Episode 101: Building Trustworthy Agentic AI at Aarda – Constantin Berthelier

But where we differ really is that we don't do things at character level. This is also a real problem I saw in every person I talked to and even studios who didn't want to embrace AI for their video games projects is that they thought it was just one character level chatbot. Okay, you have one chatbot. You talk to them. You enjoy it. Okay, they remember what you said the day before. That's nice. But it doesn't go beyond that.

Hi, my name is Demetri Panici and I'm a content creator, agency owner, and AI enthusiast. You're listening to the AI Agents podcast brought to you by Jotform and featuring our very own CEO and founder Aytekin Tank. This is the show where artificial intelligence meets innovation, productivity, and the tools shaping the future of work. Enjoy the show.

Hello and welcome back to another episode of the AI Agents Podcast. In this episode, we have Constantine, the COO of Arta AI. How you doing, Constantine?

Hey, how you doing? Thanks for having me. Super excited.

Yeah, excited to have you on. So, just to kind of get things started, tell us a little bit about what got you into the world of AI and a little bit of what you're doing at Arta.

Yeah, so I've been what we call a narrative designer and more of a writer and storyteller in the video game space for 12 years. I did pretty much everything. I did social games on Facebook, mobile, some PC and console games. So yeah, I have a strong background in telling stories in video games and yeah, I had my Chat GPT moment three years ago now, right? Time flies. It's crazy.

I know, right? It's been crazy. Yeah, I had my Chat GPT moment just like everybody. I just tried it and I was so amazed by it. I realized at that time I was working on a very big RPG project and I was given a task to give unique voices to thousands of characters and I was figuring out systems and a smarter way to do it. But yeah, when I tried Chat GPT I had this enlightening moment of realizing okay this is the way it's done from here on out. So it was a really strong moment revealing like a revelation. It's so this is how it started with AI basically.

Yeah, it was how everything got into gear. Yeah, definitely. Yeah. No, that's awesome. So what are you doing over there at Arta AI?

Continuing on that, I had been exploring AI for storytelling capabilities for games mostly because this is what I've been doing for the past decade and I've been exploring lots of tools. I was super lucky to work with a company in the US called Inorld and I've learned so much from them. But then for some reasons the project got paused I would say and I was still given the opportunity to pursue my research on exploring tools, exploring the AI, the new things we could do for telling stories in video games using AI and Arta AI is the finalization, I would say, the realization of this initiative that survived through this project.

Basically what I try to do is to embrace all the firepower we get from generative AI and to try and develop the right tool set to tell stories for games and not just games actually living worlds and try to channel all this power we get from Gen AI and to give the proper tools and the right tools and efficient tools to the people. Yeah, this is what we're doing. Very cool.

Yeah. And tell me how specifically you found an interest in this avenue.

Well, it's all of a sudden, you know, if I were to describe my job pre-AI, everything is written by hand. Everything is scripted and even games that still come out today are scripted that way. Everything is written by a person. So even a game with a super large scope AAA games like The Witcher 3 for example has so many narrative ramifications but it's all written by people and everything is scripted even though it can be very rich, super rich and unexpected it's still everything is scripted. So there is one way to play one game right even if you have many branching paths it still comes up to just one way to play one game.

What changes? With AI everything changes. This whole paradigm has shifted. All of a sudden, we have systems that are large enough in which the context window is so open that it allows us to play with so many unexpected outcomes. And all of a sudden you have also what is pleasing me the most when I explore AI is just the ability to sit down and have a natural conversation with a character is amazing and I really want to dedicate my work to making it the best way possible. So yeah, two things, channel the power and to make everyone able to build whole worlds, not just one single chatbot, but also to make conversation natural, perfect and with conversation back and forth and something players can remember. Yeah.

And I guess in general, what in the beginning got you into games or like games of this variety?

A lot of different games. You know, everyone remembers the Telltale series. Absolutely. If you remember. Of course. Yeah. Telltale was very good at making you think you had a freedom, but actually you hadn't. They were very good at it because if you looked at how structure for stories for The Walking Dead for example, maybe the most famous Telltale games they ever released, the branching was not so rich. You had three or four choices and you would always come back to the same node. It would always converge to the very one plot point story beat. So they were very good at it but you have very little freedom after all. But the illusion was really well made.

But this time it's not an illusion anymore. It's real. You do say whatever you like to a character first. That's very new. Their answers are very unpredictable. The system, the AI itself is very strange. It's the first time we play with a technology that we cannot fully control. This is what changes everything. Also in my line of work that changes everything because before AI we used to control every single thing, every single word, every single story beat, every single script. This time it's very different. We don't control everything anymore. This is what changes.

Yeah. No, absolutely. I always find what I found amazing about those games specifically is like the way that you can get immersed because whoever put those together like I think had pretty good thought obviously like you remember the Telltale series for The Walking Dead really fun, really great they managed to stretch that out. How is this going to help people tangibly make games like that that are innovative and you know I think obviously everybody loves AI for what it does and from an imaginative standpoint how does it help in the creative process for that? That's the first question I have and then I have a follow-up about execution and how it helps in execution of making games.

That's a huge one. So first it changes so many things that it's overwhelming for the whole industry. I would having worked with a very install company like Inorld in the US who is still working with major studios, they're a bit secretive about it but they have shared some demos at plus GDC for example in San Francisco and I remember one of their head product saying in a recent interview we're still trying to figure it out and that's the whole thing. The whole industry is trying to figure out how the hell are we going to make it work and this is what we're trying to fix with Arta because the AI is so powerful it's just like water it goes everywhere so you need to build the proper pipes structure to make it flow more or less where you want it to go.

From an industry standpoint it changes so many things that it's very overwhelming and I mean it's like we said in the beginning Chat GPT was three years ago now. Do you know any major game that has AI talking characters that we released recently? Not a single one. So the adoption is very slow in the industry because there is no lead tool to lead how it's made. So this is also what we're trying to provide the industry. We think we have the best tooling set to drive a story the way you want with Arta and there's so many questions also around pricing because if you have inference happening in a video game well who pays for that you need to weave a monetization system around it. It's very messy still. It's a lot of questions.

Then of course you have questions about PR because everyone has been wondering when you own a huge IP you have lots of responsibilities. IP owners will have tons of questions around okay what if the AI goes off and goes crazy and say things political harmful anything harmful there are still some fears and doubt so even though the whole games industry has always been ahead of embracing new tech and kind of has been a very strong leader now they're sort of running behind and there's a lot of questioning and wondering which is very strange. Very strange I would assume three years ago that we would have AI games everywhere now we don't at all which is very present but yeah it's a bit different on the project pipelines we now have studies saying that these studies are bringing proof that majority of production pipelines have AI tools in their chain of production. So there is progress but on the story front we have nothing.

I mean when it comes to player facing experience we have very little, very minor, very anecdotic even though I would say nice. Would you can speak to how you guys differ from other people in the industry doing what you're doing?

Yeah, absolutely. I mean, if you look at first order, if you sign up, it looks like a chatbot app, one of those character.ai apps and many looks, but where we differ really is that we don't do things at character level. This is also something a real problem I saw in every person I talked to and even studios who didn't want to embrace AI for their video games projects is that they thought it was just a one character level chatbot. Okay, you have one chatbot. You talk to them. You enjoy it. Okay, they remember what you said the other day before. That's nice. But it doesn't go beyond that. So you only get the tip of the iceberg.

We often use this image of the iceberg but in a way that we have flipped it. We're beginning with the end. We are before anything else we are a world building platform. What do we do with world building? And why world building? Not because it's a cool hobby and because it sounds fancy to storytellers professionals or not. Actually it's because AI in itself is very hungry and starving for context. It needs data. And if you want a good outcome, if you want a good player facing experience, you need to feed a lot of information about your lore, your story logic, your characters, deep emotions, and you need to feed that LLMs basically, so they can perform in a good way. And if you don't, it's going to improvise. And improvisation is not good because it's not what you want.

So this is really what we're trying to fix and this is how Arta is very different from any other tooling we saw is that first thing you need to do is world build and you need to do that in a very big way and all the app is designed in a way in which we encourage any user to first feed their story with lots of information but we also have AI tools to help you do that if you are drawing a blank or if you're not so creative which is fine you just have an idea or beginning of an idea here. We help you make it blossom.

That's very cool. And what are some of the examples of the types of games that you've helped create? You mean AI games? Games carrying AI in it?

Yeah. Very little. There is a very horrible attempt at making something really great, which is a game called Retail Mage. This studio went full on trying to deliver a relatively and they have it's a game that is on Steam. It's very high quality. There is unexpected gameplay that is fully AI driven and definitely this game is not hyped enough really. It's the early stages which is I mean so there could be more noise about those games because they really try to deliver something that is truly trying to break the covenant of not having AI in games other than just gimmicks.

So yeah, definitely. One thing very quickly I could mention is also there's a video on YouTube. Maybe this is the craziest thing I saw. It's not a game. It's a guy on YouTube. He plays Resident Evil. He lets Chat GPT play it. He asks Chat GPT, I'm in this room. I have that many rounds in my pistol. There are two monsters. What do I do? And Chat GPT decides for him. He just executes with the controller. It's one of the most amazing things I've seen. It beats everything. It beats every AI research for video game making basically.

So you can see there's not a lot of games out there. Inorld also released a game called Origins when they announced their tool set. But yeah, it was the very early stages of having AI and PCs in games were really old stages. I would say Retail Mage is much more advanced.

Very cool. What would you say is something that you're most excited about bringing forward with your product and capabilities here in the next couple months?

Yeah. So, two things. We still talk with video game studios on a daily basis. Some studios are more curious than others. There's a bit of fear of missing out aspect. They want to make sure they are on the hype train. So they are curious and we explained it to them that we can have our tech in their games. I hope I can make some very cool announcements pretty soon. But there are quite a few of them to be very honest. Most of the industry is very slow at adopting those tools.

So what we do is that we want to release this platform that we called A Play and that is for everybody because what we see and there are some good numbers and studies proving it is that who is embracing those tools it's the normies, the everyday people. You see if you go on X or on LinkedIn you see more and more of these solo dev fully AI generated games that do reach millions of players. You do get that more and more.

So normies are absolutely embracing those tools and they just create for pleasure. They don't create to make money or start a business. They just want to put their game out there because they think it's cool and if they find some audience for it, well, they're happy. This is how it goes. This is the kind of audience we're trying to appeal to builders and all that play. The platform is really for everybody. I'm really convinced myself that everyone has a story in their minds everyone. But just by listening a piece of music or you walk out of a theater, you just saw a movie that you like and you thought, wouldn't it be cool if it was somehow the same story but something different happened?

I want to give people with this very natural ideas and thinking to be able to just sit down and in 20 minutes they will have a playable experience that they can share with their friends and this is exactly what we're doing and hopefully by mid-February we'll have something that makes everyone capable to do exactly that.

Yeah. Absolutely. I'm kind of curious, you know, taking a look at what's going on on your website, what kind of variety of different types of themes and games you are mainly seeing here because I see it's kind of like a wide variety and what is necessarily your favorite when it comes to stylistically what can be created and also just practically how does it work to create these worlds?

Yeah. So basically what you're going to do is just sign up and if you want to build a world you will go to build and you will have to build the ability to create from just from blank. If you are familiar with storytelling and you already wrote some stories that's good it should be for you right but if it's the first time you ever build a story and you never wrote anything we will just ask you a few questions about what this story is about. Is it the real world? Is it fictional? Is it happening today? Did it happen 300 years ago? What age is this basically?

And then we will ask okay what is the general tone of your story? Is it humoristic? Is it sarcastic even? Is it really greedy or very realistic psychological? And just by asking questions about six, seven steps we will have a story figured out. Then you will just maybe with just one sentence tell what it is about like what is it about? Because it all comes down to this. If you tell someone you have a story well yeah what is it about? This is the only one question you need to answer and just from that we will pull the strings and make your idea blossom.

Then we will suggest some characters if you don't have any characters in mind if you don't really know who you want to play as who you want to talk to so we will make suggestions on the flies based on your previous choices and you will click some buttons and you will have a playable world and game and we will illustrate them with AI generated content images and it's going to be playable and explorable.

So you will have under 5 minutes the ability to talk to your characters first explore and adjust of course you will have your knowledge base also that is already woven and you will have the ability to edit anything you don't like and just by editing you will make this world your own and that's it. You're an author. You just made a story. That's pretty cool.

And then I guess like obviously you're talking knowledge base and that kind of stuff and you're saying under five minutes you can talk to your characters. I mean what kind of timeline are we talking about about making like a full-fledged game that you could bring to production? What does that kind of look like? And compare that to what it looks like for games like this in general without AI.

Yeah. No, that's a very good question. So we know and I have one example I always use as a reference is Roblox. Roblox is just 8% of builders. So people who actually make games and the rest are just players. So when you hear Roblox, you know that it's just a minority of very dedicated, strongly dedicated people who learn how to use Roblox because it's not that easy and who actually make games and take the proper time to make quality games. So, we know it's going to be a minority of also of builders.

If you want to build a very good story with nice illustrations to make it with a good pacing and good story development, even though we have tools to help you do that using AI or not, actually, it's going to take a while. It could take you like a whole week, two weeks to complete a very strong well fleshed out story but it should pay off. If you bring quality it will attract some attention definitely and we do provide every tool to do that.

If you want to do quality and a long feature story with development emotional evolution for your characters for example you can absolutely do that but yeah it's going to take some effort which always brings back to okay AI does a lot of things but it's not taking off the effort if you want to make quality and make something that matters and that people will remember you need to work.

No, absolutely. You do need to work. It's a big portion of something that people I think forget often times when they are getting into this kind of stuff. So what would you say to a person who's kind of in that early stage of idea for developing a game? How would you go about utilizing your tool to quickly and effectively get ideas running to try to see whether they have an actual tangible idea here with?

Absolutely. This is one argument we have with studios because as I said the adoption to have our tech running in a game engine to go in production and facing players is not going to happen now. It's going to take some time. We know that. But we do showcase Arta for exactly what you said. We tell game studios and builders you have announced the game. You need to feed your audience and your community with content and you need to provide them with the feel of the story or even the gameplay that you want to sell them. You should use Arta to make a game in less than one week you will have something of very high quality and you can definitely prototype explore ideas, explore concepts very quickly and for a very cheap price definitely.

So this is definitely something that we use and people like that when we are in meetings with execs for us studio execs this is something that they can it's not a prototype because proper prototype in game engine is a different story and it's very time consuming but yeah they can just try off ideas send it to their community see how they react if they love a vibe or something of a story or if they don't like it they can do it very fast and for very little money.

Very cool. I'm curious obviously you mentioned the amount of effort, amount of time and money as well. Would you say that with something like this, you really have the opportunity here to help out those who maybe have the idea for the game but don't really know how to maybe get started or don't have the funding to kind of get something up and running the ability to kind of test drive?

Absolutely. This is my mantra. Look, I mean it's 2025. If I asked you, can you make a video game about your vacation with your family last summer? Imagine your family asks you that, hey, why don't you make a video game based on our vacation last summer. It would take you ages. You would explore Unity. Okay, if you don't know about Unity, it's going to require weeks and weeks of learning and then you will have to deal with 3D and code. Unity, okay, out of the way. Roblox. Roblox is very complicated. You think it's easy because it looks like Lego basically. It's not. You need to learn Lua. The console is very lots of heavy lifting.

And then you have character.ai, but it's not sufficient. It's just one character. It's a quick chatbot that you make in two minutes. It's very enjoyable. I love that. But you have nothing in between, which is crazy. And this is exactly what we provide. It's a platform for normies. I'm a normie. I'm a super normie. Actually, I define myself as a super normie because I think even though I'm a professional from the space, I kept my normie thinking intact. And this all is exactly for the sort of people who want to be able to build a playable experience that looks and feels nice and that they can do overnight for practically nothing. This is exactly what we do.

So I say to everybody, finally you can make a playable game that does feel like a true story also because that's important because you can vibe code a brick game on Gemini or Chess, that's great. But if you want to go deeper and tell something, if you have a vibe, not even a story, but if you just have a vibe, you should come to Arta. It's going to be just for you.

Absolutely. And tell me a little bit more about your experience with like what the different companies like Gemini or whatever are lacking when it comes to vibe coding. So all the tools, games specifically obviously.

Yeah. I mean, you mean with the AI tool set? You mean what is missing?

Yeah. Like what is it missing in regards to getting a game started? Because my point being, right, a lot of people are talking about these multi-purpose tools and it obviously seems like you guys have a very specific thing that you're able to do. And I'm just trying to highlight, hey, this company has the ability to specifically do something different because honestly, I feel like the main difficulty when articulating more niche products like yourself is like, okay, can I just use ChatGPT or Gemini? And I'm trying to sus out the main reasons how it kind of differs from their baseline capabilities.

Mhm. Well, if you go to any mainstream AI tooling like Gemini or ChatGPT, yes, you can craft a playable story, I would say, but at some point one, it's going to lose track of your story logic because it doesn't store all the world building I've been mentioning before, it doesn't do that. So it will require you some skills to do that. If you want to do that, that's going to be hard. Second, if you want to add visuals, character sheets, make groups of people, and this group of people is opinionated against this other group of people, it's going to be a bit complicated even for a tool as powerful as ChatGPT or Gemini to keep track. That's going to be very difficult.

So what you can do is with those tools is improvise a story and enjoy it. That's very enjoyable. I did that a lot. But if you want to have strong guard rails and safeguard your story logic and make sure your world building is always addressed at the right place at the right moment, Arta is just for this is where we differ a lot and I think we are the only ones to do that. And so it's pretty comparable to all those video platforms I would say but for playable story-driven game experiences definitely.

Very cool. What would you say your take is on in general how this is going to not just your product but all products in the gaming development industry and in general how's it going to impact jobs all that kind of stuff?

I don't really have quite a footing on where that's going to go. That's the question right. Look it's clear that the whole production chain is impacted. Things will change definitely. I think the market will downsize. I mean the industry will downsize. People will still enjoy video games even more, I think. But I think it's going to play out a bit like YouTube and television. Television still exists. It's way less powerful than it used to be. YouTube is YouTube, right? I don't have to tell you what it is.

I think it's going to be that way. We already see that random people coming up with good games and they get attention and they have overnight success. It lasts for one week and then people switch off and they go to something else. So I think definitely things will change. Yes, people will lose their jobs. It's a reality. No need to sugarcoat it. And maybe lots because video game making is very tech-driven. And it seems that AI is addressing all the jobs basically whether it's 3D rigging even something as horrible and tedious as rigging is doable with AI basically today which is very hard reality.

As I always say maybe in two years from now you will have the ability to make a PS3 level game just by yourself based on your vacation with your family from last year. You will have that power. I think if three years from now you will have this ability to develop something that big that will look like a PS3 level game. But at the end of the day, we already see that kind of it's just needs to happen. People will decide what tool is going to survive, what tool won't survive. But at the end of the day, the only question that will remain is one, what is the game about? What is the gameplay loop? Is it a shooter? Is it an RPG? Is it a visual novel? And second is what is it telling? Is it a love story? Is it a conflict between two groups of people? Is it a story of just one guy trying to get back on track in his life? I don't know. You tell me. And if you don't, it all comes down to this really.

So I hate to say it, but maybe one job will survive is storytelling because it's what we'll always be needed. I mean, look at AI video. Yes, you have Hollywood at home now almost. But same question remains. Yes. Okay. You have Hollywood at home. What are you going to do with it? Are you going to make Interstellar yourself or will you just make romcoms? What are you going to do with it?

I had this very strong and overwhelming feeling with digital cameras when digital cameras became affordable. I don't know, maybe you're much younger than I am, but it was a huge thing when I was in high school. DSLRs versus real SLRs. You could buy a camera or your dad could buy it for you or it was not such a big deal. All of a sudden, you could make movies, right? But now it's that again, but very different. You have all Hollywood quality at home. But for the industry, yes, I think many things will change and I think the politics of the industry will play around who owns what IP. Definitely, you will always have big corporations owning very strong IPs that they will capitalize on. So those games will never disappear. We will always have Mario games, Final Fantasy games, Assassin's Creed games. We will always have them. It will not disappear at all. But indie space is going to become more made by amateurs, which is great. I like that. I like the fact that whoever they are on this planet can do something of quality and that they can find their audience and make people happy with it. I think it's great.

Very cool. Well, no, that's good insight. I've always felt like brand names for games are going to stick, but I have wondered kind of how it's going to impact things because even some brand names are starting to get old at this point in some ways. I don't think they just came up with Call of Duty Black Ops 7 or something.

Yeah, something ridiculous. Black Ops 7. I think it's seven. Maybe in America, there's always been this recurring joke, especially in American football. Madden is the game that has been copy and pasted for a while. I'm wondering when someone's going to be able to AI create an alternative, better version of Madden. That will be the day because the way to compete against them is pretty hard right now.

Absolutely. And you mentioned forget Gen AI for a second. Do you realize how complicated the AI of Madden games are? Can you picture the complexity of making the logic, the AI logic of a Madden game? It's crazy. They figured it out in 2005 on ES.

It's wild to me though, and the reason I'm just being tongue in cheek here is it seems like there are some games just resting on their brand loins, right? Like there was a game called ESPN NFL 2K5. It is for some reason still considered the best football game ever really because the way that they managed to have tackling and route running and all these types of things were great and they just basically yeah. It holds up better than even newer Maddens that are coming out. The way that it feels when you're playing it is insane.

Yeah. What do you think is the reason that you see these issues where companies don't quite improve in that way? And is it because of the fact that they just have the hold on the market? Can they improve within a reasonable amount of money? What do you think?

I think it's more simple. They need to release a game every year and they need to change things because if they just copy paste the same game, it's going to show and you see that a lot. So they do need to try other things and sometimes it's less good. It's very simple. I would say having been in the video game productions I would explain it really lack of time also. More games in saying cannons always faster and faster. We need more games. We need a Madden game every year and it needs to have more features because that's how it is. The publisher will say I want more features just find out just add more stuff so we can say tell that at the back of the cover. It's that simple really.

Yeah. So this is how I explained but yeah you see similar reference I have is Dragon Ball Z games on the PS2 supposedly. Those are the best Budokai 3 and Tenkai 3. That's the best game ever made. Absolutely. I agree. It's not up to debate, but you see they made a remake recently and everyone's saying it's not that great. So I would say that it's very simple. More games, less time. People get a bit tired, they just try new things for the sake of it, and they don't always hit.

No, absolutely. Are there any final thoughts you have on the industry or any final comments you want to have about your company before we close things out?

Yeah, of course. I would say as a narrative designer, as a storyteller in the games, I see also a lot of backlash and I'm absolutely comfortable with it. I recently spoke to fellow narrative designers when I tried to engage conversation around AI and storytelling in games. I've been told that they were stronger against it and they even had a commitment against LLMs and I absolutely respect that. I am myself very conservative when it comes to book writing for example. I do write novels and I do that most of the work with pen and paper. So I understand conservatism but I would like to tell my fellow narrative designers and even the whole industry that this time is very different because we can do new things that we couldn't do before and this is what matters.

It's not about going faster and producing more games for less money and it's not about replacing people at all. Even if it does that I mean but storytelling is very different from 2D assets for example which is okay that's painful 2D assets people artists yeah I do feel the pain but us it's very very different. I think it makes our jobs even more noble. All of a sudden, like I said, we need to world build with this sort of tooling. It won't do the story for you. You need to give it so much information before you reach a good level of quality that it does require so much work and skills. So that's my message to also the industry is that let's embrace those tools because we have a lot of work to do basically. So let's do it.

All right. Well, with that being said, everyone, please make sure to head over to arta.ai. That's arta.ai. Thank you so much for listening to this episode of the podcast. And if you did like it, make sure to leave a like, subscribe, and review it on Apple Podcast and Spotify. That goes for you, too, Constantine. Get the reviews up for yourself. Thanks for being here. Bye-bye.