Cote:
[0:01] Well, we had a little crack in the ceiling. You know how it goes. You get a crack in your ceiling in the kitchen or whatever room at the house you live at. And a year later, you're like, oh, that crack is still there. And I went in one day and I was like, was someone making pancakes last night? And they spilled a bunch of flour on the floor, but it was actually dust from the crack at some point. So anyhow, we got it fixed recently. I think it is being fixed. Maybe it's actually done. And they take off the, you know, how house, well, I guess we kind of know how houses are. There's like, there must be some word for it, but there's like the facade of the ceiling. You remove that and there's the actual ceiling up there. And so they had pulled apart that, you know, to redo it. Big job. And I was thinking like, oh, man, I took like five minutes to think about what are all the things I could have them put up there, right? Like, could I have them run some Ethernet into here?

Brandon:
[0:58] A little Wi-Fi access point up there.

Cote:
[0:59] Yeah, like all sorts of stuff. Maybe a home camera.

Brandon:
[1:03] Depending on where it is. Yeah, like it.

Cote:
[1:05] Yeah, yeah. And of course, I mentioned this to the CFO of the Cotec organization. And the CFO was just like, no, just move on with your life. They're just going to paint it and replace it. That's all we're doing. We're not getting complicated. But it is true, you know, because the problem is to pull Ethernet in there, we would have to pull it from somewhere. It's very complicated. to it's it's like one of these old houses that's uh you know annoying but it did get me thinking, maybe i should have gone and gotten like you know ikea lighting iot, home automation fixtures to put in so that i would at least have that but then you know a couple of years ago i went and i put a bunch of ikea light switches in or you know not like hardcore ones and they're just like they're just like trouble. I just like, there's this one that is, is this is, this is always where the issue is, is.

Cote:
[2:02] Our bedroom so like i go up there and like sometimes the switch doesn't work but i don't notice because i'm like one of these early adopter nerd people i'm like i'm just you know i'm in the game, i'm using the internet of things stuff this is this is my life and then and then meanwhile my wife is like i can't see, and so it's dark and like it is it is kind of true that uh i do sort of miss It's just like a light switch that works all the time. And I tried to buy one of those light switches that is an Internet of Things thing so that you have a physical switch and you can also use your phone with it. But then, of course, I was defeated by house built in 1923 situation where it's just like, oh, you've got the old wiring. This you need to have a green wire, not just these two wires. You need the green wire to plug this thing in. And I'm just like, I don't understand why. Like, it's a switch. But so I got defeated by that.

Brandon:
[3:01] And of course, I was- So this brings up the best question, Kote. I was, you know, we were talking a little bit before. It's like, is the home administration effort of doing any type of home automation at this point worth it in an ROI sense? Like the amount of times that you have to like, one, set it up, maintain it, your users, which is mostly your family, their unhappiness providing them support, versus whatever potential benefit is being gained, is it actually worth it? Because I've kind of come to the conclusion that it isn't. Like there is no home automation that I've set up that I think actually, other than just being a hobbyist, I believe this is a pure hobbyist domain. It's like, it's fun, but it doesn't actually offer any productivity gain to the overall household. And if anything, it's a net negative.

Cote:
[3:54] I think you nailed it. There's no outcome, as we would say. There's no like additional thing. And, you know, the CFO, I know she listens to our podcast. She gives me commentary, which is great. And so, you know, recently there's a similar story and her wisdom once again prevailed where I was like, I've taken, all the audio books and now we can listen to them through a different thing. It's easier to listen to the books. And in a very understanding way she was like, So, how is that different than listening to them in Audible? And, you know, I didn't, I was like, it's more open, but I had no answer. Now, you know, she very quickly figured out the answer, which was like, well, it would be somewhat easier to copy an MP3 or whatever to our kid's device to listen, but it's also kind of just easy enough to figure out the stupid UI in Audible so that you can family share and it works out fine, right? Like it's, it's annoying and tedious, but I think, I think both of these instances is exactly as you quantified. The outcome was just giving me something to do. And it was just sort of like a hobby. That, that was the situation.

Cote:
[5:08] There was no other reason to do it.

Brandon:
[5:09] Now, where do you come home on all the various home assistants? Like, you know, I have, uh, we've got the Alexa here at my house. We've got the, uh, I've got the, a story speaker. I don't have any Google home. Um, and of course now there's the, the actual home view from Apple. My net take on all these things is like obviously don't have the home view, so I've not seen that one. But like we've had these assistants now probably for somewhere between five and ten years and pretty much useless. The only things occasionally my wife will ask it to play music off the bad Alexa speaker. It kind of works. When we ask you questions maybe like 80% of the time because this is a reasonable answer. So 20% of the time, it's just like, the answer's Captain Hook. You're like, what do you mean? What are we even talking about that? So something that's ridiculous.

Cote:
[6:00] The answer's Captain Hook.

Brandon:
[6:02] And the only thing, useful thing that we do use it, I guess we use it as like a shared place for a shopping list because when you're in the kitchen and you're using something, you can just tell it like add such and such to the shopping list. So that's the only use case we've actually done something where I would actually say where it's like increased productivity, right? Where is that everything else, complete waste time. I'm not even picking on, I don't even think it matters. Like I would pick on these equally between Surrey and Alexa. I think they're both equally useless or equally useful, depending on how you want to say it. The Siri speaker, I think the only thing that's good about that is it's a really good speaker because it automatically connects in the Apple ecosystem. But the actual intelligence, it's not intelligent at all. I mean, you can barely set an alarm on it before it gets confused. But of course, companies are spending... I mean, think about how much money these companies have spent on these devices and the networks to run these things. I mean, it's got to be billions of dollars. And they just fundamentally don't do almost anything. Am I being too harsh? Like, have you had a different experience?

Cote:
[7:03] They just weigh, I mean, they're not useless. So we have, of course, we have Alexi. We have some of those. And then we got some, we got the Siri things. Now, the Siri thing is annoying because it always, like, I just want to listen to the, we listen to music, of course. I always want to listen to the radio. We listen to good old KOTX. And so when I tell Alexa to do it or listen to KNKX out of Seattle, it just does it fine. I do that, plays it. But then the HomePods, they get a little confused sometimes. Like I have two of them. I got a big one by the TV and a little one by the dining table. And like I even made a home automation thing to play it on both of them. But sometimes it like only plays on one of them. And then other times it's like playing from my phone instead of from the speakers natively. And then like if you want to listen to music on the HomePod connected to the TV it has to turn the TV on for some reason I don't know it's all like, goofy uh and so like it it seems it's it seems like one of those things uh as as happened so much where.

Cote:
[8:13] There there's probably two things happening the people who design and work on these things i mean i assume, they use them right let's just be kind and say they use them and they just must have maybe it's like me and the ikea light switches, like they just don't even notice like the friction because you know they're they're used to it or you know maybe they don't use it Because it seems like there's these core use scenarios where... It would be relatively easy to just like nail it and make sure that it works. Like here's like 10 scenarios for listening to music that we're going to have. And let's make sure that they all just work well. I don't know. But yeah, I don't, I don't ever like, I use my phone to set a timer. I don't ever set it. I think Kim sets timers on these things, but, but I don't.

Brandon:
[8:59] Well, I don't know, you know, if you watch the, the Homeview ad, so there was an ad, there's like a three minute ad where they kind of, it's like a fictitious family, like a busy morning and they end up using the home view. So there's been a lot of takes on this, right? Some takes is like, it's ridiculous no family would ever act like this. This whole thing is stupid. And most of the time, people would use their phones, which I genuinely agree with, right? Then there was like the countertake is like, well, of course it's contrived. It's an ad. They're just trying to show you everything you could possibly do. And of course, it's a fake family of actors. They're just trying to act out as many different things that you could do so that you could actually get a sense of what it could do. Now, I want to come out with a third thing and be like, I just think if you gave me one of these, or if anyone, if we just sent this to a random listener and we said, here.

Brandon:
[9:50] I want you to recreate everything they do in this, just task for task, not shot for shot, but just literally like do everything that they try to do. I want you to do it at home, right? I would bet they would be able to only accomplish 50%. Because the other things just wouldn't work. Like the stuff like when I was watching it, it's not even so much like, well, I want it. They're like.

Brandon:
[10:11] It will automatically switch between like, there can be up to six people in the family, right? And it'll automatically detect who you are. I'm like, no way, no way. You give me that, like, no way does everyone in my family get properly identified, when we move through it, right? I mean, the second one was like, you know, when the daughter is meeting the boyfriend on the thing, on the camera, I'm like, no, no way the camera is going to load correctly. I don't think, I think the wifi network will be wrong or the camera will go fuzzy. Uh then there was like the call oh like the recipe like i want to be able to see the recipe while i'm making i'm like no way would the recipe be properly shown and you wouldn't have to like scroll up to see what it was or you do something like it would be like in metrics units and i'd be like how many cups are in a liter or whatever or something and it would just it would just completely it wouldn't work like you just know like i'm just so it's not even so much i'm against the idea i'm like i you give me this and you just let me beta test it and you know what we'll take a video of me with my just regular normal looking family. And we'll just show you all the stuff that didn't work. In fact, someone should do that. I don't know why these Mac influencers don't do it. Be like.

Brandon:
[11:19] This seems like a perfect thing for Marquez out there to do it. Be like, listen, he's going to just put it down. He's going to do everything that they did and see if it actually works. Because you know, like, this is how like predisposed I am. I'm like, there's no way it's going to work. It's just no way this stuff you would get through all of this task without any failure. So that's my challenge to the internet influencers. Someone get one of these things, just do every single task as you possibly, just like they did it on there and show us how many of the things can you actually get through? Because I don't think you're going to get through. I think I would put the over under 50%. 50% of the things they show would actually work on the first try.

Cote:
[11:56] Yeah, yeah. I mean, having having multi users on a not Mac Apple device would be just like a fundamental shift in the reality of the Apple world. Like, it's, it's kind of crazy that like an iPad is bound to one person, a phone, a little less crazy, but like, the, you know, because it would, it would be cool to have like a panel on your wall that like listed your calendar all the time. But of course, the screen's going to lock. It has to be logged into someone's account. So then you make a fake account. And if someone wants, it's just ridiculous. So having that would be interesting. But yeah.

Brandon:
[12:33] Well, even that's even more. Go further with that. Be like, well, what calendar would you be syncing? To an Apple calendar? No. You'd probably almost sure want to use a Google calendar because that way other, people could see it. And would the Google calendar automatically sync to the Apple calendar and be shown on the device all the time without any problems? 100% no. Like, we know that does not work. We live this every single day. It's like, that does not work. There's no way that would work. That's the most basic use case. It's like, oh, everyone's storing everything on a local remote iCal calendar. No one's doing this. No one in the industry of the world that actually has to interact with other people is using iCal. I mean, it's 100%. So I don't know. So the whole thing is quite funny.

Cote:
[13:15] Well, you know, related to that, I was reading a report that some UK people put out. I forget what it's called because I'm an American. And, you know, like, I think the UK is not special, but as always in Europe, the UK is kind of sort of its own case, but still sort of representative of broader EU stuff.

Cote:
[13:36] In the sense that everyone over here, Brandon, I don't know if you've heard, but they want to have sovereign cloud. Now, I know in America, no one knows, you don't ever hear this phrase sovereign cloud or anything because, you know, in the American myopia, It's just cloud. The whole cloud is American. It's pretty much what the situation there is. And so, but of course, you know, people want to cut a dependency on American companies and things like that. And so, this is the most recent one. I think a lot of governments have been doing a study or oftentimes it's also the association of regional cloud providers that are doing studies, which is fine. I participate in such marketing activities over the years. Uh but they found and this is this is uh you know i think this is, maybe classic like this is a classic technology problem of the finding was that i think i think it's the business sector and the government so if not whatever, but the uk is highly dependent almost, they i don't think they say locked in basically locked in to aws and microsoft, and i think there's about if you read through the report which is i think, i think a lot of uk like public facing stuff is actually very well done they i think they've got a lot of principles and guides that they follow very easy to read, i'm sure there's things that aren't but i encounter a lot of good stuff um.

Cote:
[15:04] And and the reasons that i as i remember them are uh.

Cote:
[15:08] Let's i'll go all over the place one they said egress fees were very uh prohibitive which you know that probably is because like cloudflare doesn't have egress fees. So somehow you could, that might be a choice that the companies are making. I have no idea. I don't even... I don't even understand how you pay for that kind of stuff. It seems like it would just all even out in the end, but maybe not, from a seller perspective.

Brandon:
[15:33] Let's just pause for a second. Do you think egress fees is really lock-in, or is it just the convenient thing that analysts, business types, and even ourselves, podcasters, like to talk about? Because this is what I always think. It's like, once you're in there, I don't think you really want to leave. I mean, because it's like, you're so wedded to the console, but then people like us are like, you know, those egress fees are important. But like, no, no. People are like, I'm happy. The last thing I want to do is move all these VMs to another console with another infrastructure and fix it. So I actually don't think the actual egress fees is why.

Cote:
[16:08] I agree, which gets to the next thing. So there are these categories of things that were like, let me just cover the second one. Because when I was looking through it, and of course chatting with my robot friend about it, And I was like, let's divide this up into choices that the companies are making to keep people, you know, like, I don't know what to call it, like choices they make that in order to keep people in versus just like the natural order of things. Right. And the other one, another choice is that from my read of it, you get a you get a good discount on Microsoft products if you run them in Azure, which is sort of like, yes, I think that I think that's how synergies work. And so, you know, and then you pay more if you run it in another cloud. That's a choice that Microsoft makes with pricing, which is fine, but it's not really, it's not like welcome to the ELA.

Brandon:
[16:57] Welcome to some bundling magic. Yep.

Cote:
[16:59] And those, those are kind, and then, and then there's, so we move closer to, to, to the more interesting part you were getting at. So there's another point that was like. And the, I'm going to summarize a lot of what I was saying is like, they don't use open standards. So it's hard. There's not a lot of portability. And I think somewhat implied in there is like, there are no open standards, right? So, so those are some things that make it difficult to leave. But I think, I think what also is kind of written up is what you were getting at is like, also, we're happy with it, right? Like, like we, we, we actually like this, uh, and it works fine. And so that's a, it's, it's like lock in by being a good product that, that people like using and that gets the job done, which, you know, that does seem, uh, that's, that's a lot of like how, how tech sort of lock in stuff works unless, unless you somehow find a smoking gun where there was some email where they were like, Hey, you've got to delete those three lines of code. Otherwise, it's going to be really easy for people to leave, right? Like, it's just more like, no, it's just like you have favorable pricing. And, you know, for you use incentives for that. And if your product's really good, then people are also going to stay and egress fees, like, like, and so it is, it does, you know, and then I guess I'm leaving off the huge thing, which is.

Cote:
[18:28] Which is because the whole study was about dependency and lack of, I guess, competition in the market. Like you would like to have more options in there. And the biggest thing which they lead off with, which we've covered for years, everyone has, is like, it turns out it costs a fuckload of money to be in the public cloud business. And you're going to have to spend for like 20 years before anything. And so like, it's kind of like, you know, you can't just go down to like, I don't know, over there in Hackney or whatever and be like, let's start a public cloud. Like, I don't, that's not, you're going to need some money. And, and, you know, also like you've got to have actual data centers and all the physical stuff going on, which is, it is like a type of.

Cote:
[19:10] It's a barrier to entry, I guess. But it's also, I don't know, it's, it becomes an issue. And I guess, I don't know, I don't understand this type of UK stuff, but it looks like there's an interesting path they're on where I think nationalized is the wrong word. But at some point, the UK can be like, this is critical infrastructure, I think we would say it in America. And therefore, we can tell you more of what to do, or we can tell you what to do more than normally we would if you were just selling coffee somewhere. So, I don't know. We'll see how that evolves. That's an interesting way to, I don't know, crack this like sovereign cloud thing. And then speaking of the CFO, she brought to my attention another thing that I kind of glossed over. Apparently, I haven't checked in on this since last week, but in the Netherlands.

Cote:
[19:58] There was an example of why the people on this side of the Atlantic are interested in sovereign stuff is the Netherlands police went and seized a server from, I don't know what it is, some sort of Italian like.

Cote:
[20:13] Data center that was running in somewhere in Amsterdam. And on behalf of like the Americans, I think, of course, they don't really explain it. They asked the Dutch to go get a server for them. And the last I read, the police haven't given the server over to the U.S. I don't know if that means they haven't given access to it or they haven't given the hardware to it. Hopefully it's just access to it because it would be nonsense otherwise okay who needs a freaking server, you just want that you just want to be able to ssh into it i guess but it does uh you know these these uh this interesting stuff going on in the sovereign cloud world here but the first one that and and this is the other pushback that, that i always hear in the sovereign cloud sort of buzz out there gets back to the part the thing you were raising in the uk is like, yeah but it works and we're kind of fine with it as far as the users doing things. I don't think there's a complaint in Sovereign Cloud of like, it's broken. And, you know, I guess there was, that said, I saw in London, it was a Google Cloud. Some cloud went down. And so now I think it was like UK West 1. So now they have their US East 1. You know, make sure you don't just put all of your stuff in one data center there.

Brandon:
[21:30] Welcome to the family. welcome to the real cloud.

Cote:
[21:32] Yeah yeah you know shit goes down sovereign or not.

Brandon:
[21:37] So, but I guess the question I always come back to on these things is like, well, let's just go spin it ahead. It's like, okay, if you want a sovereign cloud, what are your options? You can either build, everyone builds their own data center, right? Every company. Or I actually did an interview over on the Cloud Cat, or I'm sorry, over on the Enterprise AI show, Dominic Williams of the DFINITY organization. And they've got a kind of like a decentralized way to run cloud workloads and then use some encryption to essentially, you know, you can, if you will, be like, hey, I just want to run my workloads in these various servers.

Cote:
[22:10] Like SETI at home.

Brandon:
[22:10] It's all encrypted, right? Yeah, kind of like that. It's all encrypted, but, you know, you don't necessarily control those servers. They're just part of like a larger network, right? So that's sort of like giving you the cloud ability to do things in a cloudy way. But of course, it's just a little different, right? It's one of those, like, are you willing to buy into that technology? It's like, well, this is another thing people just don't, you know, it's like, I don't know, maybe there's just something about, like, I want to be able to walk to a data center, even if I would never go. I just want to be able to go to the data center, and I want to be able to see my servers. Like, that's, like, what people want. You know what I mean? It's just like, that's what it means to be a private class. I can go physically see them. And if I can't do that, then I just don't feel like I just, I don't feel like I'd sovereign. I don't feel like that I personally own it. Like is... Because I mean, the thing, the dilemma always comes back to like, the UK being a good example. It's like, well, they could like, they could, they could fund data centers for all of this themselves, right? And, or they could have their own cloud provider just in the UK, very expensive, very hard to do. Or, you know, as we kind of, the world we've lived in for however many years before that, everybody, every company just goes and racks its own servers, buys their own stuff. They use maybe some stuff that your company provides, VMware or others, right? And they just run it that way and they don't get access to all the AWS and Azure things, right?

Brandon:
[23:40] And I know both of those vendors do have these various solutions they can, if you will, they can kind of make it like a private cloud, but of course they always want a connection to it. So it's like, I don't know, it's almost like what do people really want when they say, I want a sovereign cloud? Do they really want to do all the racking and stacking themselves? Or is that just like too much work?

Cote:
[24:00] I, I, well, you know, I think, I think, you know, obviously the people who, uh, want to start up and run commercially the clouds, they want to have business doing that, but which is fine. But I think it is like the more of the, uh, the, the, the general desire is like, uh, I don't want my shit going to American law authorities. Right. Let's, that's pretty much the entire, like there, there's a good, uh, what would you call it? There's good, um, what do you call it when there's like the story that everyone tells over and over again like there's there's several of these like what yeah there's several of these stories in sovereign cloud and the first one is always uh i think it was, uh a criminal what's the the international criminal court down there in in the hague as we would say din hog uh, like some, uh like someone probably someone in the trump administration didn't like them so they got microsoft to like turn their email off uh and like, you know.

Cote:
[25:00] I think email and calendaring is something you don't notice until you don't have it. And then you're like, holy shit, everything is stopped, especially if you work in an organization. So that, I think that was a couple of years ago, maybe, and that freaked people out a lot. And then, you know, like I was saying, the, uh, the Dutch police, uh, seized a server, which I was thinking in that situation, like there probably is also a situation where it's like, yeah, but like, what if we need a server, from america like we need to maintain this friendship and it'll be fine, not not it'll be fine for whatever reasons but like it's kind of dicey if uh if you're not helping each other out i would imagine and then uh let's see and then and then you know you have all those kind of incidents like that happening and i think there really is just like, and and it gets to that zone of um, you know when when you talk about uh.

Cote:
[25:52] Uh the only time a data center is going to be modernized is when there's a date that it's going to be shut down it's, it's in it's in a zone that is not that that's like outside of that which is like well we just like don't want to run american technology and be under american jurisdiction and then you know you could apply your rule and be like, by what date like like what what what is a date in the future that you can hang this on And there isn't one, so it'll probably never happen, right? Like, it's, I don't know. Maybe if there was a date of, like, a law was passed, and now America needs the encryption keys of everything, then there would be a date. But there's no, like, there's no definitive date. There's just, like, this, ooh. I mean, ultimately, I think it's, like, we don't like that Trump guy. And so maybe in two years, we'll see how the Sovereign Cloud thing changes around.

Brandon:
[26:39] Well, I wonder, it does kind of feel like if we're just doing some strategy work here, I'd be like, I'm surprised in some ways there hasn't been the equivalent... I don't know, some people can maybe email in because I know it's not like there's no one doing it, but there isn't like an EU de facto cloud provider, right? You know how Mistral is the de facto AI provider for, it's in France, but it's like the European AI. There's one, if you want it more European-centric, you got Mistral, right? And everything else, I'm trying to think, everything else is basically basing this. It feels like somebody would start one and be like, yeah, we specifically are catered to the EU. We're EU first, right? We're doing the EU way. Because, of course, there's China, right? Alibaba has its own, version, right? Obviously, that's huge in Asia and certainly, obviously, in China as well. So it's like, I don't know. It's kind of interesting.

Cote:
[27:39] I don't remember the names for all of it, but I think also maybe about a year ago, there's some EU initiative. And I think there's maybe like three to five identified EU-based cloud providers. One of them is like, is it like OVH? And then there's, it's not Herzog, that's a film director, but there's a German one that starts with H that's also like in there. And then there's a few other ones that are sort of like the sanctioned.

Brandon:
[28:01] But none of them are like, because I think what they're kind of saying in this thing is like, really what you want is like, okay, there's, because we can like list those off. We can be like, okay, there's obviously AWS, there's Microsoft, there's Google. We maybe see Alibaba in China. but like you want one more that you would say and everyone's like, oh yeah, that's the European version.

Cote:
[28:18] Yeah.

Brandon:
[28:18] Right, that's the big one everybody in Europe uses because... Because they make sure that there's the correct pop-ups everywhere. No matter where you log in, there's a pop-up. And people in Europe are like, this is perfect. This is exactly what we want. We have all the pop-ups and we have all the cookie warnings. This is exactly what we need. And that'd be perfect, right? It'd be like, oh, we're going to Europe. We got to use the European version of that. But I will say, I do think, from what I've seen, what I've been close up to it and see, I do think Microsoft and Google and AWS, like they are doing everything they can possibly do to adhere to the various European standards like maybe they're not doing, it fast enough I don't know but like I feel like every time I talk to those groups and we're doing stuff in other parts of the world that they, they know about it right they're like yeah we're putting this in Germany we're hosting it like we're giving you the keys like so I don't know part of me is like, maybe you kind of already have the keys to the servers like if you want to maybe just take a maybe you should just take a tour of the AWS data centers just go check them out see what's going on.

Cote:
[29:20] I think maybe the, so yes, it seems like all the American cloud providers are like being as genuine as possible. And I think the only like the only issue is, I mean, if you're on the paranoid side, it's like, yeah, but like, what if you just went to the CEO and you were like, hey, we checked out your taxes. We could either look at your taxes some more or maybe you could give us a login to that machine in Germany. Right. And so, like, it's more like just coercion that would be going on just merely by association.

Brandon:
[29:51] But do you think no one believes in encryption? Because this is the thing. Like they all, every one of these cloud providers has a story. They'll be like, and I know, you know, Cote, I know you're, because you're, you love it so much. I know you're spending some time over in a database conference this week where like, I'm sure you could say like, everybody has a thing. It's like, and if you want, if you want this data to be super, super secret, you're going to encrypt it and only you're going to have the key, right? Only you will have the key. There won't be anyone else. Like they all have these options and, and they, and yeah, and they'll even tell you, it's like, if you lose the key, you're stuck. Like it's like you know there's nothing you can do so i don't know i wonder if that message or people just don't believe that or like no one wants to take that responsibility either because then you have to manage the key and if you

Brandon:
[30:34] lose the key then you're like in real real trouble so i don't know.

Cote:
[30:37] Yeah and then you can't have like uh like search won't work or you know to some extent.

Brandon:
[30:42] Like at some.

Cote:
[30:42] Point you've got a you've got.

Brandon:
[30:44] A uh you got to let the things work right yeah.

Cote:
[30:46] Yeah yeah yeah but just for just storage yeah i'm sure you know so i am uh i'm at the the Postgres EU conference, PGConf EU. I still haven't figured out and I guess I haven't memorized it. I think Postgres SQL has two S's in it instead of one. I wonder, after listening to people talk here, I know that there was a lot of discussion about that. Oh, you know, I saw a great some time ago because I saw I think the best talk I saw here there was, I forget his name. I think it's Vic, V-I-K. He's on, you know, this is a very endearing, like nice talk. I liked it a lot. He's literally on like the SQL committee, like the people who like standardized SQL, which is like, of course that exists. And I remember now, but you know, I don't know. I just, I don't think about these as much anymore, but it was a whole, I think it was 40 minutes about how the various ways that Postgres handles null is broken and should be fixed and like really it was thorough and i was like man i remember when i would spend days on topics like this right and so you know what it what is when whenever if you were to ask i don't know postgres syntax or whatever but if you ask um.

Cote:
[32:06] If if zero is in a set that has one and null in the set, sometimes, like, do you think that returns false or null, right? And, like, so there's all these sorts of situations like that where there's kind of, like, a weird inconsistency of how you do null handling. And it was great. It's kind of like those old programming talks where it's like, what do you think this code does? And it actually doesn't do it well. But I haven't attended a lot of sessions, outside of the programming world. And these database people, they seem like they're having a lot of fun with their optimizing things. I saw one talk from...

Cote:
[32:50] I forget, is it EnterpriseDB? I think they call themselves EDB now. Not sure why. But they essentially, again, maybe this is well-known to database people, but they have a way of, I'm going to use archaic terms here, but you can have a stored procedure in the database that will call an agent, inside the database. Weird stuff. And so they're showing this, they're showing all these select statements That's basically just like executing, like calling out to an MC, calling to not out into an MCP server or calling an agent. And I don't know that that part got me. I'm like, I think I feel like when you do a select statement. You should just be returning stuff from tables like you can't like call out to some other code and just have it send some chunks of stuff to you. It seems like you got to go back to the other guy and introduce some new keyword in SQL that's like execute shit for me. Not not. I don't know. I don't know about database stuff, but it's fun. Databases. Who knew? And also, I think.

Brandon:
[33:54] Give us the AI take. I would imagine, is there, I mean, because at this point, like, you can be so abstracted from it. You don't even really need to know about it.

Cote:
[34:01] Well, that was another, obviously, I didn't go to all the talks, but there actually was very little AI stuff. Now, people would mention it, and I think, almost every slide was clearly made by Claude, right? Like, you know, how it really likes to have those little color bars on things and the exact spacing it has, for slides. And I've noticed, like, Claude slides are very busy, though. They put a lot of stuff on a slide, and it actually does it really well. But then they also has, you know, sometimes those, you know, Claude loves using call outs, little call out boxes, and those leak into slides every now and then. But you can totally like tell when a slide is like, is AI made. Now that said, Compared to slides that I've seen over the years that nerds have made, these are much better slides. They're very clear. They're great. They kind of give you the information. But it is, I think, I forget if I've said this on this podcast, but it's fine. You should have AI help you make your slides. But just set aside a day and develop your own style. If you've got a corporate style, just download the corporate template and give it to Claude and be like, listen, do this. This is your new job. This is what you do.

Brandon:
[35:23] Right?

Cote:
[35:24] And it'll be great. And then you'll be able to escape all these slides that look pretty much the same, which I think causes a new sort of problem, which is, at some point, if all the slides sort of look the same, it's harder to pay attention to them because my mind kind of fuzzes past it. But yeah, and then there was a few talks of like there was product talks like this one from the edb people right about how ai is built into there and then there were a few um, there are a few as they say war stories about how uh ai went wrong or like it was always a story of like it wasn't a story of like you know as as we encounter every now and then the ai was like oh sorry about that like it wasn't that kind of story it was more like oh it suggested that i did this and i was like sure that i do this and i was and And I did it. And then it turned out that was a bad idea, which is, there's a very human in the loop situation there. But I don't know. You know, you go to a developer conference and it's just all AI all the time, or a DevOps conference or whatever. But here I would expect to see more like, I don't write SQL ever, right? And there was a little bit of talk about that here and there, but like very low on the AI level there. I think, I think, And there wasn't, I didn't sense a lot of AI hating, like they don't like it. I don't know. I think maybe they're just comfortable.

Brandon:
[36:47] It's like it hasn't reached them this yet.

Cote:
[36:49] Yeah, yeah.

Brandon:
[36:50] I think you need to come back next year and be like, what are they doing next year? Are they all talking about AI? Is it really, is it agent first? It's like you have your agent do everything versus now where it sounds like they're just sort of just getting into it.

Cote:
[37:02] Yeah, yeah. Yeah, it's hard to tell. I don't know. I'm not quite, I'm not confident in knowing what it is DBAs do all day, right? In the sense of, there probably is a lot of discussion of this.

Cote:
[37:16] Is the DBA just configuring databases and doing all the Toily Ops work? Probably do a lot of that. Or are they working on the optimal data models and ways of organizing things and queries? Are they optimizing that based on the outcomes that they want? That's probably what most DBAs want to do the most and they end up doing the least of. or, Another activity would be like just troubleshooting stuff. There were a lot of, there was a panel I went to that was great. It was like five people in there like, let's talk about all the fuck ups that happened, right? Like all sorts of things. It always involves like moving data around and replication. Something always goes wrong with that. Or as the SQL committee guy did, someone asked him how, see, this is like the data modeling stuff. Someone was like, how would you track currency? And he was, would you use the money field? And he's like, no, don't use the money field. Because what happens in Postgres is it uses, it determines the currency by whatever locale the database is running in. So if you replicate your database from Europe to America, it just silently changes everything to dollars. But of course, it doesn't convert it. So what was one euro is now, you know, 57 cents or not 57 cents. That would be a crazy exchange rate. You know, it just it doesn't track. It doesn't track it. So just like don't use that at all. You've always got to, you know, like all this kind of stuff is fascinating. It's making me think.

Brandon:
[38:44] I feel like all the developer sessions at developer conferences or something like don't touch the database. That seems to be the main developer kind. So it's almost like this is the place like, oh, these people have to touch the database. So they have to know all these things. All the things you just said, I'm like vaguely aware that that is possible, but I've never really concerned myself with it.

Cote:
[39:03] Yeah, and that also came out. I think there was a shockingly low amount of developer hating, but many of the stories were like, well, the developers told me they needed this and I suggested something, but then I was like, all right, here's this or you know like some developers did something else you want it.

Brandon:
[39:23] Probably won't you know change one thing and it won't work.

Cote:
[39:26] Yeah yeah like there was one story of like oh yeah in order to catch errors they would uh they would do a retry on a 15 second loop and so you know one day we came in and you know we had like i'm making these numbers up we had like 10 000 retries going on constantly and you know so that was great, And on and on and on. But yeah, so yeah, it'll be, there's one in New York, I think sometime in the spring. We'll have to see if there's more AI stuff. Yeah, I don't know. I don't know. It's a lot, because, and even a lot of the tools that you would see in demos, obviously AI built, right? Like the UI just looks AI-ish. And also like, why wouldn't you? Like, of course you would use that to build these tools. Just have oodles and oodles. They love their dashboards in this database community lots of metrics a database dashboard.

Brandon:
[40:15] Made to go together.

Cote:
[40:16] I mean that's that's the peanut.

Brandon:
[40:17] Butter and jelly of enterprise software right there.

Cote:
[40:20] And well but i mean i want to hire maybe i'll get off this topic but i want to highlight the what i think putting on my uh my marketing persona uh what the highlight of the show was at the was it db tune i think booth, It was at the end of the row of the booths and what they had set up is, so I think there's maybe three or four, like 30 minute breaks, not including lunch during the day. And so what they did is they had, you know, like the silent disco thing where you've got like wireless headphones and they had maybe like, I'm guessing like 30 or 40 of them. And so they would give that to the audience and then they would be giving a presentation from the booth with the people of the silent disco, which is great because then they can hear it and it's more relaxing and it really like they're also noise canceling i tried it out and so it really focuses you on listening to the person and like, i would i counted as much as i could when this is going on and there was there was, really like 15 or 20 people there throughout the whole 10 minute pitch uh, uh each time that i would see it which is amazing right like i i i'm sure you've given booth talks before i've given lots of booth talks and seen lots of booth talks given, and it's difficult to get people to engage and look at them. And then, you know, being very, very deep in the marketing weeds, the other thing that was genius is, of course, you're at a booth, you know, even at the Postgres SQL, one or two S's, I don't know.

Cote:
[41:50] Like conference, you still got lead gens going. You got a QR code on your badge, they come around, they scan it. And so they would wait until about eight minutes into a 10-minute talk or seven minutes or so before they went and scanned people's badges so you know those were like the highest quality leads you could get the the very very highly qualified leads that someone was, paying except mine of course because i'm not buying that shit but like you know you know that the people who stayed that long, to look at a bunch of dashboards they're really into it they're very interested so that's uh i don't know i'm gonna see if we if we i think it'd be a good tactic for uh for us to use for other people, you've got to have the silent disco. And I think if you go to silentsystems.com, there's a free ad for them. They're the ones who their names are on the headphones.

Brandon:
[42:36] I've seen this a bunch of times and I have done it a couple of times to try it out, but, I still think I'm going to go old school here and be like, hey, if you really want to know what's going on, go to the booths when they're like, quote unquote, not open and just go talk to the people and be like, hey, will you tell me what this thing is? And then just do it that way. Because it is like, maybe there's a novel, because as you're saying, I get what you're saying. They've made it through eight minutes, but it's also, you kind of feel like, well, once you have the headphones on, you're like, I'm kind of stuck. I feel bad just taking them off and walking away. But there is, I don't know, it's a weird, it is sort of- So yes.

Cote:
[43:15] From a lead qualification scoring, were they just self-shamed into being polite and staying there. However, you also have, well, they got the message. So we actually got people to listen to it.

Brandon:
[43:28] There is sometimes when you see two groups doing it at one conference, it's a little dystopian. There's like, oh, well, they have their headphones on. We have their headphones on. But this kind of comes back to what I said before. It's like, this is what everyone says they don't want, but this is what I want. Everyone's like, oh, we can't have the vendors pitch what they're like, whatever you do, don't give a vendor talk. And then half the time I'm like, I see all these people here. Like, I'd love to know what they do. Like, just get up. Everyone gets 10 minutes and like, just like here, tell me explicitly what you do and why I should give you money. I would love to know that. That's exactly what I want to hear. I mean, I just like, but everyone's so allergic to it. Like, oh my God, I can't believe it. This company is trying to like sell me something. It's like, yeah, that's kind of like, I like need to, I don't, sometimes I don't even know what, like you see a weird looking booth and you're like, I don't even, that could be a database company, it could be an AI company. I don't know. Like, I just want to know. So I don't know. I think it's okay. I want to embrace it. I want to go the opposite. I'm going to be like, all we do is vendor pitches. Just 10 minutes, you get 10 minutes, go for it, right? And just like, I want to go through, I want to see whatever, 15 pitches in an hour. Like in that way, because that way I'll know, I'll know if I even need this thing, right?

Cote:
[44:40] Yeah, yeah, yeah, yeah. Well, I think from what I can tell, this Postgres PGConf, it's pretty good at balancing like, you know, sponsored keynotes, letting vendors do stuff. Like it wasn't like, you know, those kinds of conferences that try to do what you were just saying is like, just attempt to erase the reality of like, yeah, people are selling shit.

Brandon:
[45:07] So we just happen to all be here on a day where no one's here for a job at all. No one's being paid any money. We're just randomly here to like get together and just have a, it's like, no, this is business.

Cote:
[45:17] Right, and it's like for some reason, all these people, they gave us a bunch of money and they have tables over here?

Brandon:
[45:23] Yeah, tables over here. Like, I don't know what they're doing. What's going on with these people? It's just amazing that no one's here for any reason other than just because we want to be together and talk about stuff. It's like, come on.

Cote:
[45:32] But I think it's a good conference. We'll see if I can make it to another one.

Brandon:
[45:36] All right, look forward to next year's review. You know, do they use AI next year? Yeah.

Cote:
[45:41] Or more AI? It looks, so you, I know over on the, what do you got? You got the AI Fortnite over there on the Enterprise AI podcast. I think you guys cover it twice a month now, which is great. I always look forward to it. What is the situation? I've been reading the news, looking at my 401k and I'm like, what is something, like who, what are these bonds that CoreWeave has? And then who, like now Accenture only lets people spend $29 a month on their AI stuff? That's not going to work. Like what?

Brandon:
[46:14] Yeah. No, I think those are.

Cote:
[46:16] What's the bubble watch, Brandon?

Brandon:
[46:17] I think bubble watch, two things. One, CoreWeave, you know, they're like a NeoCloud or whatever, you know, basically the cost for them to borrow money has gone up significantly. Or said another way, their bonds are trading much lower than what they once were. And that's an indication that investors are worried that they may not be able to pay back their money. So it's just like a little canary in the coal mine kind of thing. It's just like, okay, good to know. It doesn't seem to be beyond core weave at this point. But if it spreads, that would be an indication like people are getting worried. So I don't know, when you see that, does it even bother you at all? Do you even care in any one bit?

Cote:
[46:58] Well, you know, it reminds me of like all the MSP stuff in the 2000s, which like, you know, rack space and all that stuff. I don't know. That's why it seems a little distressing. I guess the idea is like how much of a domino theory can you apply here? Because on the one hand, like maybe there is, it's too many, like, are we saying there's too many data centers or there's people who don't want to, they don't want to actually pay for this stuff. But then, you know, it does seem pretty evident that you're going to want AI all the time doing all the things.

Brandon:
[47:36] The fundamental thing you said right there, the question is just very simple. It's sort of like CoreWeave says, like, we have, if you will, a backlog of things we've sold. Okay, and we need to build and buy equipment to service that backlog. And how do we do that? We use debt. Very common. This is all very common. Nothing exciting here. So they, of course, have come back and said, this is irrational. People are irrational because we have the customers, we're just running a normal business. Of course, we all know, when someone says somebody's irrational, it's like, oftentimes that can be an indication that things are not going well, right? So there's that side of it. And so they're just...

Brandon:
[48:17] I mean, like everything else, we have to just wait and see. But if you are of the opinion, if you're kind of on Zitron side of the fence, you could be like, this is a really good indication. The bond market, because this is what people always say. It's like the bond market's like the most powerful market in the world. That's where everything is decided. So if the bond market loses faith in you, you're going down. So if you are inclined to believe that and you're worried about that, this is an indication that something bad could be happening. And to go pair with that story would be, if we're doing a Zitron podcast, we would do is like Accenture, of course, a company that goes to these conferences, trying to often help you implement AI to achieve some type of cost savings or increased revenue at their companies. They have come out. Now they've disputed it. And if anyone at Accenture would like to write in and tell us, we'll definitely follow up on it. But they've said that they're clamping down on the AI spend. You can only spend $29 a month, which I guess is probably just like a subscription, like, what do I have, the Pro? Because you have what? You have the Enterprise, right?

Cote:
[49:22] The max at $100.

Brandon:
[49:24] Yeah, so that's probably just them saying everybody can have the Pro subscription for the AI thing, but no, if you use that, you can't do it, which... You know, I mean, I hate to say this, like, I don't necessarily, that's probably not unreasonable, right? Like, it's for, I mean, I don't know if everybody's doing, is everybody a max user? Maybe it is, I don't know. So, but anyway, that sort of ties together in the sense of like, well, if Accenture's limiting AI use, because they're not seeing, the productivity gains, and CoreWeave is having problems with their debt, you may say to yourself that like, well, things are getting a little, a little bubbly. So, I don't know. Is your company, do you know, do you have any rules about like how many tokens you can use or have you seen?

Cote:
[50:09] Is this actually penetrated in your world? Yeah, yeah. You know, we're like any other place. There's a lot of stuff being tried out. A lot of different policy developers have their own situation. Us not developers have a situation. We used all the stuff doing all sorts of things. And yeah, it is, you know, once, once you can't buy that, uh, that consumer thing, uh, that's super cheap, uh, things get pretty expensive. Like, I don't know, uh, I, I have to remind myself not to freak out, but, you know, I'll see reports in Claude and it's like yesterday you used $298 in tokens. And I'm like, wait, wait, no, I didn't. What are you talking about? But it's just like, for some reason, the Anthropoc people are like, you know, what we should do is we should tell people what a great deal we're giving them. Right. Like it's, it's almost like, it's a strange, not strange. Is it clever? It's something I wouldn't expect from the anthropic people to be kind of like anchoring expectations like that. There's some, that's some good, uh, we used to call it growth hacking. I think some good, like growth hacking CRO type of stuff is like, we're giving them a sweet deal, but we should tell them how much they would be paying, like how much they, I mean, how much they should be paying. And so, yeah, like I think, I think I calculated, uh, in a presentation i gave about my dnd stuff like i don't know this is like uh.

Cote:
[51:31] I think to make just the presentation with research and things like that it was something like eight hundred dollars or maybe even a thousand dollars like in in this artificial token pricing thing but.

Cote:
[51:42] So like obviously i wouldn't pay that amount of money just to make a presentation i just make it myself uh but but but it is uh.

Cote:
[51:52] Yeah if you don't it it does end up getting expensive but i think i think what you're saying is right is like yeah i mean 29 30 a month for just like day-to-day office work, is probably fine like like what you would get there of course maybe it's these

Cote:
[52:08] fake tokens or fake uh you know you're not actually paying the full or just.

Brandon:
[52:11] You just have to use a different model just have to like go down go.

Cote:
[52:13] Down the sonnet.

Brandon:
[52:14] Turn off the uh the whatever the the Opus, just use on it. Well, there was this interview I listened to with Derek Thompson. He interviewed, I think, Ethan Mollick. He was a professor and he wrote this new book. And he had this phrase, nothing new here, but I thought it was just really a great way to think about it. He just said something like, right now, AI is like a single-player game. Like the people feel the productivity gains themselves. But until it moves to like, and I thought like a Scrum is a good example. It's like, if you're still meeting in a scrum team and you're still signing a backlog, that like they're probably or that probably artificially limits what the advantages an actual corporation can get. Because like you can get through your backlog fast, but you still have the scrum meeting, still have the burn down charts. You still have all of that. So until such time that you sort of assign like an agent or a team of agents, like what was the work of a scrum team. It probably doesn't bubble up. You know what I mean? It's sort of like.

Brandon:
[53:17] Because at that point it's multiplayer. So it's like, unless you're, and this is kind of back to the Accenture thing. It's like, unless Accenture is willing to take a whole group and be like, no, this is just done by agents. And there's only one person orchestrating maybe 50, what was 50 Anderson, or sorry, not Anderson, Accenture consultants. Like you're not going to see it, right? And therefore you're going to clamp down. So it's almost like you need to go big right you need to be like listen we're going to try to do all the work of 50 eccentric consultants just with agents or you're seeing the opposite you go small and you clamp down on how much each person can spend because you even though they're more productive you're not going to get the money you're not going to get that productivity gain yeah.

Cote:
[54:01] I i think i think all most of the the commentary i see on this nowadays it's like it's like the uh it's it's identical to the commentary you would have in the digital transformation days where it was just like you solved one problem and now you still have six. And it's just sort of like until you solve all the other five problems, like you're just gonna like bottleneck in a classic like Goldrattian fashion of just like, well, now you just push, you push the bottleneck further up the line.

Cote:
[54:29] And that's, you know, whether you wanna call it, it's like we need more training or like we haven't figured out how to review this code well or whatever it may be.

Cote:
[54:40] Whenever we have the humans in the loop, you end up getting to this bottleneck stage, which, I don't know. I mean, I guess that's sort of like the non-technology thing, as you were saying, is to figure out the process re-engineering, to use a 90s term, is like, yeah, how do we reorient this stuff to take advantage of it? Which I like, I think the, uh, uh, the, did you look over this, uh, swamp stuff that, that, as we like to say, friend of the podcast, Adam Jacobs, uh, like his, his company, you know, the system initiative thing, it pivoted to this. I read through the documents today or yesterday and it, it's, it's, it's, uh, you know, it's, it's a frame, it's a lot more going on, but I'm going to simplify it it's like it's like a a way of specifying the the workflows uh, that that run to do something and having like a uh having a non-ai workflow execute it and then call out to ais for various things and linking them together and all the you know they use phrases like dag, for whatever that diagram bullshit is to do things and all all this kind of phrasing so you got all that going on but it is uh, you know it's like it's like a super harness for multiplayer mode essentially, and to control tokens and stuff, which is, there's another system I don't know very well that's, you know, from the Spring guy.

Cote:
[56:02] Rod Johnson, called Mbabel, Mbabel, I don't know. That's like, that's for the Java world only, I think. But it's a similar thing of like, a lot of these frameworks being built are like, instead of having AI run everything, we need to have like the control or a control thing on top of it that just calls out to AI. They like to use words like deterministically, to make it more deterministic and more controlled, which seems fine. It is like when I was looking through it, I was thinking. Well, one, it has an interesting license in the swamp thing. Interesting. It's not as far as I can discern open source, but then I think the command line thing is a GPL or like, but then you have to log into their system in order to use it. I don't know. I didn't verify all this. It just like, it's interesting, uh, more discussion on what's up with open source stuff. But I think if I remember from a couple of years ago, Adam was basically like, meh, open source, like we're going to have a model that helps us make money. It's not, it's not a big deal. So like, you know, there's a rhetorical move that, you know, that is that sort of like, I just refuse to engage with the premise of the questions. Moving on, right? Like, you know, which is a good, that's a good way to kind of wrap up 30 years of discussing open source business models.

Cote:
[57:20] But I think, you know, the more complicated thing is like I was reading through and it seems fine. And I was like, oh, I should have my robot like figure out using this thing. And then I was thinking, ah, such a big investment of time to like learn how to use this model and build this stuff. But if something like that was just sort of built in to, what am I trying to say? There is like a lack of a standard way of establishing just like, here's a repeatable workflow that you have. Like we've got skills and MCP servers and then there's, you know, the various things like you can set up rules and stuff like this, but it's almost like you just need a straight up like automation template of like do these things in this order with this just branching stuff and that's it. But I'm sure there's like 10, 20, 500 different people chasing after this thing here. But getting that, it becomes like an adoption hurdle at some point because it's a lot of like building a data center in Europe. It's a lot of work up front to like switch over to that system rather than just, you know.

Brandon:
[58:23] Doing whatever you're currently doing what you're saying that i think is really interesting because then that same interview i listened to i think uh, the author made this point and i thought it was really good just about like, like how much better these things get and it's like we as human beings see like a problem i was like oh the problem right now is like it's a little hard to do this whole thing but like, the pace at which the ais are getting better it's like the problem may go away like in six months you know it's like and that you know and i guess the you know i was reading a little bit about you know dario at anthropic like what was his contribution, to ai right because he's obviously done a lot and but you know to oversimplify it they said like he was.

Brandon:
[59:02] Kind of back when this was kind of getting going he was the one that just said like everybody is making this too complicated we should just give it more data we should just feed it more stuff like if we if we have more compute and we put more in it will be smarter right because all these other smart people were like, we should like, you know, do this and do this and do this. And that kind of proved to be right. That seemed to be like the thing. And so it's kind of back to Adam's thing. It's like, well, you can think about all of this, but at the same time, the frontier models are like, you know, we should, let's just make these things even smarter and smarter. And then so at some point you kind of get to the point of like, they don't really need any of that. And you know, some of the points they made were like, like all these things we used to do, like remember like prompt engineering was a big deal. It was like, you know, you're supposed to tell it like, I am this role and I am trying to do this. He was just saying, none of that even matters. You don't need to do that anymore. That's just a waste of time. That's not going to make yourself any better because it's just completely all that stuff that people were talking about, irrelevant. And so, I don't know. That's why I think it's very hard. It's just so hard to like, if you're building an AI tooling solution, it's like, I don't know. It's like, the AI will probably be able to do it Bye next week.

Cote:
[1:00:17] I think, I think, I think both of these things, like the, in, in, in, in whatever, in Babel and Babel and then also the swamp thing, like it's fun to read through it and see like.

Cote:
[1:00:26] Their view of AI stuff, just like the basic concepts and how things fit together. Cause it's, you know, it's kind of fuzzy how stuff, how stuff fits together. Like you took the, uh, you're, you're like a certified, uh, Claude, uh, engineer FDE master after reading through the documentation, I think.

Cote:
[1:00:45] And so you get you get like a you get one take on what stuff looks like, uh and it the the other thing that like uh that it it can be servicing is like i think, i've got a theory we were talking about the the artificial low pricing is i think there's like i've met at least one of them and i bet there's more people and maybe i should become one who are like i'm gonna call them like, ai pricing preppers and like this guy i was talking with he was like that there's no way he was like a $200 a month guy. He's like, there's no way this is going to hold. So I've been spending a lot of time making sure I have templates set up. Like instead of it doing reinventing things, that I lock down what it does so it doesn't have, whenever the prices do finally go up, I'll be able to keep operating like I am because I'm going to have like, you know, non-AI written stuff working. Because eventually it's going to cost me like, you know, $2,000 a week to do something. And so there is like this, that prepper mentality that like, and of course that's a value prop with a lot of these systems is you're not using tokens. And then you can also like.

Cote:
[1:01:48] Be deterministic you can make sure the same thing actually happens over and over again so you've got those those things going there but the uh, i don't know we'll we'll see as with most prepper people they usually just end up with a bunch of rotten food in their basement uh and so it's not a good basis for for doing these things i think there's plenty of good reasons to uh to to think about these frameworks and do stuff but like, probably the pricing will be fine that that would be the ultimate like uh the bubble would burst if like they remove these subsidies and people would just be like i'm out like i i can't afford it so you know then your entire like that, hmm i wonder if they'll put the like whenever one of these things actually ipos if they'll split the they probably will split the enterprise revenue, from the consumer revenue and then you can kind of do the math on like how much how much subsidization is going on and then if the if they blow, right rose raise the prices on the consumer stuff to like i don't know five hundred dollars a month, and more you would probably risk like losing the majority of that and then that would be enough to create or something even though it wasn't the big enterprise thing it would just be like oh we just lost all this money yeah because we actually have to charge what what it's what's worth now classic.com problem i think.

Cote:
[1:03:06] We'll see. I don't know. Meanwhile, you know, I think, I think ever since I've been alive and conscious, we've been, we've been complaining about like wealth concentration. We got all this money laying around that people aren't doing anything with. Buy tokens. Sounds great. Let's get all, get all those billions of dollars and just use it to fund tokens and everything, everything will be good. Uh well you know uh another, one of the reasons money is lying around all the place is to help support one

Cote:
[1:03:33] of the cherished institutions of most people's lives and that's bureaucracy, do we have any this any bureaucracy that we can deterministically execute on and do in a very token efficient way.

Brandon:
[1:03:45] I think so one piece of feedback brian graceley said the barkley marathons are always in the same woods in tennessee near a prison which was the original quote escape run But the path through the woods is different every year. So I think you were asking about that. So that's the whole story. If you want to run it, you're actually going to escape for prison.

Cote:
[1:04:03] Yeah. Yeah. There's, I think I referenced this, but there's, I learned about it. There's a Tyler Cowen interview with whoever, the guy who does it. And it is like.

Brandon:
[1:04:12] The guy that comes up with it?

Cote:
[1:04:13] Yeah. I forget what his name is, but you know, it's just great. You couldn't find two people who are. Further from each other but also like very similar and just like whatever but he just sounds like some old coot from the Dukes of Hazzard just going over like his great running stuff but it seems like a challenge Well the only thing that we can hope.

Brandon:
[1:04:32] Is that Tyler Cowen enters the contest and tries to.

Cote:
[1:04:34] Complete it that would be how that needs to end I think next year we need to round up a sponsor and we'll send Matt Ray for the software to find talk Matt Ray definitely.

Brandon:
[1:04:43] Should be the one that tried to.

Cote:
[1:04:44] Do it Yeah So we need to work on that.

Brandon:
[1:04:47] That's right Also if you like stickers just send your postal address to stickers at softwaretofindtalk.com happy to send you a sticker anywhere in the world Thanksgiving's coming up Christmas is coming up you know you need some stocking stuffers you need some extra stickers so email me I'll send you plenty of stickers.

Cote:
[1:05:01] Yeah put a sticker on the turkey I think I think that's what you want put it on a ham a turkey you could put it on your, your sweet potatoes I wouldn't recommend replacing the marshmallows on top of the sweet potatoes with stickers marshmallows probably wouldn't taste good yeah,

Cote:
[1:05:16] that's hmm I should go get some dinner. Well, there's fewer conferences than there usually are. There's VMware Explorer London coming up November 18th and 19th. I'll be speaking there. You should come check it out, of course. I'll also be at Cloud Native Denmark November 19th. Well, I will be speaking. There's a discount code you can use, CSVFTW.

Cote:
[1:05:40] I don't know how much it gives you off, but try it out. Just buy a ticket and email us and tell us how much you got off. There's also build stuff coming up in Lithuania, December 2nd and 4th. And I haven't really firmed it down, but I'm going to try to be a configuration management camp as always, February 1st and 3rd. Highly recommended if you like the kind of thing. And of course, there's Scale24x in Pasadena, California, much beloved here on the podcast, April 1st to 4th. In 2027, just like the other one, no time machines involved. Now, uh, every time, every week we come up with recommendations,

Cote:
[1:06:17] Brandon, and I often have to use a time machine of the mind to go back and think about things I've done that week to recommend. But I know you, you just, uh, you know, you probably write this down in your little pocket notebook whenever it occurs to you. So you don't have to use a time machine, but what's, what's your recommendation this week?

Brandon:
[1:06:34] Well, funny enough, I'm going to go back to back and recommend the same thing twice in a row. So last week I said, I recommended the trailer for You Can See Everything. And this is the Nathan Fielder movie about Theranos. And so I did. I said I think I would go see it, and I did. So I went and saw it. The movie is three hours, five stars. Thing, absolutely excellent. I loved it. I loved it from start to finish. I think it is best if you go in and don't know much about it. Like, exactly. Not so much about the story, because there's not that much about, like, if you know about Theranos, that's all good background. Like you know but even that for the most part that's really not covered like like you know what was going on that part is sort of to the side so it's like a.

Brandon:
[1:07:17] And it's like so hard, like everybody said the same thing. So I'm kind of repeating what other people said. It's like, it's like hard to describe like what it is, like what you're seeing, because there's like many levels of it. And I did something I haven't done in a while. I was like, man, I just like, I need to talk about this movie. I was like, I was almost like, we should maybe do an hour long podcast on this, but no one probably wants to hear it. We're not a movie podcast, right? But I got out of this and I texted your friend and mine, Mr. Chippy. I was like, have you seen this movie? And when can we go to lunch? She's like, I haven't seen it. He's like, I want to see it soon, but let's go to lunch on Friday. I'm like, no, I need you to see this movie first. And then we're going to lunch. He's like, okay, I'll see you this weekend. We're going to lunch on Monday. So this is my recommendation is like, you should go see the movie, but ideally either know someone that has seen it or go with someone that likes to talk about like movies and film and just people. That way you can just get it all out of your system. I have so many takes. I have so many thoughts like in my head. That I want to talk about. So I'm going to be doing that with a good friend, Mr. Shippy, on Monday. That's my only suggestion. Go see the movie. I get to see that. But you got to be with someone.

Brandon:
[1:08:26] That wants to talk, like everybody knows, like you've got like people in your life, you got your like movie discussion people, like, you know who those people are. Some people are not like that. They're like, oh, the movie's good. Now let's go get lunch. It's like, no, no, no. I'm on a deep dive on this for like an hour. So really good, really thought provoking. I'm interested to see what the award, the various award academies, I guess the emotion picture and other, like what they do with this thing. Cause it's sort of like, I think it deserves some recognition, but I don't know how they're going to recognize it because it is in self it's like its own little like if there's ever a time to be like we're not sure what category to give put you in but we're going to give you an award this would be the year for that so there you go go check.

Cote:
[1:09:06] That movie out that's that's that's very similar to how people talk about the uh the dungeon crawler carl series like there'll be people who like have i've encountered in my little pool of people who uh a handful of people who read it, like they'll hear that you're reading it and they've read it all and they'd be like oh you got to tell me when you're done i we had to go to lunch i got a lot of a lot and then you know you got to they want to know the progress and where you are and what doing and they'll talk about because they don't they're nice they don't want to spoil it for you but they're just so excited to like talk about what's going on and they're like i don't even really want to tell you, what happens in this book because it's going to sound stupid but you got to just it's it's great it's hard to explain even.

Brandon:
[1:09:45] As you're saying this like there's parts that there's like a thing about this this film i want to say because i think i encourage people to say it but Like, I think I don't want to say it because like, even knowing that this thing that I'm not going to say exists, I think kind of takes away a little bit from it. So that's why I'm like... It's like, I can't really tell you anything. I'm like, you got to go see it. And then we got to go to lunch. Maybe I'll create a Slack channel. Maybe I'll create a channel for me and be like, only people that have seen the movie can be in here. Because otherwise, no one. And I'll almost like have like, in fact, I'm going to have like a little quiz. Like you got to answer questions, right? Before I even, to prove to me, you actually know it. Otherwise I don't, because I don't want to steal your joy. Even if you're like, no, I'm never going to see it. I'm like, that's okay. You can't be in here. I can't, I don't want to be responsible.

Cote:
[1:10:26] We need like a, we need like a spoiled channel where you can just talk about spoilers and anything.

Brandon:
[1:10:32] It's like, and I don't want to like, no one in here that hasn't seen it. I don't want to, if you ask one question, you act like you don't know I'm kicking

Brandon:
[1:10:37] you out. You're kicked out of that channel. You're just back to the general channel.

Cote:
[1:10:41] Love it. Well, I have a practice recommendation that I think I've kind of solidified back to AI usage. So, you know, as I say, I don't actually like write apps. Apps are written for me, code is written for me, but that's not what I use AI for. So I have a lot of text handily that is generated reports, summaries of things. And what I've found, it's kind of like my presentation thing, is you should figure out, you should spend some time coming up with a template that you really like. And, you know, because if that you're going to read a lot and like, it's almost like, I don't know when this happened, but at some point the solarized theme came out for everything. And now everything can be solarized. And so think about. You could even just have a solarized template, whatever that is. But like, think about, I don't know if it's a design system, but whatever that look and feel of the reports and the text and the things you want generated are. Because, you know, you can have it just output in the window and tell you stuff. You could have it output a markdown file, which is sort of fine. You can have it output its own like clod style. Like if you run the insights thing, it has its own cloddy style that it does stuff. But you should really think about like what do I want this document to look like and format it and spend a lot of time building up a skill that's like here's how we do output, in that way and then once you develop that you can spread it across to all the different uses that you have you can even.

Cote:
[1:12:09] Like you know most applications you use probably accept a custom theme and you can have a port that theme over to that and then you start you know you're deep into this land of just going AI crazy customizing everything, but I think it, uh, it brings some consistency. So the one I have, you know, I have a lot of fond memories for like the web in 1998. So I basically have a theme that makes it always look like that. I don't have like the, the people peeing on each other or peeing on a line or anything or the construction stuff.

Brandon:
[1:12:37] Uh, not yet. Go die.

Cote:
[1:12:39] Yeah. Yeah. But I have, I have like a mosaic. I did a, I did a new version of it yesterday. So it used to be a mosaic 1998 and now it's Mosaic 1999. But, you know, develop whatever style you like and just have it output things in that style. And I think it's a little happiness thing and then you have it do a skill, you evolve it. You know, as with all this stuff, the other general principle is like, You can actually just make changes all the time to it and make little tiny modifications, and it's fine. And you can enjoy that for your, maybe that's why I'm spending $5,000 fake dollars a month on AI things. But we'll see. Well, you know how much you've spent on this podcast? Zero, listener. But you've gotten, I'm going to say, at least the $200 a month max outcome and benefit from listening this entire way. Now if you want to get links to the conferences we mentioned uh articles we talked about, i guess there's really no link to the recommendation i just had but the uh to the recommendations, uh and a bunch of things we didn't talk about you can go to softwaredefined talk.com slash 595 and uh find the full show notes and with that thanks for listening bye-bye, yeah yeah so there's there's your friedman analysis that you know research by taxi driver.