[00:00:00] Whitney Lee: Uh, I understand, um, the way you started with AI, the way you learned how to use it and how it acts is with, uh, Dungeons & Dragons. Is that correct?

[00:00:12] Coté: That's, that's right. I-- That may not be incredibly accurate, but it's how it's now remembered. Uh, I, I think, uh,

[00:00:19] Whitney Lee: It's a good story, so we're going with it.

[00:00:21] Coté: The answer, the answer is yes.

[00:00:23] Michael Forrester: I, I, one, I, I don't wanna get hung up on incredibly accurate as a definition, but yes. Okay Yeah

[00:00:37] Whitney Lee: You're not supposed to talk yet until we bring you on.

[00:00:40] Michael Forrester: Oh, okay. Sorry. I'm gonna, I'm gonna go back. Incredibly accurate totally threw me. I just didn't know how to respond to that. I was like, "What in life is incredibly accurate?" I just feel like that should be a battle cry that just

[00:00:50] Coté: Yeah, yeah.

[00:00:51] Michael Forrester: Incredibly accurate. You know? Anyway, okay, keep going.

[00:00:53] Sorry.

[00:00:54] Whitney Lee: To set the tone for the podcast, and I think we just did. We have a bit of a ver- verbose guest today. Um, Coté,

[00:01:05] Coté: Yes

[00:01:06] Whitney Lee: uh, tell me about D&D, like your AI D&D experiences, just generally

[00:01:12] Coté: Oh, just, just, uh, just in summary. Well, uh, uh, well, I started doing it because, like, you know, I used to play D&D and, and I wanted to just, like... It's, it's hard to find a phrase that isn't play with yourself in English, but I wanted to, I wa- I wanted to, you know, play on my own as-- They call it solo role-playing.

[00:01:34] Um, and, and then I had just, I had just sort of like... You know, I think I did a, a few things with-- This is when ChatGPT 3.5 was out, and I'd done a st- a few things here and there, and then I had a, uh, I had so- a long stretch of time where I actually had the cognitive capacity. I was away from all my kids, uh, and also didn't have work responsibility.

[00:01:55] Whitney Lee: Quickly. I was away from my kids, so I decided to play a game.

[00:01:59] Coté: Yeah, yeah, yeah. And, uh, and then, I don't know, I just started playing with it and, you know, back when the, the, the only interaction you could do was like cut and paste basically, like in and out of it. And, and, um, yeah, so I think, I think it's, it's, it's, it's proven. It's ended up being like really handy because I don't have a lot of...

[00:02:21] Uh, this is a in- inaccurate statement, but I don't like have a lot of, uh, I haven't had a lot of other reasons to use AI stuff like deeply. You know, I just like write stuff and whatever. And so like now I do because we've got like Claude Code and stuff. But in those years before like when was it? Before November of 2025 or whenever the Claude Clo- Code

[00:02:42] Whitney Lee: Back in the

[00:02:42] Coté: wave was.

[00:02:44] Yeah, yeah. Like, like, uh,

[00:02:45] Whitney Lee: Less than a year ago.

[00:02:47] Coté: like, like there, there wasn't really-- It was difficult to do anything beyond just chatting unless you were like a programmer. Whereas if you have D&D, it's like, so this is like, you know, my jokey pitch when I give this talk is like, I'll give you the punch...

[00:03:03] Whitney Lee: me a talk right

[00:03:04] Coté: No, no, no, no. I, I'm just walking through.

[00:03:06] I'll, I'll, I'll give you the punchline is like, I think it's more complicated than any like enterprise system out there, at least as complicated. And so like if you, if, if the AI can like play D&D, it can do whatever enterprise stuff you want, like it's not a problem. And so,

[00:03:22] Whitney Lee: premise.

[00:03:23] Coté: yeah, yeah

[00:03:24] Whitney Lee: did any, did AI... Are you able... I, I don't know a lot about D&D. Are you able to play it on your own without AI, or did it, like, enable this for you? That's-

[00:03:35] Coté: yeah. No, there, there's, there's a big, there's a big... In, in all role-pl- well, not all. In many role-playing games,

[00:03:41] Whitney Lee: Mm-hmm.

[00:03:42] Coté: there are-- There's systems and mindsets for, for playing on your own. And of course, there are games that are intended to be solo player games, like classic, you know, you must have seen the Choose Your Own Adventure books.

[00:03:54] You can go to, you can g-

[00:03:55] Whitney Lee: I have.

[00:03:56] Coté: can go down to Toy Joy and buy all of them now. And then there, there's these other books like, like the Lone Wolf books. There's all these kind of books like that. And then there's also, like, there's also a few in the past 10 years, like a few systems, uh, that are built around it called-- I've never really played them, called like Ironsworn or Sworn this or Sworn that, and then some other ones.

[00:04:16] But like, I don't know. I don't think that the solo role-playing Reddit groups listen to this podcast, so I'll go ahead and say it. But like, they all kinda suck, right? Like they're all, they're all more like a way of introducing randomness into writing your own stories, which is cool. It's fine if that's what you wanna do, but it's not-- That's a lot different than like playing D&D, playing a role-playing game.

[00:04:45] Like it, it's, it's a different w- it's a different way of playing, but it's not what I look for in doing it

[00:04:51] Whitney Lee: I, I have more questions, but I'm gonna bring our guest on now. He's making a lot of faces. I can tell he has stuff to say. Uh, especially when you, Coté, asserted that, um, uh, playing D&D is as complicated as an enterprise systems. I feel like Michael, maybe you have feelings about that. Before you say your feelings, introduce yourself, please.

[00:05:16] Coté: How polite.

[00:05:17] Michael Forrester: Wait, biography before feelings? Is that

[00:05:19] Whitney Lee: Yeah.

[00:05:20] Michael Forrester: happened? man, I was just gonna open with some poetry. Um, so hi, I'm Michael Forrester. Yeah, I have been in the IT industry for quite some time, and, uh, like Coté, I use AI in my daily life. I will say, though, that I was an early, like, beta tester for Claude Code back in like February of 2025, so I've been using it quite a bit. I've been in the IT industry for 30 years, which just means I have a lot of opinions, not necessarily knowing a lot of stuff. I talk publicly like KubeCon and other spaces, and currently I was working for CodeCloud. Wait, da, da, da, da, da, CodeCloud, which is an awesome platform. I

[00:05:56] Whitney Lee: He's pointing at a CodeCloud clad flag behind him.

[00:06:00] Michael Forrester: pointing

[00:06:00] Whitney Lee: Yeah

[00:06:00] Michael Forrester: red CodeCloud flag that I still have on my wall. actually do workforce, AI workforce transformation for Accenture currently, so

[00:06:07] Whitney Lee: Oh, juicy

[00:06:08] Michael Forrester: role. Yeah. And I live, uh, in Atlanta with my son, who's 19, so he's going off to college.

[00:06:16] Whitney Lee: Awesome. Okay.

[00:06:17] Michael Forrester: have played Dungeon & Dragons

[00:06:19] Whitney Lee: Yeah

[00:06:19] Michael Forrester: since I was like 12.

[00:06:22] Whitney Lee: What do you think about Coté's ideas that it's as complicated as an enterprise system?

[00:06:27] Michael Forrester: Mm. remember back when I was a, a, a wee lad and, uh, I w- uh, got my first, like, player's handbook, like Dungeons Master guide, like all the books that you need to kind of play. And I remember opening these things up and they were entire systems. It was like, uh, in a sense, like mathematical and qualitative ways to define ecosystems like combat, role-playing, exploration. you could get specialized books on like how to sur- like what you needed to do to, to rule and check in the wilderness

[00:07:00] Coté: So many books

[00:07:02] Michael Forrester: underground or, uh, w- the Manual of the Planes. This was like

[00:07:06] Coté: Mm.

[00:07:06] Michael Forrester: which was a book, it had this great like monster beast on it, but you could go to other planes of existence and there were rules and dice rolls that you could make about behavior and choices in those.

[00:07:18] So I agree. I mean, any one of those books would've been an entire enterprise system in and of itself if you added any, like the additional books, like for different contexts. Yeah, I, I agree. They're, they, they are extremely complex

[00:07:33] Whitney Lee: D&D is for people who like systems but don't like money

[00:07:39] Michael Forrester: Well, I mean, th- they don't want to make money themselves. They just wanna make other people money. Like Hasbro, for example, who owns, right? Or TSR was the original owner, but yes, I think Hasbro is

[00:07:50] Coté: there, there, there is a good joke there that D&D books cost like 50 or $80. So yes, they enjoy helping other people make money.

[00:07:56] Michael Forrester: Yeah, yeah. Yeah. And then

[00:07:58] Whitney Lee: And if you have that skill set, if you use it toward cloud stuff, you're very

[00:08:02] Coté: Yeah, yeah. Yeah. It's, it's true. I, I think, I think there's a perhaps un- not n- a, a very unsurprising crossover between people who've played D&D and people who like to talk to computers. There's,

[00:08:13] Michael Forrester: Oh

[00:08:14] Coté: li- a com... Yes. It's systems people

[00:08:17] Michael Forrester: Yeah. Well, and

[00:08:18] Whitney Lee: Is it

[00:08:19] Michael Forrester: I think it does help communication from a storytelling perspective, but I think also the ability to learn the rules of a system maximize the output of those rules for the thing you're trying to accomplish, I think, I think you're right.

[00:08:34] Whitney Lee: Are we,

[00:08:34] Michael Forrester: talking to computers is the same

[00:08:36] Whitney Lee: are we talking about D&D or, or g- uh, systems now? Or like cloud systems? I don't even know.

[00:08:41] Michael Forrester: Why do we have to draw lines, Whitney? I don't understand. Are you trying to put us in a

[00:08:46] Coté: Yeah. Yeah, you, you could-- you're, you're either figuring out how to, uh, get, get an invoice through the system or pick the pocket of a wizard.

[00:08:54] Michael Forrester: Yes.

[00:08:54] Coté: same thing.

[00:08:55] Michael Forrester: yeah, it's the same, same difference, right? Wizard, invoice. is the world we live in, right? Or, or the first time I ran my son and my nephew through a campaign, like during COVID, they had never played before. And, they were... I have to say that because they had not read any fantasy books and didn't have any kind of sense of like, you know, King Arthur or Lord of the Rings or whatever, they did some of the craziest things

[00:09:22] Coté: Mm.

[00:09:23] Michael Forrester: know the system and didn't have any sense of like, you know, propriety for, you know, lack of a better word, or like what other, you know, context had done. So, I mean, they were... Liam, uh, my son, who's 19, creature that we saw, it didn't matter if it was a floating eyeball or a giant thousand-year-old dragon, he was like, "Can I, can I make that into a pet?" Like, that was his question like every time. I was like, "So I just wanna mention like this thousand-year-old dragon's like trying to cook you with magic, and you wanna see if you can make it a pet."

[00:09:51] He's like, "Is there something I can offer it to kind of entice it to kind of be my pet?" And he would like ask me these questions, and I'd be like, "I wanna reward your bravery and like boldness, and I just don't really know, like, I'm not really hearing like solutions for this kind of thing." So I did never took it off the table, but it, it is fascinating about how people game systems if they don't have a sense of what they think is appropriate

[00:10:13] Coté: Yeah, yeah, yeah

[00:10:15] Michael Forrester: So like invoices and wizards makes total sense to me because they're both just processes somewhere.

[00:10:21] Whitney Lee: So, so how does AI affect, I guess this is, is for both of you, how, how you, um, do Dungeons and Dragons differently? Like I, I assume there's, there's some creativity involved, there's world-building involved. Sometimes it seems like you're being handed the worlds also. Uh, I'm a little fuzzy on it, but I'm wondering where does it, like, are you outsourcing the creativity to AI, or are you finding, are you able to be creative at a, a different scale?

[00:10:50] Michael Forrester: Hmm. Coté, you wanna go, go for that one or shall--

[00:10:52] Coté: You, you, you're the guest. You go first. Otherwise, I'll talk for the rest of the episode.

[00:10:56] Whitney Lee: He really will.

[00:10:57] Michael Forrester: it like 30 seconds and then Coté's gonna catch us up. All right. Uh, I would say that in this day and age, considering that there's-- w- since we were just kind of waxing about priorities before we got on live, is that the one thing that's interesting for me around AI, around entertainment and games and systems, is that it normally, for examp- and just I'm gonna give specific context just to keep this short, is that like my son approached me at one point and said, "Hey, I wanna do a game, but I don't necessarily wanna revisit Dungeons & Dragons 'cause we spent three years on that."

[00:11:29] And so then he came to me and he said, "I wanna do a sci-fi game. I wanna switch off of Dun- Dungeons & Dragons for half a second." And I said, "Okay." I said, "What do you wanna do?" And he's like, "I'm thinking about this thing called Lancer." And so I was like, "Great." He's like, "But I wanna do my own thing. I don't wanna do what Lancer does.

[00:11:42] I wanna do a home brew, like a, a custom created campaign."

[00:11:45] Whitney Lee: What's the answer, please?

[00:11:47] Michael Forrester: So Lancer is a sci-fi mech game, and mechs are giant, basically multi-story machines, kind of like Pacific Rim, um, where you can like hop in them and control these giant robots. And some of them are fast, some of them are slow. They all have different characteristics.

[00:12:01] Some of them can fly into space, some of them are stuck on the ground. And so you basically play a pilot for one of these giant

[00:12:10] Whitney Lee: Okay

[00:12:10] Michael Forrester: they're powered by like, you know, magic, and sometimes they're powered by like nuclear generators. And so he wanted a game where he could pilot a giant mech and, and with the family and friends, and they could do adventures while piloting these giant mechs. So funny thing that the AI enabled, not to get too far into that, is that the thing that AI enabled for me is that normally I would say to myself, "Oh, I don't know that I have time to flesh that out

[00:12:35] Coté: Hmm.

[00:12:35] Michael Forrester: way." Like I don't, I don't, I don't know that I could spend the 60 to 100 hours necessary to get that to a usable state for Liam,

[00:12:44] Whitney Lee: And

[00:12:44] Michael Forrester: based on his request.

[00:12:45] And AI

[00:12:45] Whitney Lee: what

[00:12:46] Michael Forrester: made it so that it was 20 or 30 hours

[00:12:48] Whitney Lee: So this may be me being naive to the field, but what are you fleshing out exactly? What is that 100? Is it the world? Is it... Yeah, what, what are you figuring out?

[00:12:58] Michael Forrester: Yeah, so that's a great question. So imagine that you've decided, like, you, you have a recipe for making pizza, but someone comes to you and says, "I want my own custom pizza," right? And in this

[00:13:06] Whitney Lee: Yeah

[00:13:07] Michael Forrester: let's say it's egregious and they want ham and pineapple and all kinds of stuff on there that you normally wouldn't put on there.

[00:13:12] Whitney Lee: Uh-huh

[00:13:12] Michael Forrester: But let's say you have to make it. Let's say it's a special cheese and all this other stuff, and so there has to be all these ingredients. So what I did is I borrowed the base recipe, like I would borrow the base recipe for a pizza, from Lancer. So Lancer had a

[00:13:23] Whitney Lee: Mm.

[00:13:24] Michael Forrester: that said, "This is what combat looks like, this is what exploration looks like." And so what I did is I borrowed that base recipe, and then I added my own ingredients on top of it. And

[00:13:33] Whitney Lee: Okay

[00:13:34] Michael Forrester: happen is I would have to learn all the ins and outs of the Lancer system, which is not big, by the way, way,

[00:13:38] Whitney Lee: Mm-hmm.

[00:13:39] Michael Forrester: than Dungeons & Dragons. And I would have to learn all of the ins and outs of that system in detail before I could home brew my own campaign above it. then, and then if I had an idea of like, I'm thinking about something that's like, you know, I want zucchini or I wanna make my own cheese or whatever, normally I'd have to flesh out all the edges of that idea to even get to a d- like a recipe or delivery. With AI, it was a lot easier to be like, "Okay, I have 15 ideas want you to round out the edges of them, and we're gonna work together for that. it's still my idea, it's my creativity, but you're gonna help me fill out the edges so I can actually make this into a consumable pizza."

[00:14:15] Whitney Lee: Mm-hmm.

[00:14:15] Michael Forrester: I did with Lancer, is I had the Lancer base and then I added my pieces on t- on top of

[00:14:19] Whitney Lee: So you're making like the characters and you're making the

[00:14:22] Coté: Right, right. Like, like, like, uh, like imagine if, uh, everyone knows about Monopoly the game, right? Like, like although I don't know, I wouldn't be surprised if Whitney was like, "No, never played it." That, that would be amazing.

[00:14:32] Whitney Lee: I was raised under a rock

[00:14:35] Coté: But,

[00:14:36] Whitney Lee: it

[00:14:36] Coté: but, but, but you could imagine if, if the, uh, if the task was like, "I lo- we've been playing Monopoly for three years, and now Dad, what I wanna do is like let's keep playing Monopoly, but imagine if before you buy a house or a hotel, you go through a sub-game where you've gotta hire some contractors, get an architect, and figure out like do the layout of the house that you wanna build, and then we're gonna roll a bunch of dice for like three hours, and then we may or may not have a house at the end."

[00:15:03] And, and like and, and, and, and then he's like, "So I mean, we're gonna need rules for that that involve dice." And, and then so it builds all that out for you.

[00:15:14] Whitney Lee: Yeah. Going back to the pizza analogy, what I'm actually imagining is like a cookie pizza or like Lancer is the idea of

[00:15:21] Coté: Hmm

[00:15:21] Whitney Lee: like I'm making up what, what is crust for me? What is cheese when it's supposed to be sweet in a cookie? Like, what is like the quote... Yeah, the pepperoni looking thing in my world, in my cookie pizza world.

[00:15:33] Coté: Yeah, yeah

[00:15:33] Michael Forrester: you're borrowing the shape of the thing and

[00:15:35] Whitney Lee: Mm-hmm.

[00:15:36] Michael Forrester: thing, but the substance is completely different.

[00:15:38] Coté: Yeah, yeah, that's, that's a good way of thinking about it. Yeah, I like it

[00:15:41] Michael Forrester: Yeah.

[00:15:42] Whitney Lee: Uh,

[00:15:42] Michael Forrester: So

[00:15:43] Whitney Lee: yeah. So AI enabled.

[00:15:45] Michael Forrester: was possible,

[00:15:46] Whitney Lee: again, this is just like Coté at the beginning, this is something that you just would plain would not have done

[00:15:51] Coté: Yeah, yeah, yeah. Yeah

[00:15:53] Whitney Lee: do it

[00:15:53] Michael Forrester: I

[00:15:54] Whitney Lee: with AI.

[00:15:54] Michael Forrester: have attempted it, honestly, 'cause I knew from doing it when I was young, you know, tw- 10, 20 years ago, I knew the amount of effort it typically took. so I didn't-- And I, and honestly, I did it during COVID, so I actually knew. I ran a massive campaign, Dungeons & Dragons campaign, and so I had that frame of reference, and so I knew it took 60 to 100 hours to get it to a, a deliverable state. I wouldn't have even attempted it 'cause it was like 60 or 100 hours. Like, where am I gonna get that from? That's gonna take six months to even get to. I would argue that it probably took about 30 days or less, honestly, to get to something deliverable. And

[00:16:29] Whitney Lee: That,

[00:16:29] Michael Forrester: I

[00:16:30] Whitney Lee: that's just as in many hours.

[00:16:32] Michael Forrester: Say again?

[00:16:33] Whitney Lee: he just compared 120 hours to 30 days. I mean, that's the same, isn't it?

[00:16:38] Michael Forrester: Yeah, but it was probably like five hours a week, maybe,

[00:16:40] Whitney Lee: okay

[00:16:41] Michael Forrester: but probably m- actually more like 20 hours to get to something consumable. And not only that, when I say consumable, way more consumable in that a campaign would be basically writing down a bunch of documentation. is what the pizza looks like, here's the cheese, here's the character, you know, like here's the monsters, here's the storyline, here's the arc, all that stuff.

[00:17:00] And all of that would just be written down, and then you would pull all the players into a table, virtually or physically, and you would set the frame for them, give them the story, help them develop their characters, and then you would journey through that whole process. When I say it took 20 hours, I mean not only did it take 20 hours to set up the frame, the context, the world, the world-building, the characters, I have a website, I have more documentation than I've ever seen in my ent- any campaign ever.

[00:17:30] I have a virtual tabletop board. I have Discord bots integrated with the thing in question. That's what 20 hours got me. Not

[00:17:38] Whitney Lee: Yeah

[00:17:38] Michael Forrester: the setup, but more setup than I've ever had.

[00:17:40] Coté: It was turnkey

[00:17:42] Michael Forrester: Yes

[00:17:42] Whitney Lee: Well, and I think it's worth saying here, uh, Michael and I are friends, so I know that you have a really beefy AI build system where it's building all the time and you have it running on a NetCup, if I remember correctly. And so when you say, like, you did that in 30 hours, I don't think that's normal.

[00:18:01] Michael Forrester: Oh yeah. You,

[00:18:03] Whitney Lee: I mean, I don't know other people, but yeah

[00:18:06] Michael Forrester: people aren't paying for a 16-core, 64 gig RAM system that runs 24/7 to do Cla- Claude Code over 50 projects. That's

[00:18:14] Whitney Lee: Yes. That's-- I don't think so. I don't know.

[00:18:17] Michael Forrester: don't have one of those in your back pocket running, you know?

[00:18:20] Coté: No, no, no. Unfortunately not. It's, it's, uh, in the shop

[00:18:26] Michael Forrester: Your computer's in the shop. I w- I will say that I was at a co- uh, AI Engineer World's Fair when you and I were co-speaking actually at the, at an event, and, uh, uh, there was one moment where I was in a talk and they were showing, I think it was the keynote, they were showing people's laptops like folded down. Like, they were almost closed but not completely closed because they didn't wanna interrupt their Claude code sessions. So Coté, just so you're aware in reference to what, uh, Whitney just referred to as far as my build system, all of my Claude code sessions run locally on any device that I'm on, but they're auto SSH'd and Tmuxed, so they're like screened, they're

[00:19:02] Coté: Sure, sure.

[00:19:03] Michael Forrester: over on a remote server.

[00:19:05] So if I close my screen, it just keeps running on the remote server

[00:19:08] Coté: Yeah, yeah

[00:19:09] Michael Forrester: I'm on. So that way also I have remote control on my phone the entire time.

[00:19:13] Coté: Yeah

[00:19:14] Michael Forrester: I'm at in the world, I can actually go and look at my sessions and ki- kick them off or keep them going. So, uh, when I, I, I think it's, it is worth stating, Whitney, that, that is a disparity that I have a build system that can just run continuously that I can monitor from my phone or any device I have, so

[00:19:31] Whitney Lee: So what I was getting at with the line of questioning around AI was, was kind of trying to, um, AI and D&D or Lancer, I was trying to get at like, is the creativity outsourced? And it sounds like you're using just as much creativity as you would have without it, but just at a different level, at a different scope.

[00:19:51] Does that feel accurate? And then can we extrapolate that to

[00:19:57] Coté: Yeah, yeah. I, I mean, I think, I think, I think, w- I, I, I agree, I agree with, like, the, the, the, the experi- I agree with the experience Michael had. But the, it's to, to, to, to phrase it, like, there is It, it'll, it'll, it'll... In, in the D&-- in the role-playing game, in the D- I'll just say D&D, but in the role-playing game world, uh, there is like an infinite amount of things you could be doing and work you could do.

[00:20:26] And so what you can choose is you can, you can prioritize, like what work you wanna do based on the time that you have. So like, you know, I don't, I don't know the stories, but just like in the software world, there's stories like this, but I imagine in the role-playing game world, there's like, "Oh, there's Jerry.

[00:20:43] They've been working on a new game for 15 years, and they never actually show it to anyone," right? And that's probably because Jerry there is infinitely tinkering and doing things. And so in that case, I forget what they call this, but there's another phenomena in, in role-playing games that's essentially like you play the game by preparing the game.

[00:21:04] Like you never actually get around to like playing the game. You're spending all your time like world-building and doing this stuff. I don't even know if it has a name, but it definitely has a criticism. Like people will be like, "Oh, you're spending too much time on that. You should be playing instead." Um,

[00:21:18] Whitney Lee: Uh-huh

[00:21:18] Coté: and so I think, I think the point of, of AI usage, uh, is like you can pick the things that you don't necessarily wanna spend time on and then refocus on the things you do wanna spend time on.

[00:21:32] So to put it in reverse, like I haven't actually done this, but it would be fun to like-- Well, it would be fun. Like, so you would think the point of, of a role-playing game is to play the game, right? To be the, um, uh, figuring out if you wanna make a, a thousand-year-old dragon your pet or not, being a player, right?

[00:21:50] Whitney Lee: Mm-hmm.

[00:21:51] Coté: But you could also imagine a way of playing it, just like I was saying, where you like to do the world-building, and so you have the AI be the player. And you're like, "I like to build all the worlds, but I can never fi- but I don't wanna find actual humans to play with me, so I'm gonna set up AIs to be the players and figure out like interacting with it."

[00:22:08] So I think you just like choose what part you wanna automate. And I don't know. There's like more nuance, uh, to, to a lot of the things, but it allows you to, uh, pick which creativity you wanna do. And I mean, I guess there's a bit of the nuance. You can, you can not only-- You, you can not, not only categorically choose the, the creativity you want, like, uh, "I don't wanna specify what this adventure is."

[00:22:36] Uh, like you, you could also just like pick and choose what you wanna do. Like, uh, I think, I think section five of the adventure that we have is really boring, so I don't wanna think about that part, whereas I wanna do section seven or whatever. So I don't know. It's, it's-- There's not a universal like sucking out of joy that the AI does.

[00:22:55] It just like... It just, it...

[00:22:58] Michael Forrester: wait, there's a non-universal sucking out of joy?

[00:23:00] Coté: It's, it's just, it's just more, uh, it's more targeted. You, you can, you can direct, like you can say, "This isn't my joy, so suck away at it, and, uh, I wanna get to the other stuff."

[00:23:11] Whitney Lee: pros and cons of that. Like, you're not learning, you're not pushing yourself to learn stuff outside of your comfort zone, or you're not teaming up with someone who has compli- like

[00:23:18] Coté: That's--

[00:23:19] Whitney Lee: whose,

[00:23:20] Coté: yeah, yeah. I m- I mean, that, that is, that is the-- Like, I think, I think when you're using AI for entertainment or, or for playing games, a lot of the things that, you know, when you go see presentations about using AI for work and enterprise that are problems are actually benefits, right? Like, so, so if, if, if, if you think about, like, like, it hallucinates all the time, it's inconsistent, blah, blah, blah.

[00:23:47] It's like, yes, that's exactly what I want. Like, even, even in D&D, like, you can decide to not follow the rules if it's cool, right? Like, so there's this, like, there's this phrase, the rule of cool, which is like, "Yeah, you're not supposed to do that, but it'd be pretty cool if you could," right? And so, so if, if you can, like, if you can portray something as cool, then it's kind of like that's the point.

[00:24:13] You're s- the point is to have fun. So, like, whereas, you know, in, in the world of invoices, the point isn't to have fun. And so, like, that, that is,

[00:24:22] Michael Forrester: invoices

[00:24:23] Coté: that is an overlap that doesn't quite apply. And indeed, I think, indeed, I think introduces some of, to your point, the creativity stuff is... I can't think of an example off the top of my head, but there are, there are moments when you're using AI in role-playing where something interesting and new happens that you wouldn't have really thought would happen, probably due to the non-deterministic nature of, of, of AI and all of its limitations.

[00:24:50] Whitney Lee: So, so then that, my question is AI, and this is probably for Michael, if AI's building the whole world, I mean, you can have it do everything. So you can have it build the characters and each mission and every mechanism for how things are, what dice rolls mean what or, I don't...

[00:25:09] Michael Forrester: Mm-hmm.

[00:25:10] Coté: The rules.

[00:25:11] Whitney Lee: don't know.

[00:25:12] Coté: Yeah, yeah

[00:25:13] Whitney Lee: Okay, so like porting, so if, so you could technically have it make the game and play the game and never be a part of it at all, right? Like, it's over there doing something really cool.

[00:25:23] Michael Forrester: Yes

[00:25:24] Whitney Lee: that not what enterprises seem to be working toward in terms of software building?

[00:25:32] Michael Forrester: Yes. But like I would say this is that like, at least for, for me, I, I think the only time I would play solo is if I wanted the experience and I d- I couldn't, I couldn't arrange my friends or I couldn't get access to people to do the thing in question, and I was just dying to have a creative experience.

[00:25:49] So I

[00:25:49] Whitney Lee: Mm-hmm.

[00:25:50] Michael Forrester: with Coté that oddly enough, the hallucinations and the making up and the kind of variability around probability, like around the outputs for AI, is actually a net positive when it comes to like having a solo experience, because then you get the same randomness that you would get with people.

[00:26:03] Which by the way, I think is an interesting point, by the way,

[00:26:06] Whitney Lee: Mm-hmm.

[00:26:06] Coté: Yeah

[00:26:07] Michael Forrester: AI to a much different standard than the same people who, by the way, also hallucinate and make stuff up and misremember and whatever. But, you know, let's, let's go back to your point about isn't that what enterprises are trying to

[00:26:17] Whitney Lee: We treat AI like software when we should treat it like we treat people

[00:26:21] Michael Forrester: Honestly, truly, right?

[00:26:23] Whitney Lee: Mm-hmm

[00:26:23] Michael Forrester: and, you know, like you, you and I gave a talk basically at AI Engineer World's Day about that whole thing, about having guardrails because you wouldn't give a junior developer unfettered access. So framing it like that instead of deterministic software, right? And so I think, a- and especially because I talk to multiple C-levels every week, I think your frame about, oh, aren't we chasing that level of independent autonomy where it just kind of goes and does the thing, is

[00:26:48] Whitney Lee: Yeah

[00:26:49] Michael Forrester: a sense, and I'm gonna use a phrase like the Holy Grail, the Ambrosia, the secret treasure that

[00:26:55] Whitney Lee: Mm-hmm.

[00:26:56] Michael Forrester: to be chasing from enterprise. And quite frankly, it is the riskiest and wildest idea that I've ever seen, because we wouldn't expect that of people, and yet we expect it of this interesting probabilistic word calculator.

[00:27:09] Whitney Lee: Mm-hmm.

[00:27:11] Michael Forrester: you know, we're like trying to put guardrails around it to kind of like have it kind of do autonomous outputs and whatever, but we would never allow people to go autonomous, honestly, and do their own things without the same guardrails.

[00:27:20] So there's kind of an interesting kind of sidebar around enterprise perspectives around autonomy that don't exist in the entertainment world. But-- And we love the, like we love the randomness in the entertainment world, but we don't love the randomness in the enterprise world. And yet the perspectives, like Coté, you are like, "I love that it's random." But

[00:27:39] Whitney Lee: S-

[00:27:39] Michael Forrester: we do that for the enterprise, we're just like, "No, I, I don't want

[00:27:41] Coté: Yeah, it's not-- And, and I think, I think also, like-- So first, I think that's, that's a great, great point is that like, uh, yeah, if, if, if you had, if you had agentic AI in like playing D&D, then it's like, uh, it's, it's, it's kinda, it's kind of like, uh, I like to summarize a lot of things, but there's some podcasts I don't summarize because actually listening to it is the point.

[00:28:02] It, it's like enjoyable. And, but that's right. But, but I, I think, I think the, um, the, the, the follow-up to the joke of like, you know, if, if the robot can play D&D, it can play ERP or whatever, is that like it can't really play D&D without a lot of human intervention. Like it's actually-- You can't just like load up, just go to Wi- load up a chat session and be like, "Let's play D&D," and even like give it like your, your character, uh, and even give it like, "I wanna play this kind of adventure."

[00:28:33] It just like, that won't go well at all, right? And there's so much like, in the same way, uh, that with-- that you need a harness to do anything... Anything's the wrong word. To do intermediate to advanced things with AI, like you need a harness to play D&D with AI. And even then, it's not great, right? Like even then it takes a lot of human, uh, centauring or in or out of the loop, whatever loopiness that you want, like humans still need to be involved

[00:29:04] Michael Forrester: So when you say harness, by the way, you're talking about like an application harness, like an, an application wrapped around

[00:29:09] Coté: yeah. I mean, and then, and then even like a stable and you might-- And you like everything. Like you need all the stuff behind it. Like, you know, you think of, as you were saying, like the... You could imagine if you stacked up all of the papers from a three, three years of playing a campaign, and then all the stuff in your head, like that would take up a room full of s- not maybe a roo- maybe a closet.

[00:29:33] Uh, but it would take up a, a European room full of, of, uh, of space. And, and so there's, there's like it's, it's not only like all of the little, you know, your little markdown files and scripts and stuff that you need and instructions on how to do it, but it's also just like all of the, the data and the memory, I don't know, on and on and on.

[00:29:51] Like there's all this stuff that you need to have like beyond a very basic short-lived D- short-lived D&D thing. And I think, you know, you, you, one, or I should say I follow and read all the stuff that's g- goes on in the, the more enterprise-y space, and it's the same thing, right? Like you need all that same stuff for it to actually do anything.

[00:30:11] Um, so that's what I got.

[00:30:14] Michael Forrester: Yeah. Yeah, yeah. I, I, I will say this, I, I re- I-- and Whitney and Coté, I actually researched this before I started the Lancer D&D thing. and by the way, my general process with AI is before I actually do anything, I just research. What is the community doing? What does everybody do using AI? What did we do before AI?

[00:30:31] Like, I usually do a bunch of research spikes to find out background. One thing I did is I said, "Surely," you know, 'cause I got to this only a couple months ago. I, I thought to myself, "Surely other people are doing things with AI, that, like, there's stuff out there, like people are already into this." 'Cause I've been using it for enterprise development and other pieces, and so I wasn't using it for leisure. But I thought, "Someone's probably already solved these problems. Let me not reinvent the wheel."

[00:30:53] Coté: Mm.

[00:30:53] Michael Forrester: So one thing that clearly came out as a consensus across all research, whether it was ChatGPT or Claude or whatever, Gemini, um, it was, by the way, don't let it do the math. point,

[00:31:06] Coté: Right, right, right

[00:31:07] Michael Forrester: Coté. It said get an application

[00:31:10] Coté: That's right

[00:31:11] Michael Forrester: the model, and make sure that all of the math is as deterministic as you g- like, as, like calculator, pure code, do not use LLM as a judge.

[00:31:20] Like, it was just like, sure that all of the numbers are not being done by the LLM. It can interpret the, the numbers, it can pull them in, it can do the result responses. But like d- if, if you need to roll a to hit, the math, roll the number, use some random generator if you can, make sure that you're doing the comparisons, has a mathematical differential like equation.

[00:31:43] Like, don't let the LLM be like, "Okay, you rolled a You needed a 12 to hit, um, but I'm gonna subtract this and do this." Because it, it was basically saying that if you allow LLM to do math for Dungeon & Dragons,

[00:31:55] Coté: Yeah, it's terrible.

[00:31:56] Michael Forrester: times it's terrible. It's just gonna

[00:31:58] Coté: Yeah

[00:31:58] Michael Forrester: and you're gonna be like, "Hold on."

[00:32:00] And so like even last night, so last night I ran a campaign this space sci-fi mech Lancer thing that I was talking about. It's called, uh, Antiphon, by the way. Antiphon, like, which is the Lancer game, we were running just a space combat mission. Like, they just had like a bunch of like spaceships, and they were just shooting at each other because we couldn't run the normal campaign for reasons. So five of the ships, and by the way, the ships are all based on marine creatures, so Whitney, to your earlier thing of like, would you ever outsource creativity to the, to the AI? Just so you know, at some point going back and forth with AI 'cause I was using a spec file for this one, the AI was like, "You know what we could do? We could just actually base all of your spaceships on sea creatures." And I was like, that's, a great idea." But I don't wanna do the toil of finding every one of the sea creatures. So what I'm gonna do is I'm gonna rattle off seven sea creatures that I absolutely love, like the pistol shrimp, the mantis shrimp, the anglerfish.

[00:32:58] Like, I'm gonna rattle off some, and then I want you to do the remaining 20, and it did an excellent job. I can actually pull up my board here and probably describe it visually

[00:33:05] Whitney Lee: Is it a public, uh, is there a public website?

[00:33:09] Michael Forrester: Yes.

[00:33:10] Whitney Lee: Okay, we'll...

[00:33:11] Michael Forrester: The Game." I can

[00:33:11] Whitney Lee: Okay, we'll put it in the show notes

[00:33:13] Michael Forrester: Yeah. And so "Antiphon: The Game," and so anyway, so they were playing last night, and by the way, the whales, 'cause, you know, we're based on sea creatures,

[00:33:21] Whitney Lee: Mm-hmm.

[00:33:22] Michael Forrester: battleships have broadsides, like it's an 1800s, like battle- like wooden battleship.

[00:33:27] They

[00:33:27] Coté: Hmm.

[00:33:28] Michael Forrester: so they can shoot each other, right? 'Cause, you know, k- uh, part of, part of "Dungeons & Dragons" is combat, exploration, and role play. So the combat piece is actually a part of it. Anyway, they're shooting broadsides and they keep doing ones. And I'm like, "Why?" Y- why... Okay, so the computer's doing rolls and that roll is supposed to be deterministic. Why is it only doing one damage? These are huge battleships space with huge big guns on them, and they're doing one damage to each other. And so at some point last night, and by the way, this is the awesomeness of AI as well, I'm like, "Huh, why are we rolling ones?" And so they started getting frustrated.

[00:34:05] They were like, "We're just doing one damage." So they did like 20 rounds and they're just doing one damage to each other. This doesn't make any sense. And by the way, these whales have 100 hit points, so one damage means we're gonna sit here for

[00:34:15] Whitney Lee: F-

[00:34:15] Michael Forrester: hours, right? This is not... They, they don't want that. They want action.

[00:34:19] They want things to catch on fire. They want, you know, they want drama. So I'm literally talking to Claude, who is hooked into the system, and I said, "Hey, I'm doing one damage." And instantly it's like, "Oh, that's because I'm doing the math." I'm like, "Why are you doing the math?" And it's like, "Oh, because I decided to ignore the harness and I'm doing the math myself."

[00:34:37] And I was like, "Oh, so can we get back to letting the harness tell you?" And the next thing you know, all the rolls evened out. They're doing like 20 damage to each other. They're like super happy. They're like, "Woo! I'm on fire and I'm dying." And it's like, okay, cool. We're back to this. But to your point, it slipped out of using the application harness and it was stepping itself in, and it was just horrible at the math, and we would've been there for, I don't know, hours for them to kind of

[00:34:59] Coté: Yeah, yeah. Yeah, I, I mean, that's, that's, that's another good example of, like, the, the, the, I don't know, elf to enterprise spectrum, is like you, you like all-- If, if, if, if you play role-playing games with AI, you can encounter all the problems you'll have in the enterprise space before you mess up an invoice.

[00:35:18] And, you know,

[00:35:18] Michael Forrester: them

[00:35:19] Coté: that's, that's like a, that's like one that you generalize pretty quickly, is just, like, you really should verify things. Like, obviously you can't-- You, you don't wanna put in, like, a hook to verify every single turn because that's gonna be resource intensive and take time. But, like, every now and then you should probably be like, "I don't think you're doing something.

[00:35:36] That y- I don't think you're actually rolling dice." And then, as I'm sure you know, and, and as you find in other AI stuff, immediately be like, "Oh yeah, I totally wasn't doing that." And you're just like, "Well, what the fuck?" But the, it's, uh... Yeah, like, like I think, I think all the, that layer of harnessy stuff you end up encountering in, uh, in, in, I mean, pretty quickly when you're, when you're playing games with it

[00:35:58] Michael Forrester: Agreed. Yeah. And, and I wanna say this also is that, is that the gut check was instant. minute it started showing ones, both of the boys were like, "Hey, we've got two rounds of ones. What's going on?"

[00:36:14] Whitney Lee: Mm-hmm.

[00:36:14] Michael Forrester: instantly they were like, "Something doesn't-- It doesn't

[00:36:16] Coté: Yeah, yeah. That, that doesn't happen

[00:36:18] Michael Forrester: what's funny about that is that them being experienced rollers, like they had done it enough manually to know that a random roll doesn't elicit this many ones with this one damage. I thought that was a great example, by the way, of kind of the intuition and the judgment side of experience when we do like human in the loop pairing with AI. thought

[00:36:39] Whitney Lee: Mm-hmm.

[00:36:40] Michael Forrester: Like I-- Like just as an example, I have to give a talk to the Air Force, my first ever in like, I don't know, 10 years. And I haven't really been

[00:36:47] Whitney Lee: First ever what?

[00:36:48] Michael Forrester: talk for the Air Force. I, I haven't done

[00:36:50] Whitney Lee: A-

[00:36:50] Michael Forrester: in like 10,

[00:36:50] Whitney Lee: Air Force spec- specific talk. Okay.

[00:36:52] Michael Forrester: have to give a talk about

[00:36:54] Whitney Lee: Okay

[00:36:55] Michael Forrester: I don't know if you're aware of this, but as I've studied the doctrine of all this, and this is totally relevant for what we're talking about by whatnot, is that the Air Force has this perspective that says, we want the speed, the precision that the, that AI can have, but we want the judgment, the intuition, and there was some other word that they use.

[00:37:14] We want the judg-- I thought the Air Force is using the word intuition, which for them is a subconscious synthesis of all of your experience and your ability to tell that something is off. And that usually leads to some curiosity where you do further discovery, and so that you can find out then like, is the AI on point or is it not?

[00:37:33] Like in this case, the boys were like, "Hey, we're doing one damage. There's been like four rounds of this. Something's off." And immediately I go in and like, "Hey, what's happening?" And immediately from a diagnostic perspective, it's like, "Hey, this doesn't work." To me, I thought that was a great example to your point, Coté, of like running into like all the enterprise stuff in this simulcrum, is like that's exactly what happened, is that immediately human in the loop caught some kind of sniff test that said, "Hmm, this isn't, this doesn't feel right."

[00:38:00] Further discovery yielded the fact that there was an actual problem. And then now, by the way, I have it check every fifth-- every five rounds, it checks to see if it's sourcing from the harness

[00:38:10] Coté: Mm.

[00:38:11] Michael Forrester: if the LLM has actually changed the game in some form or fashion. So to your point, now I've got checks,

[00:38:17] Whitney Lee: So, so earlier you said we, we talked about like having an entire D&D where AI built it, AI's playing it, AI's doing all the things except the math. Maybe we built harnesses so that like ... We talked about this being like the enterprise holy grail where, where it's AI's doing everything, no humans are needed anymore, all the software is getting built as fast as people can dream it up. Um, or AI dreams it up apparently. Um, uh, if that And would you-- You kinda said it was dangerous and did you say impossible? Or like what were the words you used to describe a fully, fully 100 AI? Yeah. Y-

[00:38:54] Michael Forrester: Was that like past, past Michael talking or was that past Coté

[00:38:56] Whitney Lee: yeah, yeah, no, you in the, during this conversation. Um, regardless, my question is, if it's not, if it, if it shouldn't be doing all of that, what parts of it is it good at?

[00:39:07] Like what, which parts should it be doing? And like what's impactful if you're an org wanting AI transformation, what are the most impactful things you can do in the short term in terms of incorporating AI?

[00:39:19] Michael Forrester: Well, I mean, one, you know, like people always talk to me about w- like how do we get people familiar with AI, right?

[00:39:25] Whitney Lee: Mm-hmm.

[00:39:25] Michael Forrester: I think this conversation, right, both the, the questions and the perspectives that you're asking and also like kinda the whole thing is representative. Just play with it. Play with it

[00:39:35] Whitney Lee: Mm-hmm.

[00:39:35] Michael Forrester: stakes environment. That, that-- I think that's the idea. Because I think all the lessons that I've y- and, and I used it primarily in an enterprise environment before I got to gamification, but all the gamification lessons have taught me more about how to fail and avoid that failure in a production environment than anything I did in the enterprise, because in the enterprise risk. could,

[00:39:57] Whitney Lee: Mm-hmm.

[00:39:57] Michael Forrester: I couldn't do a bunch of things. But like in a Dungeons and Dragons environment, I can sit here and do everything I want, explore all the failure edges and modes, and if I'm paying attention, I can take that back to the enterprise. So honestly, I think playing with it, like that's just like the first thing, is just play with it as much as you can. met a bunch of people who unfortunately think that AI is a fad and that it's gonna go away, I would love for that to happen on some level. Although I will say that

[00:40:21] Whitney Lee: Okay

[00:40:22] Michael Forrester: has changed a bunch of things, and that's just not gonna happen. Silicon Valley is so hot and hungry for the, for the fantasy of gen AI, right?

[00:40:31] Whitney Lee: Mm-hmm.

[00:40:32] Michael Forrester: there's no way they're gonna let this go, It's a, it's like there was a study, the meta study, and I, I, and I know like Whitney, you and I have had conversations about it. Coté, unfortunately we haven't had face time to have the conversations. But like there was that meta study, like the threat modeling study that where developers thought they were 20% faster and they were actually like 20% slower, and they tried to recreate the study in 2026 and they couldn't because developers refused to do the toil work without AI. So they couldn't create a control group because the minute they were like, "Hey, can you just go in and alphabet, like list these files like manually or whatever," the developers were like, "No, I'm just gonna get AI to do it. Why would I do that manually?" And so they couldn't create a con-control group to do the 2026 version because they couldn't get, uh, the developers to not use AI for the stuff they found boring.

[00:41:18] Coté: The-- You know, that, that, that, that makes, that makes me think of... Well, at the moment, here, here's my best theory for things. And,

[00:41:27] Michael Forrester: Uh-oh, here it is. Let's hear it

[00:41:29] Coté: uh, and, and, and, and, and I say, I say theory because, like, I don't, I don't know. But, but it seems-- It's, it's like seeing a bunch of one- ones rolled for damage. Like, it just seems like something is like-- It, it seems like, in my experience and, and what other people-- what I hear other people say is, like, an individual gets a lot of value out of AI.

[00:41:49] But thus far, organizations have found it near impossible to get value out of it in aggregate. Like, whatever the rolling up you do from an individual, maybe even a team, to, like, the organizational level, like, I don't know if that ever really happens. Like, you can go, you know, every year for, um... I get the Google conferences mixed up.

[00:42:10] I don't know if it's I/O or Next or whatever.

[00:42:12] Michael Forrester: Yes.

[00:42:13] Coté: But, but e- every year when they've had it for the last few years, they have, they have this one great-- They have many blog posts, many, many blog posts. But they have one that I like the most, which is, like, "Here's 500 ways people have been using AI with Google," right?

[00:42:25] And they seem cool, but they're always kind of small, little individual things that... So, so there's that. But then you go talk to an individual, like developers who refuse to give up AI, so you can't find a control group. And clearly, the individuals are like, "I think this is fantastic." And then on the upper end, you've got the executives who are like, "I'm not sure if GPT is spelled with a T or a G, but we should be using it."

[00:42:57] And, and, and, like, and, like, they think it's great. But then, like, every other week, there's, like, this schizophrenic, like, one week Bain says AI is great. The next week, McKinsey says it's not great. MIT says 90% of it doesn't work. And then that does- that's not valid anymore. And then, you know, it just, like, flops back and forth.

[00:43:16] There's no organizational slam dunk for enterprise ROI. And so I don't know. It seems like maybe it doesn't work at the organization level? Like, it, it's weird.

[00:43:28] Michael Forrester: Well, okay. So I'm, I-- and I, by the way, Whitney, I haven't forgotten your original question, like what's,

[00:43:32] Whitney Lee: Okay.

[00:43:33] Michael Forrester: for enterprises

[00:43:34] Whitney Lee: Yeah

[00:43:35] Michael Forrester: And so I think this framing is excellent, and I think it speaks to the whole gamification piece really well, right? have been talking about individual workflows, individual projects in like isolated, like I, I want to get to an outcome like kind of workflows.

[00:43:51] And like the outcome might be I wanna play Dungeons & Dragons by myself, or I wanna run a virtual tabletop campaign for my son, or whatever it is. Like I wanted to recreate Zork, if I don't know if anybody remembers Zork from

[00:44:01] Whitney Lee: I love Zork, yeah

[00:44:03] Michael Forrester: who's, who doesn't love being eaten by a Gru? And so, and I think that that framing is no different.

[00:44:10] I think when we say, "Hey, what's good for an organization?" Coté, I think that you're like, you're, y- I think you are espousing the 100% perspective that I hear from C-levels all the time. I mean, I'm talking like Fortune 50, 100. They're like, "We're gonna do like, "Okay, cool." And like, "We're gonna do it now."

[00:44:26] "Okay, cool." And like, "We, we've gotta do it." "Okay, great. Um, where would you like to start?" Like, "All of it." And I'm like, "Well, hold on."

[00:44:32] Coté: Exactly

[00:44:33] Michael Forrester: how many companies that are like, "We're doing AI." And they, like some companies, and I'm not g- I'm 100% obviously not gonna name names, have given Claude code unfettered with no security restrictions and no cost controls and no oversight to all of their

[00:44:45] Coté: Ooh, I want to go work there. That'd be fantastic.

[00:44:48] Michael Forrester: and I'm like, "How come you guys aren't in the news?"

[00:44:50] Like, and by the way, I probably just gave it away. There's no-- none of these companies have shown up in the news. They've just kind of managed it themselves, but they figured out pretty quickly that that was a mistake, right? And I think there's this linearization of AI as the magic pill from a, from an executive kind of investor perspective, like this is gonna happen. the, the truth on the ground is, I think, reflective, Coté, of what you're talking about, and I think this answers W-uh, Whitney, your specific question, which I think is gold, it actually comes down to every single individual workflow, like seven questions that you can ask, like, could a human do it cheaper?

[00:45:23] Is this a bound system? Does it have-- There's like all these questions you can ask. One of the questions that you ask is b- is basically, is it, is it automatable? Like, is it truly automatable, or does it, are there portions of it that need to stay manual so you can inject AI in it? So there's seven questions that we typically ask. But the reason I mention that is that people are chasing this holy grail of automation, and I don't think it works for teams. I think it works for a workflow. And if the team works in that workflow, I think AI can help you. the one thing that everybody seems to be chasing right now is, "I want AI to make changes to a workflow." And I'm just gonna frame this up right now in, in j- you know, August 2026, doesn't seem to be working too well

[00:46:05] Coté: Hmm

[00:46:05] Michael Forrester: enterprises, and that's the part where people are failing. But if you said, "Hey, I want to create an agent that will read all of my log files, all of l- my telemetry..." Oh, wow, I feel like we're kind of leaning towards Whitney now, because as, as you

[00:46:17] Whitney Lee: Mm.

[00:46:17] Michael Forrester: she works for Datadog,

[00:46:18] Whitney Lee: Yeah

[00:46:19] Michael Forrester: I want to read all of my enterprise data because it's too big for a human being to read, and we're gonna miss it anyway, and it's too much alert fatigue, and it's too much information fatigue. kind of perspective, which I think honestly, just to say this, and I'm gonna say this, Whitney's not saying this, Datadog is actually a great example of a company that's doing agentic really well

[00:46:38] Whitney Lee: Yeah

[00:46:39] Michael Forrester: actually using a lot of it for read-only ingestion and toil that we can't do en masse.

[00:46:44] Now, are there things possibly dropped? I'm not saying Datadog does that, but, like, you can lose information in there. But a human would lose that information anyway. Like,

[00:46:55] Whitney Lee: Mm-hmm.

[00:46:55] Michael Forrester: do the same thing. If you gave it 1,000 streams of information, I would drop a bunch of them.

[00:47:00] Whitney Lee: Mm-hmm.

[00:47:01] Michael Forrester: no risk or cost for us to just replace that with, uh, several agents that actually may be more consistent, plus deterministic checks and evaluations to check that. And so to answer your question is that I, I th- I don't think it's a benefit for organizations. I think it's a benefit for individual workflows inside of an organization. And I think the holy grail right now a multi-agentic thing that d- like reacts to changes and makes like risky, like updates code and pieces like that. think one that reads an en masse system that's too large for us, which could be code, by the way.

[00:47:37] Whitney Lee: Mm-hmm.

[00:47:38] Michael Forrester: could be log files, it could be traces, it could be metrics. think that's where most organizations ideally would be pointed now.

[00:47:45] Coté: Mm-hmm.

[00:47:45] Michael Forrester: But also I just want to point out that almost everything I mentioned is a bound system. Dragons has a finite set of rules. Lancer has a finite set of rules. Zork has a finite set of rules. Code creation has a finite output and a finite set of rules.

[00:48:01] Whitney Lee: I

[00:48:02] Michael Forrester: Reading infrastructure is the same way

[00:48:04] Whitney Lee: I don't understand. Uh, instances can only understand so much context at once.

[00:48:12] Michael Forrester: Mm-hmm.

[00:48:12] Whitney Lee: So when you say its value is to read all the logs or all the everything of something, I don't understand. at-- That's not one instance. Is it several instances? Are you scoping what one AI instance can do?

[00:48:27] Michael Forrester: Uh,

[00:48:27] Whitney Lee: R- and

[00:48:28] Michael Forrester: say that context clearing is probably the most important thing, like managing context,

[00:48:32] Whitney Lee: Mm-hmm.

[00:48:32] Michael Forrester: for a particular agent. But just like we can with software processes where we can scope and parallelize, right? As far as like running threads and ingestion connections and whatnot, same thing can be done with agents.

[00:48:43] So if you need to ingest a bunch of information, you can either serialize it and just reset context every time, or you can run a bunch of agents with like very small context windows that are spawned for a particular finite ingestion point.

[00:48:55] Whitney Lee: Uh-huh.

[00:48:55] Michael Forrester: this out is that when it comes to things that can be done in code and the agent just invokes the tool and then perhaps reads the aggregate or does some kind of management, that's ideally what we're doing, is that we're still actually leaning on deterministic application wrappers and deterministic tools and determi- even like MCP servers to kind of do the actual heavy lifting and the work that can be done deterministically.

[00:49:21] Whitney Lee: Uh-huh

[00:49:21] Michael Forrester: greatest value is the ability for us to create code and documents and artifacts that then can be verified, tested, and reused, and not in the pure like let's get the LLM to do all the things of work

[00:49:37] Whitney Lee: When you said it's bad at workflows, what did you mean exactly? Bad at creating workflows? What did you mean? Like, what's an example of that?

[00:49:44] Michael Forrester: So i-in a sense, let's say the, a workflow has, in, is in a bound system and like I wanna-- Let's, let's you do invoicing, for example. Invoice comes in, invoice needs to be read, invoice needs to be categorized. There's all these steps that kind of happen with an invoice.

[00:49:59] Whitney Lee: Mm-hmm.

[00:49:59] Michael Forrester: invoice... Is there attribution?

[00:50:00] Like, does it have to be attributed and put into some kind of financial system? And then is the invoice already paid? Do we need to text that system again? Do we need to then pay the system? What is the process for that? So there could be like a 20-step process for that whole piece

[00:50:13] Coté: If everything goes well.

[00:50:15] Whitney Lee: Mm-hmm.

[00:50:16] Michael Forrester: Yes. And my, my, my sister, by the way, works for a dental company, and she's in invoice. She does, she does accounts receivable, and so I hear about this kind of stuff all the time, oddly enough. Unfortunately, there's no wizards being pickpocketed, but you know, I digress. And so what's interesting is that that whole kind of step flow should actually be done inside of an application harness, ideally. Like a Lang graph, which is, you know, like could be Python, TypeScript, whatever. That should be an application process. The part where someone needs to like read the invoice and interpret the meaning of the words, because for some reason we failed at the deterministic match of the company name to the list of companies that we have, and we're now we're falling back to an LLM that maybe can just read the words and maybe we're not using fuzzy matching or whatever. So maybe this is a third step. That to me is where the, the agent, like the LLM comes in, not in the known process. Why would we get some random Tasmanian devil agent is gonna deviate at any point in time off of a 20-step process? Why would we get it to manage a known process that

[00:51:20] Whitney Lee: But

[00:51:21] Michael Forrester: steps?

[00:51:21] Whitney Lee: wouldn't that be a 20 agent step pro- like each agent does one step?

[00:51:26] Michael Forrester: No, because, because the processes are very well known and the gates about where we go this way or this way are actually also relatively known,

[00:51:34] Whitney Lee: Uh-huh

[00:51:34] Michael Forrester: honestly still just be a deterministic application.

[00:51:37] Whitney Lee: Oh yeah, agreed.

[00:51:38] Michael Forrester: that we're running into is I

[00:51:39] Whitney Lee: Mm-hmm

[00:51:40] Michael Forrester: like, "Well, we don't have to write code anymore. We'll just get the agent to do the 20-step process, and we'll get the

[00:51:45] Whitney Lee: Hmm

[00:51:45] Michael Forrester: that we got if we did it in like .NET, Java or Python." And the

[00:51:49] Coté: Yeah, like, like I, I, I think, I think you're, you're, you're like go-- You're, you're explaining an- an- another thing that, that I f- I f- I feel like I feel a lot, so to speak, is,

[00:51:58] Whitney Lee: Mm-hmm.

[00:51:58] Coté: is, I don't know. All I can think of is that like, you know, someone invents ketchup and they wanna put it on everything, and you're like, "No, w- that doesn't go on everything."

[00:52:08] Ketchup is delicious except for those people who think it's not. But like, ketchup's great, but you don't put it on like noodles, n- unless you do. But like

[00:52:16] Michael Forrester: Wait, what's wrong

[00:52:17] Whitney Lee: And in the, in the Midwest we put it on mac and

[00:52:19] Coté: I, I, I know, I know. You, you know.

[00:52:23] Michael Forrester: okay, hold on. Can we just, can we just do ketchup on ice cream? I think ketchup on ice cream would be the universally

[00:52:28] Coté: There we go. There we go. The-- So, uh, and, and so

[00:52:32] Whitney Lee: did catch up

[00:52:32] Coté: a-analogously, right, like if, if you're at the, uh, let's just say the executive level to make a cartoon of it,

[00:52:38] Michael Forrester: Hmm

[00:52:39] Coté: like you're like, "Oh, let's just put AI in everything 'cause it'll speed everything up and make everything better," whatever better means, like faster, cheaper, et cetera.

[00:52:47] And like it seems like kind of what you're saying with the invoice thing, that, uh, it's difficult to get big results applying AI to existing processes because these existing processes, like you can't really just like improve the invoicing thing by like pointing Claude at it and being like, "Uh, make this 20% more KPI," and, and just like have it do it.

[00:53:14] And instead kind of, kind of as, as you were saying and as Whitney was, was also kind of saying like, well, there's actually a point in the invoice processing where, uh, you know, it doesn't know the difference between J-H and J hyphen H, and a human would be like, "This is the same thing. What's the problem?"

[00:53:32] And so like that's-- It, it's almost like you either have to pick and choose these little tiny thing-- Well, tiny diminishes it, but these narrow things that you're gonna apply an AI to, or you just like throw out the workflow and put an entirely new one in. And like analogous to that is like, I think, I don't know, it's foggy in my memory, but I think w- before DevOps existed, there, there was a build pipeline.

[00:53:59] There was like CruiseControl and Jenkins and Hudson and all of that. But like, but like a lot of work went into recrafting what an automatic build pipeline was based on DevOps and SRE stuff. Like that whole, that whole process as done in like 2003 doesn't exist anymore 'cause that workflow was completely redone.

[00:54:19] And so similarly, like it feels like, well, if you wanna spread, put ketchup on everything, you gotta redefine what the food is so that it actually like mixes well together. Like it seems like... And that's the kind of thing that I think at an executive level, it's not only jokingly like not within their comprehension to do, but it's also like not their job.

[00:54:37] Like th- they don't know about all these little stuff on the, on the floor to do. So I don't know. It seems like another thing that would contribute to like we don't have organizational AI ROI because like there's some mismatch between the two things. We're talking about like what paintings to have in the, uh, in, in the, the, the lobby instead of like what robots to have on the factory floor.

[00:55:00] Whitney Lee: Mm-hmm.

[00:55:01] Coté: last part didn't work out, but it was fun to say.

[00:55:03] Michael Forrester: No, no. But, but I, I mean, I think you're onto something. I think there's this sense that somehow if we just put AI in there revenue and ROI, like all that's gonna go up, right? And know, the, I think the answer unequivocally, and I, I'm not gonna cite the Nanda report with the 95% MIT, 'cause by the way,

[00:55:20] Whitney Lee: He just did

[00:55:21] Michael Forrester: just wanted to be clear about that.

[00:55:22] People should really go and look at that 'cause there's a lot of questionable... That's a qualitative study. It actually wasn't part of MIT, it was actually in f- anyway, there's like a b- like I know people quote that a lot 'cause they like kinda want that to be true, and I'm not saying people are not getting value, like I think people are like failing in some ways, but this is also new technology.

[00:55:40] I mean, it was November 2022 when this first came out. We're in early days. Does anyone remember where Kubernetes was at in its first three years, right? I mean, I, I, I did my first production deployment two years after Kubernetes went 1.0, and it ate six months of my life. Like it

[00:55:55] Whitney Lee: Mm-hmm.

[00:55:55] Michael Forrester: worst, right? And it's totally different than what it is now, right?

[00:55:59] But I mean, that's what you do when the first two or three years of a technology. But I think Coté and, and Whitney, I think you're leaning into this also as with the questions, is that like people are thinking, "Oh, we're just gonna shotgun AI into the system," and like what's the organizational ROI? That def- that framing I think is completely misapplied.

[00:56:19] Whitney Lee: Mm-hmm.

[00:56:19] Michael Forrester: that the, the ROI is always at the workflow level, and I think that everything we already know about CI/CD and good SDLC and all those other pieces and how to run applications at scale and how to do all that 100% applies. I think AI should be applied as a last resort to the areas where it can add improvement and should not primarily replace these deterministic setups. If anything, now that we can bring an AI partner into code, would argue in some cases better, faster, quicker, depending on the coder, right?

[00:56:55] Whitney Lee: Mm-hmm.

[00:56:56] Michael Forrester: would say like let's bring it in and have it help us write these artifacts and write these deterministic processes at speeds that we've never seen. I mean, the whole premise of this conversation was that I created an entire system of role-playing for my son that I couldn't have done within the 20 hours that I had, 'cause it would've taken me 60 to 100 hours to actually deliver. And in this particular case, because I had AI in the, in the path, I got to delivery way quicker than I ever would, but as an assistant, like as someone who was using it to generate artifacts that then could be deterministically consumed, not as someone who took the AI and said, "Hey, by the way," like to your point, Coté, you can't just launch a chatbot and be like, "I wanna play Dungeons & Dragons.

[00:57:39] Here's my character. Let's go."

[00:57:42] Whitney Lee: Hmm

[00:57:42] Michael Forrester: can, but it'll be a sloppy, bad experience, and that might be acceptable for a Dungeons & Dragons campaign, but it's not gonna be acceptable for an enterprise. And so I, I think that the still, like the value that we're realizing with AI now in that context is that we partner with it, it allows us to generate these great artifacts.

[00:57:58] If we wanna go to en masse kind of read-only workflows where we're digesting massive volumes of information with a fleet of agents, I think that's a great way to do it. I think the fantasy of we're gonna get-- This, by the way, was literally a request I heard from a large Fortune 50 company. "We wanna replace all our software engineers with agents and have it write code at speed."

[00:58:20] And I was like, "No, not gonna happen." And they're like, "Why not?" I was like, "How much money do you wanna spend to do this compared to getting your human beings to do it, who are gonna do it more reliably and cheaper and more consistent? You're gonna spend so much time wrangling with a system that's doing all this change.

[00:58:39] Why don't you just get a system that just reads before it starts writing? And then once you get the read process down, then you can start thinking about small workflows in which you can introduce its ability to change things." But like, so anyway, I, I think that's the basic premise in August 2026 is that let's partner up with it. Let's have it enable us to deliver things at speed that we've never seen before. But read-only is probably the best enterprise application in each individual workflow, but I don't think there is an organizational ROI. I don't think there is one

[00:59:09] Whitney Lee: I'm, I'm hearing too, everything that can be deterministic should be deterministic. Everything that can be verified deterministically should be verified deterministically. When you use an LLM, ideally you're using it to write deterministic things. What's actually running is deterministic. Or if you need to use an LLM in a way that's not deterministic, you keep it very tightly scoped

[00:59:31] Michael Forrester: Okay. So true confessions here, Coté, just to make sure we're clear. I was doing that and had not labeled it until Whitney and I had a talk for one of our talks, and she said, "Oh, it sounds like you're doing everything deterministically first." And I was like, "I am?" Right?

[00:59:49] Whitney Lee: But yeah.

[00:59:50] Michael Forrester: she goes, "Wait, you're, you're running hooks to, like, do all these pre-checks, right?"

[00:59:54] And I was like, I, I am?" And I was like, uh, I didn't even realize that I had set up all this stuff. And Whitney, you were doing the same thing, by the

[01:00:03] Whitney Lee: Yeah, I was doing that for my, the software I'm building too. Mm-hmm.

[01:00:06] Michael Forrester: the one who said, "It sounds like we should be doing things deterministically first." And then immediately I was like, "Oh yeah, that's a great idea. We should do things determin-," and I was already, like, doing it. But like, she was like, "I'm already doing it, too, but like, do you realize that you're doing it?" I was like, "No, I don't real-" She's like, "Yeah," it was

[01:00:20] Coté: Yeah

[01:00:21] Michael Forrester: it." I was... And so instantly the frame was, yeah, you should go determinism first."

[01:00:25] And so just so you know, anybody who ever hears me say that, it was Whitney who was like, "Hey, by the way, do you know that you're already doing determinism first?"

[01:00:32] Coté: Yeah, yeah. And, you know, as, as, as always, I think, I think Whitney actually knows what she's doing instead of just crashing into and discovering how things work.

[01:00:39] Michael Forrester: Yeah.

[01:00:40] Whitney Lee: Well, I do both, but then I'm good about... Once I crash and burn and figure it out, I'm pretty good about being able to talk about it.

[01:00:47] Coté: That's right

[01:00:47] Michael Forrester: I'm over there like a gorilla, like, "Yeah, this thing just works. It's great." I'm like doing great when he's like, "Hey, by the way, you know you're doing this like thing." And so anyway, yes, I just have to give attribution 'cause like I don't think I

[01:00:56] Whitney Lee: What?

[01:00:57] Michael Forrester: model in my head before

[01:00:58] Coté: Uh, and I, I like, I, I like, I like what you like, like several times you, you've, you've, uh, explained Michael, is that like, uh, with, with the, um, understanding of like, yes, you can't read all of Wikipedia in one chat session that, that Whitney correctly pointed out. But it, it is-- I, I, I would be comfortable saying something like, "AI is really good at comprehending a big mass of stuff," right?

[01:01:25] Or comprehending is the wrong word, whatever. Not to be too precise, but helping you understand something that's hard at human scale to understand, right? And so that probably is bringing the cartoon characters in again, the executives. Like it, it, it do- it does-- I mean, I wish this was the case two years ago, but it seems like a great thing would to be, would be is just say like, "Hey, how about you just use it for a week, and then I'll come back.

[01:01:52] And, uh, I'm gonna, I'm gonna expense that whole week flying back and forth and eating meals as a consultant. But this is your assignment. Do that." And, and it's not only to like learn what's possible, but I think, I think if you, if you had the framing you were saying is like, "What I want to give you as a consultant, ironically, is what a consultant would give you, is I'm gonna go take the time to understand your organization or even what it is you're asking me to do, uh, and, and give you kind of a read on that."

[01:02:22] Like, there's the joke about consultants that, you know, you hire a consultant to tell you what time it is. And back when I did kind of consultancy work, I would think like, "Yes, well, if you could find your watch and look at it, you wouldn't have to hire me." And so, like,

[01:02:35] Michael Forrester: Right

[01:02:35] Coté: know, I think AI's pretty good at that kind of thing.

[01:02:38] And, um, I don't know, maybe that's... And, and, and maybe we should hold off on that automation part until 2028, un- until we get our hopes and dreams up too high

[01:02:50] Michael Forrester: You mean the full on I'm gonna read and write to a

[01:02:53] Coté: Your, your however many, uh, however many developers you were gonna fire or, or, or whatever

[01:02:58] Michael Forrester: I mean, but, you know, uh, I was-- By, by the way, I just want to mention this in passing, which was funny. The company in question actually required SOC 2 compliance. I'm just gonna state this, and this, unless this has changed somehow miraculously in the last 30 days.

[01:03:09] SOC 2 compliance requires that every piece of AI-written code be manually reviewed line by line by a human developer.

[01:03:17] Coté: Right? And see

[01:03:19] Michael Forrester: gonna get beyond human speed, as was rightfully pointed out by another engineer, if we get alert fatigue looking at logs, long are the engineers gonna be reviewing code before they just hit accept because they're fatigued at the 10 times speed of code production that the agents are

[01:03:39] Coté: I'd-- See, see that, that would be, uh, while, while you're having, uh, you know, uh, expensive meals and en- enjoying, uh, the crisp linen in a hotel room for that week. Like, th- that, that, that would be like a good exercise to go through is like, "This project you asked me for, I'm gonna open up a chat window for you, and I want you to start typing it in there."

[01:03:57] And so for example, the compre- the read- the readability or whatever would be, all right, so we've got SOC 2, 3, 4, whatever pair of socks you want that you gotta comply to.

[01:04:06] Michael Forrester: Yes

[01:04:07] Coté: How about you have it go through that and answer questions for you? Because, you know, as, as, as we know, if you can give it all the fine print y- you want to, and it's pretty good at telling you like what's in the fine print.

[01:04:17] And that would be more of a, uh, I don't know. It, it would kind of set the expectation of what you can do with AI at, at, at an organization-- not an organization level, in an enterprise setting. It would just tell you like, if you're a highly... Everyone's always talking about highly regulated industries, but what does that mean?

[01:04:34] Like just load it up in there and have it search over it and bounce questions off of it of, of what it can do or not

[01:04:40] Michael Forrester: Well, and I will also say this, is that a, a lot of people aren't realizing how much they touch regulated indus- industries. So I don't necessarily wanna deep dive into this, but like I have worked for a number of ed tech training companies, like I formerly worked for CodeCloud, which was awesome, right? And one thing that they had to do is they had to get SOC 2 compliance.

[01:04:56] Well, guess what happened getting SOC 2 compliance? had usage of AI internally that now required a different level of

[01:05:03] Coté: Right, right

[01:05:04] Michael Forrester: even though they're not a regulated industry. But because they can't do business with a regulated industry without themselves being reg- you know, adherent to a certain level of standards, they next thing they know, they found out that they couldn't do full-on rewrite automation.

[01:05:15] They actually had to themselves be like, uh, adherent to the standard. So I, you know, I wanna point that out is that I think people think, "Oh, we're not SOC-- you know, we're not a regulated industry. We don't have to do that." Well, if you do business with one and you want to be certified as a business partner, for most of them, you are actually going to have to produce a certain measure of compliance. to my point, uh, like partnering with it, keeping humans in the loop, like keeping the read capability, I think still is super valid. The change part of it, oof. Especially if you have it like generate like PRs to make code changes, you're gonna have to have a human being review every single line of that code

[01:05:51] Whitney Lee: It-- Another thing about getting rid, like laying off people is your competitors may not choose to lay off people. They may choose to, like, take those people and have them do a lot more with AI as their sidekick. So the layoff paradigm only works if your whole industry is doing l-layoffs, I would say. If you, if you d- uh, m- uh, to bring it back around to D&D and to, um, Lancer, if instead you, like, you could get rid of the 20-hour person 'cause you could build Lancer or, like, you took something you could do in 20 hours and that would take hundreds of hours and did it in 20 hours.

[01:06:26] And so lot of people should just take their engineers and magnif-magnify what they can do. Multiply, that's the word. Multiply what they could do as opposed to getting rid of them altogether.

[01:06:40] Michael Forrester: So that, that to me is a super valid point, I think.

[01:06:43] Whitney Lee: Mm-hmm.

[01:06:44] Michael Forrester: I would say, like Coté, you made reference to like running CI/CD pipelines in like 2003, like before DevOps was the thing, right? In 2009 onwards. And me, what's interesting about AI is exactly the whole Dungeons & Dragons lancer reference. I am now able to do things that would normally just sit in a backlog that I would never touch

[01:07:02] Coté: That's right

[01:07:03] Michael Forrester: it's just too much effort.

[01:07:04] I don't have the time to do it, and the value of it is not equal to the amount of time it would take to get it done.

[01:07:12] Whitney Lee: Mm-hmm.

[01:07:12] Michael Forrester: me, revitalized my career because I have set up CI, CI/CD pipelines, IDPs. I have done all of the edges of SDLC more than I have ever done because the cost of doing it and the time sink to do it is so much lower than it's ever been.

[01:07:27] Whitney Lee: Mm-hmm.

[01:07:28] Michael Forrester: me, and I, and I wanna be fair, there was only one company that made a statement like that, and I think they were being a little glib about it, and I, I don't, I don't think... And traditionally this company, by the way, looking at their history, doesn't have a history of doing layoffs. Matter of fact, I don't think they've ever done layoffs. And so I think the C-level was kind of joking, but I haven't heard anybody else say anything like that from an executive level. I, so, you know, I, it's hard to tell sometimes 'cause there is a certain measure of insanity, uh, from my perspective when dealing sometimes with executives. like the other day I had, uh, people who had been using Claude for 30 days tell me that they were power users and they didn't want a beginner's class because they wanted to be the power user class.

[01:08:02] And I was like, I was like, "I've been using Claude for like, since like February of 2025, and I don't know that I would call myself a power user, but

[01:08:10] Coté: And, and you just recently found out about hooks?

[01:08:14] Michael Forrester: Right. Well, yeah. Thank you. Thank you. Thank you

[01:08:19] Whitney Lee: I'm kidding. I'm joking, joking.

[01:08:21] Michael Forrester: yeah, so I mean, I think somewhere, uh, you know, I, I think that this has enabled us to do more with less than we've ever done.

[01:08:29] Whitney Lee: Yeah

[01:08:30] Michael Forrester: of getting rid of your engineers who have a backlog of 1,000 items that would all have been super nice to have that would've increased, you know, decreased cost, increased performance, increased reliability, increased security, increased

[01:08:39] Whitney Lee: Mm-hmm. Make it.

[01:08:41] Michael Forrester: you have an accelerator that can amplify the engineer's capabilities.

[01:08:46] 'Cause I actually

[01:08:46] Whitney Lee: Mm-hmm.

[01:08:47] Michael Forrester: not even an accelerator, I think it's an amplifier.

[01:08:50] Whitney Lee: Mm-hmm.

[01:08:50] Michael Forrester: doing 2X y- it can probably amplify you to 10X work. If you're doing 10X work, it can amplify you to more. And I think all of these things in the backlog that you previously haven't been able to touch, the dream is now, is that if you can figure out how to get AI to act as a good partner for you, you can now check off all the boxes for all the things that you ever wanted to get done inside of your organization in the specific workflow a rate that is probably faster and honestly higher quality.

[01:09:18] If you're doing TDD and spectrum development, I'm just gonna put that plug out there, right?

[01:09:21] Whitney Lee: Uh-huh.

[01:09:22] Michael Forrester: TDD and spectrum development in a pipeline, then I think you can get to the kind of acceleration you're looking for. If you're not doing those things, by the way, you are really opening yourself up for disaster,

[01:09:33] Coté: Well,

[01:09:33] Michael Forrester: opinion

[01:09:34] Coté: you know, one thing that, uh, I think we can check off our list is listening to another episode of Software Defined Interviews, which you, the listener, have done. And, and, uh, you know, uh, maybe you automated some of your, your processes so you could have the time to do it, which, uh, which was just probably great.

[01:09:50] So

[01:09:51] Michael Forrester: summarizing

[01:09:52] Coté: no, no, you don't wanna summarize this episode. See, I, I think I, I li- I like your framing. Like, like the, uh, or, or the point is like instead of just doing what you're doing better, why don't you do things you've wanted to do or, or that you're not already doing? And, and you can layer that on there. And, uh, I don't know.

[01:10:10] That sounds better than firing people, don't, don't you think, Whitney? No.

[01:10:16] Michael Forrester: How are you supposed to answer that? No, let's fire people. 

[01:10:19] That

[01:10:19] Coté: That's right. But, uh, so, uh, so Michael, if people wanted to check out your stuff on the worldwide web, do you have a homepage? You got a LinkedIn? Where, where would you send people?

[01:10:29] Michael Forrester: Yeah. So, uh, Michael Rishi Forrester, R-I-S-H-I. So michaelrishiforester.com is available. We'll make sure it's in-- I think it's okay to put it in the show notes. And then the, the, the game was called Stonesong, but because of copyright,

[01:10:41] Coté: Ah.

[01:10:42] Michael Forrester: Antiphon. So anti-P-H-O-N, antiphonthegame.com. We'll put that in the notes if that's acceptable as well, and you can catch it.

[01:10:50] But I'm frequently on LinkedIn, is probably one of the primary methods that you might see me socialize. But I post a lot of stuff just to my website. But Antiphon the game, you can actually read the entire story of mechs sung alive from stone

[01:11:03] Coté: Ooh,

[01:11:03] Michael Forrester: this whole

[01:11:04] Coté: that's a fun angle

[01:11:05] Michael Forrester: Oh, yeah, and it's very psychological as well, which is a whole kind of other thing.

[01:11:08] But yeah. So yeah, you can-- Antiphon the game is where you can go. You can read the whole thing. You can read the story. There's a short story out there. You can actually see the decks. You can actually see the entire board and all the missions and everything out there live, including the weird sea creatures that are all of the capital ships for the space battles.

[01:11:22] So

[01:11:23] Coté: And, and various types of shrimp it sounds like, which, which

[01:11:26] Michael Forrester: shrimp, cavitation shrimp. I mean, it's all kinds of, yeah, lots of shrimp.

[01:11:30] Coté: Do you, you know, you know it, you, you know, you know in,

[01:11:33] Whitney Lee: Sure

[01:11:33] Coté: in, uh, in, in Dutch the way that one of the way- idioms for saying someone is dumb is to say that they're shrimp-brained. Like, not, not small, but that they have the brain of the animal, the shrimp.

[01:11:44] Michael Forrester: Well, I mean, considering that some of these shrimp punch, punch hotter than the surface of the sun, you know? It might be

[01:11:51] Whitney Lee: Wait, that's real life or that's in your game?

[01:11:53] Michael Forrester: No, it's called, uh, it's called, uh, s- uh, cavitation. So there's

[01:11:56] Whitney Lee: Uh-huh

[01:11:56] Michael Forrester: punch so hard that for a moment that the point of contact,

[01:11:59] Coté: Ah,

[01:12:00] Michael Forrester: than the surface of the sun

[01:12:01] Coté: love it

[01:12:02] Whitney Lee: didn't know shrimp could punch, period.

[01:12:06] Michael Forrester: they're

[01:12:06] Whitney Lee: So

[01:12:06] Michael Forrester: experts. They're, they are the martial artists of the sea. Okay? Like, you

[01:12:11] Whitney Lee: Okay.

[01:12:11] Michael Forrester: the chicken of the sea, these guys are the martial artists of the sea.

[01:12:14] Coté: Yeah. Yeah. I, I think, I think they, they, they're definitely in that, uh, I'm gonna call it like the, the, the Lovecraft filter, where you find some everyday creature, like a shrimp, and you stare at it longer and longer, and you're like, "This is the most fucked up thing I have ever seen."

[01:12:29] Michael Forrester: Yeah, yeah. Yeah, yeah

[01:12:30] Coté: And, and you're like, "I don't think this is of this earth."

[01:12:33] And, and then you just follow that trail as far as possible

[01:12:36] Michael Forrester: Yeah, it's like when I was a kid, someone told me that electric eels are a thing, like they electrify thing, and I was like, "Wait, what? No, that can't be real." Like electricity, it's water. Like even like at eight years old, I was like, "You can't have electricity underwater." And then eventually I was like, "Hold on, wait. eels are a thing. That's weird."

[01:12:51] Coté: Yeah. I can't even swim when it's about to rain outside, and you're telling me this? That doesn't make any sense.

[01:12:56] Michael Forrester: Exactly. Exactly. So we have underwater martial artists that can punch hotter than the sun. So I

[01:13:02] Whitney Lee: Yeah.

[01:13:02] Michael Forrester: that's

[01:13:02] Coté: Exactly.

[01:13:03] Michael Forrester: in

[01:13:03] Coté: Well, this has been episode 125. So if you wanna get those, uh, the, the links to those things we talked about, uh, you can go to softwaredefinedinterviews.com/125. And, uh, you know, speaking of enterprise awesomeness, that's why I have to wrap up 'cause I'm going back to my enterprise life after this.

[01:13:21] And with that, don't despair. Go play a game. It'll be fine. See everyone next time. Bye-bye.