Sign in or Join the community to continue
Inside a Game Where AI Agents Never Sleep
Posted Oct 08, 2026 | Views 5
# AI Agents
# Game Development
# Agentic AI
Share
Speakers
Issac Lee
Founder @ OnchainLabs
Demetrios Brinkmann
Chief Happiness Engineer @ MLOps Community
At the moment Demetrios is immersing himself in Machine Learning by interviewing experts from around the world in the weekly MLOps.community meetups. Demetrios is constantly learning and engaging in new activities to get uncomfortable and learn from his mistakes. He tries to bring creativity into every aspect of his life, whether that be analyzing the best paths forward, overcoming obstacles, or building lego houses with his daughter.
+ Read More
SUMMARY
What happens to an MMORPG when half the players never log off? Issac Lee, who runs AI and blockchain experiments at NEXUS, has been testing that idea. His team is filling game worlds with agent players instead of scripted NPCs, so a new title feels alive from day one.
+ Read More
CONTENT & TRANSCRIPT
Issac Lee: [00:00:00] If you populate the, the, an MMORPG with, um, agent players instead of NPCs, you're able to have a much more, uh, vast and dynamic experiences.
Demetrios: The thing is, and again, this would be difficult, right? They're relentless. So a player has to sleep at some point. An agent doesn't. You can almost have undercover agents that are just testing the game for you to see and make sure that everything is working.
Issac Lee: One of the earliest, um, adoption of Agentic future, I think would be shopping. So agent has, as we said, agent has a lot of context of myself.
Demetrios: My agent usually just tells me to stop smoking weed when that
Issac Lee: happens.
Demetrios: All right, Isaac, you're working at [00:01:00] Nexus. You all are doing a lot of stuff with games. You particularly are working on the AI and blockchain field. I know you have a little bit of a lab going on there. You've got some cool skunk work things happening. Can you break down what you're thinking about when it comes to AI in gaming world?
Issac Lee: I think there's been a lot of experiments across the globe from large studios to very small, um, indie developers using AI to develop games. Um, and one of the key things that I, I was-- I learned at GDC this year as I was presenting was using AI as a companion within existing games to increase the retention of the users.
Issac Lee: Hmm. So really interesting thing that, um, it was a Chinese, uh, it was a subs- I think it was a subsidiary of a Tencent. They presented a lot of metrics on how they enable, um, users to have a playmate [00:02:00] within the game that actually knows how you play. Uh, the AI agent has data access to, to your gaming play, and it's able to kind of help you become a friend in playing the game.
Issac Lee: So that really kind of struck me when-- And that was like s- Five month ago.
Demetrios: Uh-huh.
Issac Lee: About a half year ago, right? And a lot of, um, interesting developments has happened since then. So for us at Nexus, um, we're doing a lot of experiments from A to Z in we're investing into companies that are, uh, building AI platforms to create games, uh, from, from just text to, to actual serviceable games.
Issac Lee: Mm-hmm. We're developing, um, simple tools to help users create games. For example, uh, we're building-- we, I think we recently built a tool called Text to 3D, and d- we give AI a preset [00:03:00] of engines that it could use in order to generate some sort of a physical world, and when you prompt the AI, you, you tell them, "Hey, I wanna, I wanna build a town.
Issac Lee: I wanna build, um, some kind of map that's very similar to some game in Roblox."
Demetrios: Mm-hmm.
Issac Lee: And, um, AI is able to go find the right places where to integrate this engine, where to use this setup, and then it's able to create, like, a map that is, that could actually be serviced in, in, in Roblox. So this was one of the experiments that we're doing.
Issac Lee: We're trying to push down on it to, to, to refine it. And, and I guess a lot of the people c- would agree that AI helps you to bring the completeness of the result of, of a product, say by 80%, 90%. Makes you really- Mm-hmm. It helps you a lot in develop- in getting to that 90%, but after, after that- That's the hard part
Issac Lee: it's a lot of refinement work, right? So the u- the users doesn't really care whether you used AI to [00:04:00] develop A or, uh, right? So, um, it's, we're, we're focusing on that remaining extra 10% so that the product that is built through our tools and products are actually serviceable.
Demetrios: Mm-hmm. So it's not just that 80% done and then You have to work so hard to get that extra 20% that you give up, and you have a graveyard of 80% done things- Mm-hmm
Demetrios: in your repertoire. I personally can re- relate with that one quite a bit because of how many things I've started, and then that extra effort to finish it is just not worth it at the end of the day, or it's- Right ... I don't have the time to hit it. I wanna go back to this idea that you were talking about of using the, using AI as a companion to increase the engagement, and I presume the hard questions in that are making the AI at your level.
Demetrios: Mm-hmm. And it just doesn't go really, [00:05:00] really good, and it, it's not on, like, God mode, and it's also not on really dumb. And so you as a player, wherever you are, the, the level it can kinda meet you at is where it should be. But then also, you can interact with it as if you were interacting with somebody else that is on the other side of that.
Demetrios: And so you can tell it to go and do something, or you can tell it that you're going to go and do something, and that it should wait there, and it will understand those commands.
Issac Lee: Mm-hmm. Few, few things. I guess, um, one of the research that's being done by Krafton, um, the operator of, um, Battlegrounds-
Demetrios: Mm-hmm
Issac Lee: They're actually doing a lot of research in this front.
Issac Lee: So the, the, the, the playmate agent that, that knows what you're doing in the game is able to kind of direct you to say, "Hey, you know what? You're, you're kinda weak when you're doing something. You know, you're, you're [00:06:00] kind of vulnerable. This is, this is sort of, you know, based on your past play history, this is sort of where you kinda die out."
Issac Lee: So maybe you should- So
Demetrios: it's not like another character. It's like a coach that's there-
Issac Lee: Yeah, yeah ...
Demetrios: that's kind of om- omnipresent.
Issac Lee: Yeah, yeah. That, that's one way to think- Yeah ... about things, but if you think about how AI, how AI agent can be part of your, um, part of your play in the games, it really depends on what sort of game you're playing.
Issac Lee: Mm-hmm. For example, um, if you're playing like, you know, um, FPS games, it could kind of help you, direct you, coach you. But if you're paying, if you're playing something like, um, MMORPG, um, one thing that we're actively researching is how to, how to enable a lot of, uh, play, a lot of NPCs as sort of, um, agent players in the game.
Issac Lee: So, um, you know, if we look at the MMORPG landscape, um, if you, if [00:07:00] we look at successful MMORPGs, obviously the game is fun, but when, when we say the game is fun, it retains a lot of, uh, users, right? It retain, because it retains a lot of users, there's, um, competition within the users to spend in the game to get ahead, so it creates that dynamic.
Issac Lee: Um, we're thinking that we might be able to bootstrap that specific phase with, by deploying a lot of agents, agent players in the game so that the human player can come in and, and feel the similar social dynamics they feel in, in, um, natural human-only games- Mm-hmm ... so to speak.
Demetrios: And why is that different, forgive my ignorance, but- Why is that different from just having traditional NPCs?
Demetrios: Why is like infusing them with AI or having them be AI game players versus regular NPC different?
Issac Lee: Right. Regular NPCs, everything's scripted, right? So there's no, [00:08:00] um, spontaneity or, you know, if, if you're-- So for example, my interaction with you would be very different than someone else's interaction with you.
Issac Lee: It's not gonna be the same, right? Time and place is different. You know, the thought that's going through your, your head is gonna be different.
Demetrios: The questions I ask you, yeah,
Issac Lee: of
Demetrios: course.
Issac Lee: Exactly. Um, that's sort of the experience that we think could happen with, um, agent-populated games. So, um, if we look at it, if we look at the development of, or history of, of games, we have these studio-generated content where, whereas, uh, the studio, the developers define everything that you could experience in a, a certain, in, in a setting, and that was it.
Issac Lee: Once you finish the contents, that was it, right? And then the next evolution was user-generated contents, right? You give the users an open world, and then they create this sort of, you know, additional maps, features, um, some sort of [00:09:00] loops, and users were able to generate the content, right? And I think the next, um, next phase of the development is agent-generated content within the game.
Issac Lee: So whereas if you populate the, the, an MMORPG with, um, agent players instead of NPCs, you're able to have a much more, uh, vast and dynamic experiences as you interact within that world.
Demetrios: From day one for you too, as you were saying, you launch the game and it already has the tipping block or the tipping point of players that you would need.
Demetrios: Because a funny thing about This scenario, which is similar to a marketplace, is that you have the chicken and the egg problem. And so you're kind of passing over that chicken and the egg problem if you already have the game filled with agents that can be spontaneous- Mm-hmm ... in those moments.
Issac Lee: Exactly.
Issac Lee: Exactly. That, that, that is what [00:10:00] we think will help the games to enhance their monetization and also to prolong their, um, um, half-life, so to call it. Mm-hmm. For every MMORPG, there's a certain set of, um, as the game matures, um, the players are able to reach greater levels, a lot of inflation happens within the game.
Issac Lee: So th- there's, there's, there's a lifetime value to that game. Yeah. Um, we're thinking that we can increase that as well with use of, um, agents in the game.
Demetrios: And would you give some agents the ability to have, like some agents show up and they're rich, and you can trade with them? Or is it everybody's, all agents are starting from ground zero, and they have to all work?
Issac Lee: I think, I think the way we see it is that they should have the same restrictions of, of, of play. They should have the same starting ground as the players.
Demetrios: Oh, interesting. Yeah. And they also have the same [00:11:00] tests that they have to go through or side quests- Right ... that they can go on, and they can choose if they wanna do that.
Issac Lee: Right. Right.
Demetrios: The thing is, and again, this would be difficult, right? They're relentless, so-
Issac Lee: Mm-hmm ...
Demetrios: a player has to sleep at some point. An agent doesn't.
Issac Lee: Right. But if you think about-- Well, i- if you look at the current landscape of MMORPGs, uh, most of them offer auto sort of, um, farming.
Demetrios: Uh-huh.
Issac Lee: So they're able to go on, um, go kill monsters on, on the auto mode.
Issac Lee: So already, already the players, you know, it's the, the experience for players in MMORPG is that they could pick the segment that they wanna enjoy. Uh-huh. Um, rest of, rest of the time, the player's character is on auto mode- Uh-huh ... doing whatever quests, farms, killing mon- monsters, et cetera. It's already on auto mode, so that's why we thought that maybe introducing [00:12:00] the agent concept wouldn't be as difficult as, as, as building something ground up.
Issac Lee: Yeah. Because from, like, you know, I guess if you think about MMORPG from, like, 10 years ago, you have to spend a lot of time, you have to spend a lot of time actually playing the game.
Demetrios: Yeah.
Issac Lee: Right? Doing, doing- All your
Demetrios: life.
Issac Lee: Yeah, doing all the hassle-related stuff, right? Mm-hmm. But right now, the experience already is at where, um, the player can just cherry-pick the moments that they wanna play.
Demetrios: Uh-huh.
Issac Lee: Um, and they're here to experience the, the, so to speak, uh, dopamine loop within the game. So they're, they're, they're here to pick those, pick the best moments to kinda play at. Um, and I think with the addition of agents, I think we can increase the length of such experience as to also, um, provide the variety of the experience.
Demetrios: So now that's [00:13:00] just one sub-sector of how you could be using AI- Yeah ... within the games. You did mention Prompt to full game-
Issac Lee: Mm-hmm ...
Demetrios: as one potential outcome. But that 10%, trying to make that 10% work is where you feel like you have to put in a lot of effort. How are you doing that now? Can you explain, like, how you're helping make that 10% easier?
Demetrios: Because I have the biggest qualms about the chat-
Issac Lee: Mm-hmm ...
Demetrios: being that language is fuzzy. So a lot of times when we want traditional GUI options like a slider, or we wanna just, like, turn a knob just a little bit, or bring-- I think about histograms when you're correcting photo coloring or something. Mm-hmm.
Demetrios: There's a lot more fine detail and granularity to being able to adjust things-
Issac Lee: Yep ... [00:14:00]
Demetrios: in that level versus if you say, "Okay, make it pink. No, make it more pink." Right? Or- Yeah ... "Make it big, bigger." Yeah. It's not really the same kind of granularity that you can get, uh, big or grandiose. It's, it's also very subjective in a way.
Issac Lee: Right. So I think, um, we're-- I, I think I could give you two examples, right? Um One is, um, so Nexus recently acquired OneStore, and we're expanding into the, um, application marketplace. Nice. Well, actually, we, we already took a hold of application marketplace. And OneStore has a lot of, um, different, uh, businesses, and one of the businesses that it does is it, it, um, distributes content, so you're able to read, uh, comic books, novels within this, this app.
Issac Lee: And using this, um, existing IP, we thought to create, [00:15:00] um, some sort of, um, short web dramas that users are, users can, um, um, watch using AI. And, um, the problem becomes that you're able to... You're-- We, we, we don't develop our own model, so to speak, but we use whatever model is out there, um, to the best of our ability.
Issac Lee: And the, at best, you could generate probably, you know, 20 to 30 seconds of, of a video. Let's say that we wanna build an, uh, we wanna turn a novel into some sort of, um, video series, right? So we need to, we need to create maybe at least hundreds, if not thousands, of these 30-second clips, and they have to... It, it has to be very natural for, for human eyes to, to, to see that the story is actually flowing through, right?
Issac Lee: Mm-hmm. Um, but the, the trouble is that these segments are, [00:16:00] tend to be different. Even though they're generated with similar prompts, it, it, it tends to be different. So, um, what we do is we give very specific instructions on how to generate video, at which point... So I, uh, uh, going back to your question, right?
Issac Lee: In order to refine the work, um, you have to give it very specific, um, instructions. So as, as you said, make it pink, make it more pink. You have to define how pink it is. Yeah. Right? So, um, when we build any sort of, um, AI -induced products, we try to define as much as what AI can do so that it doesn't have to assume much.
Demetrios: Hmm.
Issac Lee: We try to define, say, if we're building a video, there's gotta be some sort of, um, human characters in it. We try to define, we, we, we give instructions so that the AI goes, say, "Okay, the first thing I'm gonna do is that I'm gonna create a video using this [00:17:00] story. Um, so first I'm gonna build a scenario, and within this scenario, um, after building the scenario, I'm gonna, I'm gonna describe how the characters look.
Issac Lee: And in this 30-second segment, the characters are-- they gotta be wearing A, B, and C. This character, when it speaks, it has to be looking at that way. When it interacts, it has to sound like this," et cetera, right? So, um, the way we, we do this is to, we try to generate as, um, we try to create a very high-level framework of what the AI should be doing.
Issac Lee: And then, um, after that, it's a lot of iterations and trying to fix that, the, trying to color in that last missing piece, um, so that when you use this, um, system to generate an output, it, it, it lives up to your expectations. This is kind of similar as to one, one other tool that we built was to, if you create, um, 2D pixel games, [00:18:00] um, using AI to generate a pixel is, is not that difficult.
Issac Lee: But when you play the pixel games, it's, it's, it, it, it's got a lot of motions, right? But when you try to make the lo- motion look very natural to human eyes, it's a lot of effort. Yeah. It's a lot of human effort to make that happen, right? But using AI, um, when we first tried it, it was, it was awful. It was just, it was like you, you, you, you built, we built the 2D character, and it's, it's moving through a lot of different ma- a lot of different, um, settings, and it just doesn't look natural at all.
Issac Lee: So what we did was, um, we laid out what we want from this character and what we hope to achieve, and, um, we created, uh, um, a loop where the agent goes around and round and tries to fill in that, that, that blank, so to speak.
Demetrios: Yeah. So this again is, um [00:19:00] One other way that you can use AI, and I love the fact that you're trying to incorporate it into many different areas within the game design.
Demetrios: One area that we haven't talked to that I feel is probably the most relevant right now is how the folks that are coding are actually using it. Like, are you shipping with all of the agents, your coding agents, you're creating those? Like, walk me through what the engineers are doing.
Issac Lee: Yeah. So, um, we built our own, um, internal development tool called, um, DaVinci.
Demetrios: Nice.
Issac Lee: Yeah. And, um, it's got-- One of our AI engineers, um, he's been, he's been doing a lot of AI work prior to ChatGPT blowing up. Um, and back then he was focusing a lot on doing, um, AI, uh, to use to [00:20:00] analyze and increase the marketing performances. Um, but we've been doing a lot of research on, on that front.
Issac Lee: Um, so we use our own internally built framework to ship product. Um, and within this framework, um, let's say that I'm building, um, say, a note-taker app, right? Um, when I say that I'm building a note take-- Uh, so we, we, we go into this ID and we say, "Hey, I wanna build a note-taker app." When we say that, it, it classifies that, okay, this is, this guy's trying to build an app.
Issac Lee: When-- What, what do we know when we build an app? So it, it's got, it's got some sort of a blueprint on how to build an app. Um, and when that happens, it has multiple different agents that are-- We have back-end agents, front-end agents, um, the, the agents that are defining what sort of features that has to be in there.
Issac Lee: We have, uh, QA agents. So all [00:21:00] these agents have different roles with different, um... They have different defined jobs, right? And then DaVinci, um, we use, uh, loop engineering within it. So there's gonna be an agent that says, "Okay, this is the task that we're trying to do, and this is what you guys are gonna work on.
Issac Lee: When you're done, report back to me." Uh, it, it reviews it and then, okay, we still have some missing, missing pieces, and we're gonna do it again. Reiterate, reiterate. So this is the sort of the process that we use to develop. Um, and a lot of the time is actually spent on designing, um, I'd say, core architecture of the product that we're trying to build and, and evaluating the output.
Issac Lee: Yeah. So development process in itself has, has been Almost fully replaced by, by AI. Um, and you know, like [00:22:00] half a year ago, deployment was deploying any sort of product service was kind of, uh, uh, there, there used to be a lot of human effort that needed to kind of, that was needed to push the product, um, out.
Issac Lee: But even, but now it's, it's just fully automated.
Demetrios: So one thing that I have my agents do when I'm building is when it-- Before I push something, I will ask it to go through, and it's almost like a QA agent, and I ask it to write a report With screenshots
Issac Lee: Mm-hmm ...
Demetrios: and click every button, click every link inside of whatever I'm building to make sure that everything's working properly.
Demetrios: Mm. And so it writes a report, it has screenshots, it says, "Yes, this is the correct page. This is, uh, working properly." And then it'll click a button, and it will say, "Okay, this is actually going to the right place." And if it, if there is a broken button or there is something that's not rendering properly, usually in the [00:23:00] screenshot it will catch that.
Demetrios: Mm-hmm. And then it can go and it can kick off another subagent to try and fix it, or it will fix it itself, and then, uh, push that through. What came to my mind as you were talking about this, bringing that original idea of having agents playing inside the games, you can almost have undercover agents that are just testing the game for you to see and make sure that everything is working.
Demetrios: Because you said that a lot of your time now goes into the evaluating.
Issac Lee: Yeah.
Demetrios: I imagine evaluating is playing the game.
Issac Lee: Right. Right.
Demetrios: And so how does the evaluating happen? Is it just you have a, you have game players there? You sit down on the weekend after you push something and you play it and, or is there, like, more of a mapped out way?
Issac Lee: Interesting. Um, we haven't, we haven't actually applied this, this QA concept to, to games yet. Mm-hmm. Um, not yet. Not yet. But I think, I think that's definitely an interesting [00:24:00] method for us to try out. Right now, we do human QAs. Um, obviously a lot of, a lot of, I guess, basic stuff has been abstracted by AI, but still, um, th- the thing that we're optimizing for is the user experience.
Issac Lee: So, um, I think it, it, it still requires human, human touch to say that, you know, "This is good. This is good to go out."
Demetrios: Yeah. Um- There is a little bit of magic there that is, that it's beyond that 10% that we were talking about earlier. When the game is prompt into existence, you still need some- something that will hook the person, and that you can't really get.
Demetrios: Even if every pixel is in the right place and everything moves the way it should be moving, since it's a game, you need that gameplay.
Issac Lee: Yeah. You need that human experience. Games are, games are for humans, I would say, for now.
Demetrios: Okay. So you could try. Maybe it would work- I think- ... on certain levels.
Issac Lee: [00:25:00] Yeah. I think, I think, um, I think doing so would help us to maybe, uh, find minimal bugs a lot faster- Mm-hmm
Issac Lee: before it's reported by players. So definitely I think there's an upside to it.
Demetrios: Yeah.
Issac Lee: Yeah. We're gonna try that out.
Demetrios: Yeah? Yeah. Well, yeah.
Issac Lee: Thanks for the
Demetrios: idea. You're gonna have to tell me how it goes. I do love the idea of, like, some of these, you fill the game with agents, like you were saying, and some of them are just undercover QA agents.
Demetrios: Mm-hmm. And they're reporting back like, "Hey, fix this," and it kicks off another agent loop, and it gets pushed, and so you have that, that type of thing. It would probably be a much bigger report than my agents make with my little dinky apps, you know? If you have a whole world that these QA agents are reporting back on and taking- Mm-hmm
Demetrios: screenshots of or making sure it works. But I, I do find that fascinating. So you've set up your own development environment. [00:26:00] You chose not to use some of the ones that were already out there. M- was that just because you started so early?
Issac Lee: I think it's partly that we, we also research a lot of existing open, uh, frameworks, and then we wanna kind of handpick the features that we want in development.
Issac Lee: Some of the things aren't really necessary to us, and, and I guess a lot of, a lot of, um, effort in, in designing this sort of development process is so that we can continue to refine it in a way that is, um, excuse me, um, that is, that is right for us. What's right for us might not be right for other sort of, um, development companies.
Issac Lee: Say, you know, if you're developing a financial app, what they are gonna be doing, h- their development process is gonna be a lot different than ours, right? So, um, I think that's the kind of the core reason why we decided to develop our, develop and maintain our internal framework. Another [00:27:00] thing is, um, security, right?
Issac Lee: So if you use external frameworks, um, it's, it's always... Yeah. You never know. The, uh, security team always wants, wants some sort of closed systems.
Demetrios: And am I right in saying, like, this framework, in a way, it's like you've built a harness for agents?
Issac Lee: Right.
Demetrios: And so this is fascinating to me because harnesses are, were very popular to be building feels like a, a few iterations ago.
Issac Lee: Mm-hmm.
Demetrios: Because of all these reasons that you're talking about, you wanna have that controllability, you wanna be able to use it and, and make sure that it understands your use case. Have you noticed as time goes on that it becomes less and less relevant? Because I think now it feels like the industry is moving in the other direction where they're saying, "You know what's the best [00:28:00] harness?
Demetrios: No harness." Or like the minimal harness that you get with Pi. Yeah. And that's, that's the good enough one.
Issac Lee: Right. So I think, I think harness is, is it, it could be a, a, it could be a very low thing as like, you know, um Whenever you're prompted A, you should go look at skill B, and then you should try to, um, do things in C order.
Issac Lee: That's a very-- That's a harness, right? Yeah. And it could be as complicated as, as, as anyone wants to be. But before, before harness was kind of a thing, we, we, we had our agents. We, we, we used these sort of methods to define, to tell AI that, you know, "Your job is A, B, and C. You should do it in this order," et cetera.
Issac Lee: This was, like, the initial sort of harness that we used, and back then we had, like, um, 200 different, um, sort of agents role defined.
Demetrios: [00:29:00] Uh-huh.
Issac Lee: Um, but frankly, as the models got better and better, this is, it becomes a lot- Stopped making sense. Yeah, it becomes not necessary. So, um, harness, it, it ca- it could be very useful when you know exactly what you wanna do, um, and you want agents to, um, kind of fall into this repetitive system, whereas doing so would get, would, would, um, help you guarantee some sort of quality of, of the work.
Issac Lee: Because if you do it in a certain way, the, the results are gonna be, the outputs are gonna, uh, are gonna be similar. So, um, from, from that point of view, harness is useful in making developments, um, in, in developing a product. Um, but as you s- as, as, as, as you said, um, I think as the model gets smarter, um, I think harness wouldn't, would not be necessary.
Demetrios: Yeah,
Issac Lee: all of these- The [00:30:00] importance of harness would decrease over time as models get smarter.
Demetrios: Exactly. The harnesses become thinner and thinner, and they're just there to really help you figure out... I, I, I just bring it up because I remember, what was it? Like a day or two ago, I saw or I was listening to a podcast about how exactly what you're saying, we used to have to define every agent.
Demetrios: We used to have to create every tool-
Issac Lee: Mm-hmm ...
Demetrios: in order for it to use. But now it's like, oh yeah, just give it to Claude or ChatGPT or Codex, and it will use Bash-
Issac Lee: Yep ...
Demetrios: if it needs to. Or like the-- It's, it's wild now how much we've moved away from that paradigm. And needing to really like say, "Here's the tools that you have, here's the..."
Demetrios: And if you give it GitHub, it's got infinite tools type thing. Right, right. Or you create your own [00:31:00] tools, so and you are a QA agent, you have these five tools, but now we don't necessarily need to do it that prescriptive, and we don't need that many different types of agents.
Issac Lee: Right, right. I think Codex and, and Claude, they're, they're harness, right?
Issac Lee: Exactly. So they're, they're a very, very good form of harness. It's-- I think if you try to like deviate away from that, it's because you have a specific need- Yeah ... that, that these, say, I'd say general harness is not kind of, it's not, it's not doing something for you, right? It's not that these are bad, it's that these are very general.
Issac Lee: But say that, you know, we want the agent that's developing the game to do a certain sort of, um, um, work, that, that's sort of where we would use harness.
Demetrios: Yeah. Are you guys using a lot of open source models as well as the closed source labs, [00:32:00] or do you-
Issac Lee: Yeah. Well, primarily we think we are currently developing with, um, Frontier Labs models.
Issac Lee: Um, I think, I think they're definitely, um, a bit better. They used to be much, much better- Yeah ... than open source, but I think they're still a bit better than open source models. But, um, since last month, um, I think, what did we see? The Kimi-
Demetrios: Kimi models.
Issac Lee: Yeah.
Demetrios: They blew it
Issac Lee: up. The, the- But
Demetrios: how are you doing-- I always like asking people, like, how are you accessing the Kimi models?
Demetrios: Because if you wanna run it on your own hardware, you need a few GPUs lying around.
Issac Lee: Yeah. So we're actually building that.
Demetrios: Really?
Issac Lee: Yeah. After, after we saw the Kimi news, um, last month, we, you know, we, we as a development company spent a lot of tokens every month.
Demetrios: Yeah.
Issac Lee: And-
Demetrios: You're not the only one.
Issac Lee: Yeah. Yeah.
Issac Lee: Um, and you know, when we look at news like that and see, you [00:33:00] know, if we could reach 80, 90% of the work that we're doing at, what, one tenth of the cost, yeah, we gotta look into it. So, uh, we've started to look into, um, look into, uh, building a small internal data center so that we would set up GPUs and then run it on our own.
Issac Lee: Have
Demetrios: your own rigs.
Issac Lee: Yeah.
Demetrios: That's very cool. That is, uh, where- In theory, it's 10% of the token cost, but then you always have to remember, like, how expensive is the engineer that knows how to work those GPUs- Yeah ... and the people that know how to, like, hook up the GPUs. Yeah. All of that stuff is, is part of it, which I think at the end of the day, if you amortize it over- Mm-hmm
Demetrios: a long, or depreciate it over a long horizon-
Issac Lee: Yep ...
Demetrios: then you're gonna come out on top, especially because you can continuously swap out the models as they can get better and better, and so you're [00:34:00] not stuck there. And these Like these GPUs, you're seeing that even A100s, which are, I don't know how many generations ago.
Demetrios: It's so, it feels like they're ages ago. The A100s came out and it was like, wow, that was amazing.
Issac Lee: Yeah.
Demetrios: You could still use those. Those are still very good machines, right? Like you could put a lot of model onto those machines, especially if you get a few of them together. Mm-hmm. So if you're to buy a few very beefy machines today, they're gonna last you a while.
Issac Lee: Yeah, this is something that we've just started to look into, but I think from, from what we've gathered so far is that we're not looking to train anything on these GPUs. Yeah. So we're trying to inference. And for inference, I think we're looking at at least five, five years- Yeah ... to use GPUs. So, um, if we spend X amount on ex- tokens [00:35:00] that are currently going to the Frontier Labs, it, I think it would make economic sense for us to set up and then just self-serve our needs.
Issac Lee: I guess one, um, one concern that I have is that, is, um, open source labs, they could always choose not to open source their weights-
Demetrios: Yeah,
Issac Lee: like- ... at some point in the future, right? ...
Demetrios: Meta kinda did that.
Issac Lee: Right.
Demetrios: They said, "Oh yeah, we're Llama fully open." And then they went and they closed the doors, but it seems like they're going back out to the open again.
Issac Lee: Mm-hmm. Mm-hmm.
Demetrios: Yeah. So it is a, it's definitely a valid concern.
Issac Lee: Yeah. Yeah. But even, even then, I think it might be worth it. I think it's worth for us to give it a try.
Demetrios: Yeah.
Issac Lee: Yeah.
Demetrios: Yeah, see what happens. And I th- I get the feeling that you'll be able to use the machines well longer than five years because as time goes on, and as you can start, we're already seeing that the [00:36:00] router- Mm-hmm
Demetrios: and the, the gateway type of pattern-
Issac Lee: Mm-hmm ...
Demetrios: for model routing is a very common pattern. And so as the machines get older, you can just put smaller models on them.
Issac Lee: Mm.
Demetrios: Or what today is quote unquote small maybe. And then- Right. ... as you get, then you get more machines, and you can, so you can continue after five years with those machines that you have today.
Demetrios: They're just might not be running the best and newest- Mm-hmm ... and greatest models on them. But it doesn't matter because they are of a certain quality.
Issac Lee: Right.
Demetrios: And so the model can, or the router can still route to them when they're needed.
Issac Lee: Right.
Demetrios: I find that stuff fascinating. It is a huge investment. I know some friends in Switzerland are doing it right now.
Demetrios: Mm-hmm. And the biggest headache that they've had is just, like, getting and setting up the GPUs. GPUs are very finicky. Mm-hmm. They're very-- They don't play nicely all the time, and you really [00:37:00] have to get down to a low level to do that, so.
Issac Lee: Interesting. Um, what, what-- Are they trying to build it for their own internal needs?
Demetrios: Yeah.
Issac Lee: Okay.
Demetrios: Yeah, exactly. And it's, they, they have it downstairs in their basement and heating their whole building basically. And, uh, it's, it's one of those things where they have some incredible engineers on it, but NVIDIA doesn't necessarily... They're not so open with how GPUs work that well. Like you- Oh
Demetrios: you program in CUDA.
Issac Lee: Uh-huh.
Demetrios: But if you really wanna know what's behind that, like what's lower than CUDA type thing, that's where I think it starts to get a little bit obscured.
Issac Lee: Okay. Okay. So from, I guess, running that software on GPU, um, it might require more knowledge than programming on CUDA, and that's not really not, it's not really, uh, common knowledge, so.
Demetrios: Yeah, [00:38:00] especially if you wanna... Uh, I, it's not common knowledge, but also if you really wanna optimize like the the GPU for what you're trying to do. I would say though, there's this great book. I actually interviewed this guy, Chris, uh, I'll have to find his last name and put it in the show notes because he wrote an amazing book on like optimizing inference for, for this specifically and he went through like, hey I...
Demetrios: He reverse engineered it. Oh, really? Because his friend, he had a lot of friends at NVIDIA, but they wouldn't tell him. They, they can't tell him. Right, right. I guess they've signed pretty hardcore NDAs, but he's able to poke around enough-
Issac Lee: Mm-hmm ...
Demetrios: to reverse engineer it and then he, he wrote it in this book and- Wow
Demetrios: he kind of gave his, all of his secrets. He shared all of them, at least as of when he wrote it, you know, a few months ago or-
Issac Lee: I'd, I'd love to get the name of the [00:39:00] book.
Demetrios: Yeah. I'll, I'll share it with you for sure. Thank
Issac Lee: you.
Demetrios: It's a, it's a great one. The other piece on inference-
Issac Lee: Mm-hmm ...
Demetrios: specifically is you've got vLLM, which is a killer open source product, and that helps a ton for running-
Issac Lee: Mm-hmm
Demetrios: these models and on the GPUs. And I know there's different companies that have worked really hard at making that even faster.
Issac Lee: Oh, wow.
Demetrios: And so you can hopefully find stuff out there that will, um, that will get you like super fast inference-
Issac Lee: Mm-hmm ...
Demetrios: on top of this, you know, lower cost. So there's all the GPU providers now, right?
Demetrios: Like there's the, uh Neo clouds is what they're- Yeah ... called.
Issac Lee: Yeah.
Demetrios: And some of them have different value props. Like I look at, [00:40:00] have you heard of this com- like Modal or Base10, these neo clouds?
Issac Lee: Mm-hmm.
Demetrios: Their value props, like Modal's value prop is you can fire up a GPU really fast. So you can go to serverless-
Issac Lee: Mm-hmm
Demetrios: and then you can go out of it and wake up the GPU in like seconds as opposed to minutes.
Issac Lee: Okay.
Demetrios: And so it's just really fast, and they've done a lot of work, really, really hard work on the kernel to be able to boot it up that quick and load everything in, and get your model and your data, everything that you need, there ready for you.
Demetrios: And then you've got Base10, and from what I've understood from them is they did a lot of hard work on the inference side. Mm-hmm. And so they said, "vLLM is great. We're going to optimize it even more." And so they've written like this Frankenstein version that is very fast, uh, of that.
Issac Lee: That's awesome.
Demetrios: Yeah.
Demetrios: There's, there's all kinds of ways that folks are trying to optimize all of this. And when you start [00:41:00] rolling your own- Mm-hmm ... you get, you go down these rabbit holes, you know? Right. There's a really good, um, community, I think it's called like GPU Mode, and they basically just talk about GPUs all day- ... and this stuff.
Issac Lee: Definitely something that I need to look into.
Demetrios: Yeah. Yeah. Yeah. If you're gonna- Yeah ... if you're gonna
Issac Lee: p- Yeah ...
Demetrios: you know, pay the price of a GPU-
Issac Lee: Uh-huh ...
Demetrios: or a few of them-
Issac Lee: Uh-huh
Demetrios: Then you should find the, the communities and the literature where people are, you know, talking about this stuff.
Issac Lee: Will do. Will
Demetrios: do.
Issac Lee: Yeah.
Demetrios: Well, I know you've been working on a ton of different experiments in the AI lab. Can you walk me through a few of these? We already talked about DaVinci, which was one, right? Yep. What's-- what are other ones?
Issac Lee: Right. So one other one is called, uh, Text Through 3D. I think I talked about this briefly, um, where you prompt a text, and then it's able to build you some sort of a world, and within that [00:42:00] world, it's able to build you structures.
Issac Lee: And then in this structure, um, you're able to kind of go into, say, you know, "I wanna play an FPS in this map." Then it's gonna give you the abilities to go into the, um, first-person kind of settings, and then you, you'll be able to shoot around. Wow. So it integrates multiple different, uh, physical engines, and AI is able to help you build this sort of, uh, environment.
Issac Lee: So that's, that's one thing that we did.
Demetrios: You've got like something that is a, an engine where physics works.
Issac Lee: Yeah. It's not that we develop our own engines. I think we integrate existing engines- Oh, okay ... to call upon it. So for example, like let's say one of the, one of, one of the reasons, I guess what the background on how we started this was we were thinking of, um, creating, uh, sort of like a map editor for Roblox.
Issac Lee: A lot of people build on Roblox, and if you try [00:43:00] to actually build on it, it's, it's a lot of hassle. You have to learn how to use it.
Demetrios: Uh-huh.
Issac Lee: Um, and the existing AI enhancers are not that useful, right? So we, we try to- Ironically ... we try to use the physics engine that is provided by Roblox and then, you know, help creators build games, um, for Roblox.
Demetrios: Uh-huh.
Issac Lee: Um, so that's how we initially started. Uh, we used some multiple different engines to kind of, um, do a POC. Um, and as I said, it, it, it's, it's you, you integrate a... Let's say that You have to first define what you're gonna enable creators to do, what tools you're gonna be able to help them with. What-- Well, sorry, what tools are available, and that- And
Demetrios: tools you mean like which engines- Yeah
Demetrios: not like which tool calls the model can make.
Issac Lee: Right. Right.
Demetrios: Yeah. Well, I guess in a way it is kind of a tool call if it's calling- It's very similar to that ... the [00:44:00]
Issac Lee: engine. Yeah. Yeah. You, you have to tell the model to use, you know, when you create a world, think about this specific construct or this specific, um, construct.
Issac Lee: Say that, you know, when you, when you're building a city, you have to have A, B, and C, um, and whenever you're-- And then that would be the sort of, um, baseline of where we start. And when a character interacts within that, um, environment, that world, um, it should have these sort of, um, capabilities. Yeah. So it, it's very similar to a tool call.
Issac Lee: Yeah. Um, it's just that you have to, you have to lay out what you, what the model has to do in order to achieve the desired outcome. That, that's one thing. Um, another thing that we built was a short form generator. So as a game developer and publisher, we have to market our product, right? Yeah. So, um, one thing we built was that you're able to-- So the, the model [00:45:00] is able to watch the video and then create, reverse engineer the prompt for it You're able to create, input, um, any sort of image along with, um, prompts to generate a video.
Issac Lee: So it's, it's, it's even it-- I guess it, it helps you to create pretty okay videos with, say, like two sentences of prompt and a few snapshots of the game.
Demetrios: And videos like gameplay video-
Issac Lee: Yep. Yep ...
Demetrios: so that you can showcase what is possible.
Issac Lee: Right. Right.
Demetrios: Wow.
Issac Lee: So that, that's one thing that we built, um, that we're using it.
Issac Lee: We're-- we-- this is a product that we're using in production to market our, our cur- our last game, w- well, our current game that we published like a month ago.
Demetrios: Wow.
Issac Lee: Um, so that's one thing. And then, um, we have-- we built our own internal meeting note, uh, meeting note- Uh-huh ... taker. Yeah. We call it Bob. And one of the things that it does is that you're able [00:46:00] to turn it on and you, you talk...
Issac Lee: It, it runs on local model.
Demetrios: Mm-hmm.
Issac Lee: It runs on, uh, the local, local GPU of the computer, and then, um, it, it, it, it's able to kind of whenever-- Let's say that we're discussing about, um, some theory, some idea, and like GPUs, right? We're saying that, you know, A100, H100, et cetera, what model is it leased? How much does it cost?
Issac Lee: You know, if we- Yeah ... run inference on it, what's that gonna come out to? Bob is-- Bob takes notes on these, and then it's able to, on, on a side chat, it's able to, "Hey, can you verify that?" And it's gonna, it's gonna go on and verify that. Oh. It's gonna make another call to a model to verify any sort of important things that we're discussing real time.
Issac Lee: So, um, we think that could be very handy. It's not, it's not fully, um- We don't fully use it. It's still under development. We don't fully use it within the company yet, but I think in, in a few weeks, [00:47:00] I think it's gonna go out.
Demetrios: Oh,
Issac Lee: that is cool. Um, one other, I think, extra thing that, um, we're doing is, uh, it's a product called Keyro.
Issac Lee: You know
Demetrios: what... Sorry, just- Yeah ... one other thing that came up when I think about Bob-
Issac Lee: Yeah ...
Demetrios: is a lot of times I talk with the same people.
Issac Lee: Mm.
Demetrios: E- especially at work. I would love if Bob knew who I was, you know? Like- Right ... so that it doesn't have to-- So it has context around me as a- Mm-hmm, mm-hmm ... person. What's my job?
Demetrios: What do I do? What are my tendencies? I like my notes like this.
Issac Lee: Mm-hmm.
Demetrios: Bob, when you send me notes, send them to me in my style because this is related to my job. And then maybe a sales guy who's in a call with you guys- Yeah ... he likes his notes in a different style. Right. So send him those notes. Right. And make sure that, like, each individual person or each individual team- Right
Demetrios: has their own [00:48:00] style that they get from Bob. Right.
Issac Lee: So you're asking for personalization.
Demetrios: Yeah, a little bit more personalization. Interesting. Yeah. That, and especially around... 'Cause I, I got this-- Actually, this is quite nice. If you have a new note taker and you really wanna make sure that it understands everything and it has the full context.
Issac Lee: Mm-hmm.
Demetrios: My friend Fausto used to, before we would actually start the meeting, start the meetings, he would make us introduce ourselves.
Issac Lee: Mm-hmm. '
Demetrios: Cause then the note taker would have that context.
Issac Lee: Mm-hmm.
Demetrios: And so if you don't have to do that, or so you don't have to do that every time, every time you have a meeting, you can have the note taker just remember that.
Issac Lee: Yeah, I think that's, that's, yeah, definitely something that we're gonna, we're gonna try it out. Yeah. Thanks for the idea. So we, I think we could build, um, say a profile of- Yep ... everyone that's in the meeting based on the discussions, and then we could also, you know, tell [00:49:00] any sort of profile to Bob. Exactly.
Issac Lee: And Bob would build, um, outputs based on the profile.
Demetrios: Yeah, and you could potentially have this just with some... I mean, if you wanted to get wild, you could create a, you could embed Different chunks of what folks are saying and then have the, um, the profiles and have it in a vector store and all that fun stuff.
Demetrios: And at the end of the day, you almost are like building a recommendation system on top of Bob, which would be very cool, but you don't, probably don't even need to do that. You just create some simple file system-
Issac Lee: Mm-hmm ...
Demetrios: that is like, "Hey, this is Demetrios' file. Bob, whenever you write, like invoke the Demetrios skill that explains how Demetrios likes to get his stuff back."
Demetrios: Mm-hmm. "And here's a few like short sentences on Demetrios."
Issac Lee: Yeah. Yeah.
Demetrios: And so you can take it to like two extremes. I don't know
Issac Lee: how- But, but I think, [00:50:00] I think that's kind of the future that we're gonna live in pretty soon. Whether, whether or not we build it, I think we're gonna see these sort of products with AI.
Issac Lee: I, you know, I, I'm sure, I'm sure you'd agree is one of the things that's very, very critical for AI is, is giving it context, right? Yeah. So if we, if, if someone finds a very, um A very good way of delivering context of the things that we're trying to get it done as a company, um, I think that would be very interesting for us.
Demetrios: Yeah.
Issac Lee: Not, not just for us, but for a lot of the, a lot of the, a lot of people out there.
Demetrios: Yeah. 100%. And the thing that I think about is, like, how much context is too much context. Mm-hmm. Because on my system at home, I am-- It's fully untethered, and I'm like, "Whatever you need-
Issac Lee: Mm-hmm ...
Demetrios: you can have [00:51:00] it." I-- The more data, the better.
Demetrios: Mm-hmm. So, like I notice you're wearing an Apple Watch. I normally wear, um, one of those health bands.
Issac Lee: Mm-hmm.
Demetrios: And if I can, I wanna give all of that data via the API to my system also, so it can just, like, go and make a daily report. "Oh, check this out. I see that when you were in this meeting with this coworker, your heart rate spiked.
Demetrios: Were you angry? Were you happy? What was it?" You know? Like, so you can- You can reflect ... reflect back on that type of stuff.
Issac Lee: That's interesting.
Demetrios: But at work, you may not want that. Like, you don't wanna give all of that health data to the- Right ... job so that they can see, like, "Dude, you get stressed every time you're in our standups.
Demetrios: What's going on?"
Issac Lee: You know, um, I saw, I saw an X post some time ago that was, there was a woman that recorded all of her, um, all of her dates. Oh. And then she uploaded that to Claude and analyzed it. And that was, it was, it was [00:52:00] kinda getting a lot of heat from, from the fact that she was recording it- Yeah ... et cetera.
Issac Lee: But, like, she was able to analyze her, her dating partners based on their interaction. But, you know, I think- Exactly. Yeah.
Demetrios: It does get weird. Yeah. That's why I'm saying, how much context is too much context? Right. 'Cause you can take it a little too far, and these-- Well, who am I to say that it's too far, right?
Demetrios: Like-
Issac Lee: Yeah ...
Demetrios: we're exploring right now what is too far. Uh, is that weird? Is it not? I don't know.
Issac Lee: Yeah. We don't know. We don't know. Yeah. I- Well, well, that, that actually might be the sort of, sort of the, um, future that we might have to get accustomed to, whereas- Yeah ... it's not just human that's making any sort of decisions.
Issac Lee: But there's always some sort of, um, some sort of, uh, watcher or AI observer- Yeah ... that kind of i- sh- that kind of studies the interaction and tries to tell the human to iterate based on that. Um, not sure if that's bad.
Demetrios: I don't know either. I, I know that it gives [00:53:00] fear to a lot of people.
Issac Lee: Mm-hmm.
Demetrios: I remember when I was first getting into college, university.
Issac Lee: Mm.
Demetrios: Photos, like taking photos of people-
Issac Lee: Mm
Demetrios: was not as commonplace as it is now, 'cause it's on our phones.
Issac Lee: Right.
Demetrios: So you had to have a camera and you had to actually whip it out and then take a photo. Right. And it was more of a thing-
Issac Lee: Mm ...
Demetrios: to take a photo, and a lot of times you would get caught in the back of a photo and there was like, then photo bombing became a thing.
Demetrios: You would jump into other people's photos. Now, if I'm caught in the back of somebody's photo, whatever. Like- Right ... I, that probably happens to me 100 times a day. Back in those days it was like, "Oh no, I'm, those people are taking a photo, let me get out of their photo range." Gotcha. You know? If I did that today, I'd never get anything done.
Demetrios: Right. Right. There's so many photos being taken all around, and so it, it kinda feels like that, [00:54:00] like we're gonna get more accustomed to-
Issac Lee: Having less sort of privacy, I guess ...
Demetrios: yeah, or having more AI in our life, more context that we're giving the AI. Who knows? Maybe that's a dystopian future, maybe not. I like how you're saying it could actually be useful for us.
Issac Lee: Yeah. Yeah.
Demetrios: Gotta get used to it. Depends on how you
Issac Lee: use it, right?
Demetrios: Yeah. Yeah. Is there another experiment that you had in
Issac Lee: there? Um, we did a marketing, we built a marketing agent. So, um, it helps you to, um, analyze the Presence of your, um, product, of your, say, game, a product, a service, whatever, whatever, whatever you're trying to push.
Issac Lee: Um, it, it, it, it, it's connected to the SNS, mainly as X, and then it cause all the discussions on how your product is being discussed and how people are talking about it, what sort of how you should position your [00:55:00] product. It gives you an analysis. It helps you to push it. So it also runs on a loop, so it helps you.
Issac Lee: So it's got a few different agents. The, an a- analyst agent that analyzes your product and says, you know, "Maybe the things that you should position for are A, B, and C." And then it's got agents that collect the data and say, you know, "You're only mentioned X times, uh, per day. You should try to get it up by doing A, B, and C."
Demetrios: Yeah. It's giving you advice.
Issac Lee: Yep. Yep.
Demetrios: Where my mind goes on that is not necessarily on this product, but I think about how I used to pay for API access to some of these. Or for example, like LinkedIn, you can't even get API access- Yeah ... if you want it. And so it's, it was notoriously hard to get that kind of social listening for a platform like LinkedIn, but now [00:56:00] I just use computer use.
Demetrios: And-
Issac Lee: Right ...
Demetrios: so it's like you bypass all APIs. Which to me is like, oh man, they're going to figure out, like, that I'm doing this, and there's danger there. But Until they do, this is quite useful.
Issac Lee: Mm.
Demetrios: You can s- you can just bypass the API. And granted, it's much slower.
Issac Lee: It is a lot slower.
Demetrios: And you- Yeah ... so if you're doing it on a scale that you guys are probably doing it on, it's a much different thing.
Issac Lee: Mm.
Demetrios: But it is a, a way to try and bypass, like- ... the LinkedIn policies.
Issac Lee: Yeah. But I think, I think going forth, like not providing an API would probably mean that you're not-- Uh, it comes with-- It's, it's a double-edged sword, so to speak. Um, you're gonna get hit by a lot of the agents.
Demetrios: Yeah.
Issac Lee: But if you're not being hit by the agents, you might not get traffic at all, 'cause like, you know, if you-- [00:57:00] Like even for little things, right?
Issac Lee: I think, I think one of the earliest, um, adoption of Agentic future, I think would be shopping.
Demetrios: Yeah.
Issac Lee: Yeah. So agent has, as we said, agent has a lot of context of, of myself, what I eat, you know, my health data, what I do, and then, you know, let's say that if I was searching for, you know, my eyes are a little dry, I want some, you know, eye drops.
Issac Lee: What should I buy? And agent is gonna, it's gonna call via API and then figure out what the best thing is for me, and then say- Yeah ...
Demetrios: you
Issac Lee: know?
Demetrios: My agent usually just tells me to stop smoking weed when that happens. So...
Demetrios: But I guess that's the personalization we were talking about before, right?
Demetrios: But no, I, I was thinking about this a lot because Perplexity got sued by Amazon- Mm. ... for their shopping agent, and it was very much doing-- [00:58:00] just using computer use and then getting everything on Amazon.
Issac Lee: Right.
Demetrios: And Amazon said, "Hey, you can't do that," and so they sued. But recently, I think I saw that the court ruled that it is okay for Amaz- or for Perplexity to have a shopping agent for you-
Issac Lee: Mm-hmm, mm-hmm
Demetrios: uh, on Amazon, and so Amazon can't-- They, like, have no grounds to sue, at least at this point in time.
Issac Lee: Right.
Demetrios: And that's a huge loss for Amazon.
Issac Lee: Mm-hmm. '
Demetrios: Cause now they are not getting that first-party data from their customer. Well, they are kind of getting it, but it's through a agent.
Issac Lee: Yep.
Demetrios: And then second of all, everything on Amazon or any e-commerce website, for that example, they're-- It's been hyper-optimized for humans.
Demetrios: So you have the pop-ups. Yeah, yeah. "Hey, save 20%," uh, "if you give us your email." That's a human thing. Like, we do that because [00:59:00] hopefully it gets you to buy. Right. Or they upsell you with, "Oh, well, if you get three more dollars' worth of products, we'll give you free shipping."
Issac Lee: Mm-hmm.
Demetrios: And those things work on human psychology.
Issac Lee: Yeah.
Demetrios: But for an agent, it's like, "No, I just want pants. I want pants." Right, right. "I want pants. I don't care about no email. I don't care about the save $3. I just was ordered to get pants." And so we have to think about optimizing now for agents, and what does that even look like? What does that even mean?
Demetrios: Mm-hmm. Because an agent is tasked with doing something, it goes right directly and it does it. Or it gives you 10 different options.
Issac Lee: Mm-hmm.
Demetrios: And one of those options is Amazon, another one is just the random e-commerce store.
Issac Lee: Yeah. Yeah. So I think, I think going forward, like as you said, optimize, should we optimize for, for human users or should we optimize for agents?
Issac Lee: Yeah. And, and optimizing for agents, if I had to guess, is I guess providing [01:00:00] as straightforward info as possible.
Demetrios: Mm-hmm.
Issac Lee: Um, or if you're able to kind of retrace the origin of the agent. Uh, right now, um, I guess the, most of the... Well, all of the API doesn't kind of authenticate on who the calling party is.
Issac Lee: You have no information on who the caller is, right? So perhaps I think from a product or service perspective, that might be an interesting way to gather information about, "Okay, so you want access to some sort of functions or calls. Who are you?"
Demetrios: Yeah. "
Issac Lee: Tell, tell me about you so that we can give you a set of-"
Demetrios: The right things- Right
Demetrios: that you're looking for. Right. Or we can show you some sponsored images also. Yeah. Because that's another piece that like, where's that gonna go? All that sponsor revenue- Mm-hmm ... or the, the ads revenue inside of a e-commerce store like Amazon.
Issac Lee: Mm-hmm.
Demetrios: This has been great, dude. Yeah. It's been fun. I appreciate you doing [01:01:00] it.
+ Read More
Watch More
Inside Uber’s AI Revolution - Everything about how they use AI/ML
Posted Jul 04, 2025 | Views 2K
# Uber
# AI
# Machine Learning
Generative AI Agents in Production: Best Practices and Lessons Learned // Patrick Marlow // Agents in Production
Posted Nov 15, 2024 | Views 6.5K
# Generative AI Agents
# Vertex Applied AI
# Agents in Production