kind of well, it's the less of all evils and we can hire in many places, but yes, it's hard. People are efficient on their own, but inefficient towards company goals, etc. So, it's a it's we don't have to go into it. It's a different topic. It's it's very the type of a topic that always stays

because it feels like startups are forced to choose a model versus scale-ups can choose a model that they want to work with because they have so much power over employees.
Yes. Yes. Yeah, absolutely. Well, we I will say it's you know, one thing given the diversity of of backgrounds that we have from like sort of the veteran game devs to the AI researchers and you know, sometimes the the sort of classic over lunch conversations are kind of where some of the gold comes from. So, which we do try to enable that. That's probably the key. Like we just try and provide that place, but but yeah, I I agree. It you know, I think while we're small, we we're we're benefiting from from some of that human contact, but we'll see where we go from here.
Well, Andrew, it was a pleasure having you on the podcast talk about um building an engine, which is very very difficult. Uh yet you have all the backgrounds to succeed in that. A very

impressive background. So, um would look forward to talking to you in a about 12 months to see where you're at.
We'll see where we get to.
see some demos.
Yeah. Yeah.
On the immersive gameplay. From Iconic.
Yeah. No, I look forward to that. Definitely. Well, thanks so much for having me on today and it's been a real pleasure.
Check out Iconic for open job positions as you said.
Yes, on our site. Love to hear from you.
you so much and thank you everybody for listening.
Great. Thanks all. Bye-bye.
You did it. You made it to the end of the episode. As a fan of the show, it would help us out if you'd subscribe and leave us a review on the podcast [music] service of your choice. Those stars and likes, while a small action on your part, mean a lot to us and help to grow the show. More importantly, are you a member of the Deconstructor of Fun Slack group? If you have five years or more of games industry experience, go to deconstructoroffun.com/slack [music] and apply to join.

Join the games industry's best professional community filled with peers always willing to lend a hand. Or, [music] subscribe to our newsletter to get all the latest insights from the Deconstructor of Fun content creators. Thanks for listening. [music]