0:00 Scott What changed? Well, we know what changed. Agentic software factories.
0:04 Brad People aren't reading the code and it's all about getting humans out of the way of the code. That's a pretty big change.
0:27 Scott Ladies and gentlemen, welcome to this week's episode of how many CTOs does it take to screw in a light bulb? My name is Scott Porad and my trusty co-host.
0:37 Brad It's Brad Haftig out.
0:39 Scott Hello, Brad. How are you doing today?
0:40 Brad I'm doing really well, Scott. How are you doing?
0:42 Scott Well, I'm doing pretty good. This week, I sent you a link to a summary of a talk. about the state of the tech industry by someone whose name I don't know how to pronounce, Gergurly Oroz, who writes the programmatic programmer. And
1:02 Brad I just
1:02 Scott you probably...
1:02 Brad call him the pragmatic engineer. That way I don't have to figure out how to pronounce his name.
1:06 Scott And you promptly said, hey, I sent this to you last week.
1:10 Brad Yep, I did say that.
1:12 Scott What did you think of that video I sent you?
1:17 Brad I thought it was great. I sent it to you a week earlier.
1:20 Scott I think you have a time machine and you went back and inserted it in our text thread and got it in there.
1:27 Brad By the way, I just saw a meme on the internet about time machines. where it's like you can't just have a time machine because you know our galaxy is is moving in in the universe so if you were to go back in time like the planet isn't where it was when you left so you got to be able to travel in space and time at the same time so i don't have a time machine because that might leave me suspended in space
1:59 Scott Time travel is a complicated topic.
2:02 Brad Are we going to try to tackle that one on this episode?
2:07 Scott No, but it reminds me of this little clip I saw this week from a professor teaching quantum mechanics. And I'm going to paraphrase it, but he said something along the lines of this is, you know, we're going to start quantum mechanics today and we'll be discussing this for the rest of the semester. This is a professor at Yale. Yale, I remember seeing that. And he said, you know, the bad news is nobody understands how quantum mechanics works. But the good news is you don't not understand it yet. And so right now I'm the only person that doesn't understand it. But by the end of this semester, all of us will not understand it. Okay, so yeah, time travel, let's talk about that. We are actually talking about the state of the tech industry,
2:53 Scott and if we could go back in time, did we live in a happier time five years ago?
2:59 Brad Ooh, in the tech industry?
3:01 Scott Yeah, happier, better. Well, how was the time? Obviously, the time was different because of AI, but you know. We recorded that episode last week with Kent Beck, and he was so focused on the experience of the people and the engineers, it got me to thinking during that episode, and I'd bring it up because of the time travel question, was that a better time for engineers or is it getting better for us?
3:33 Brad man i that is you you've stumped me on that one i mean i think if we really go back five years i think that it was the middle of the pandemic and like there was a lot of a lot of things that were not great about 2021 so so
3:48 Scott Well, yeah, okay. Okay. I
3:52 Brad i'm just putting that out there
3:54 Scott will say these things about 2021. Uh, you and I worked together in 2021,
4:00 Brad Well,
4:01 Scott so
4:01 Brad that
4:01 Scott it was, it was definitely a better time.
4:03 Brad was bonus. That's true.
4:06 Scott Uh, the tech market was in fuego.
4:10 Brad Yeah.
4:11 Scott So people were getting paid.
4:13 Brad By the way, as a hiring manager, that wasn't exactly my favorite time, to
4:17 Scott That's true.
4:18 Brad be perfectly candid.
4:20 Scott There was something else I thought about it that might have made it better, but I lost my train of thought. So that's okay. I don't know. I don't know if it was better or worse. It certainly was more certain, but to some degree it was a little dull.
4:31 Brad I think that if we try to interrogate, I'm assuming that you're asking the question to try to interrogate. negate both what we talked about with kent uh beck last week as well as kind of this this um state of the tech industry in 2026 and man i i'll admit i am biased on this scott but i feel like this is the golden age of building stuff and
4:57 Scott Yeah.
4:57 Brad it's so much easier to build stuff now so i will just say like that part of me thinks if i could if i if i was back if if we were back at 98.6 together and we had some of the tools that we have right now i feel like i feel like we would be There were so many trade off conversations where it was like, do we have enough people to do this? We have these ideas that we want to implement. We had a product team that had we had two different we had kind of like two different pieces of the product org trying to go after two different things. And we were constantly kind of challenged with the classic resource constraints.
5:33 Scott Mm-hmm.
5:34 Brad And man, I feel like I feel like these tools would make some of those hard challenges a lot easier. So that's the one thing I would say about kind of trying to put my if I was to give myself the tools that I have today or that we can give engineers the tools we have today, I feel like, man, I feel like some of this I feel like some stuff we could do, we could do better.
6:01 Scott Yeah, it's interesting. If 98.6 were today, it might have ended up a viable business because
6:08 Brad Yeah.
6:09 Scott the core of that business, the AI part of that business was, I mean, it was just a little too early.
6:19 Brad Yeah.
6:19 Scott And now, you know, stuff that they were trying to build, we were trying to build, you know, it was kind of. pushing it pretty hard is kind of you just get a clock or all the models provided for free now so
6:32 Brad Yeah.
6:33 Scott now you'd be in the business of just like using all these frameworks to make an slm small language model right just focused on health care it'd be a lot easier
6:40 Brad Plus, I also think that like the fact that it had a user interface, like a big piece of that, there was a whole teams working on the user interface for these things and that stuff is so much easier to build now. So I just
6:51 Scott Yeah,
6:51 Brad feel like,
6:52 Scott it's
6:52 Brad yeah.
6:52 Scott true. Well, I want to add something to it. I would say that at this moment today, the industry might be a little worse because we're experiencing what the capitalists call creative destruction. But I would also say our industry was fat and this is thinning it out. and when it comes out the other side it will be leaner and better that was one thing i saw during the pandemic for instance a lot of businesses got hit really hard during the pandemic that was terrible but i think many of them came out much stronger as a result of it some of them didn't come out at all but a lot of them came out much stronger because they really trimmed
7:41 Scott You know, this is an expression people use about headcount. That's what I'm really talking about is I'm not talking about people. I'm just talking about everything. Like they just trimmed the fat, the things that really weren't having an impact and weren't making a difference out of necessity got thrown by the wayside and companies and teams just got a lot more focused and a lot stronger.
7:59 Brad Yeah, and I mean, if you think about the creative destruction that's happening right now. Again, like we record this and then we release later, but, you know, recently in the last week, the reverse engineer anything. Open source project came out.
8:14 Scott I'm not familiar with that. I just saw the Adobe thing
8:17 Brad Yep.
8:17 Scott that happened, and then I've blinked for a minute because I was away at a work on-site. Did I miss something else?
8:23 Brad Yeah, so basically there's a new open source GitHub package that's called Reverse Engineer Anything. I think it was what was used by the guy to reverse engineer the entire Adobe suite. If not, it's a comparable open source project. project and basically yeah basically you give it any any compiled application any mobile app any any web app that you have login credentials to as a user and it goes to town and it basically builds you a copy of it
8:52 Scott What do we think about, let's talk about that for a second. What do we think about that? Just in terms of like the legalities of it, not the coolness of the tech. I'm not sure what to make of that.
9:04 Brad It's it's definitely like if you use it if you log in with your you know, Strava account and then use it to copy Strava, then you're surely violating the terms of service of the Strava account. Like all the terms
9:16 Scott Yep,
9:16 Brad of service for every website says you can't use it to reverse engineer and stuff. So from an individual person's perspective, they're violating the terms of service for the piece of software that they're using in all likelihood. Now, I don't know how it's questionable how enforceable that is, right? Right. Like those those there's a whole bunch of like to your point, like we're just in this this era of lawlessness, maybe.
9:42 Scott Yeah.
9:42 Brad But, you know, like one hot take I saw from from an artist was, aha, Mr. Software Engineer, you're finally getting to taste your own medicine.
9:51 Scott Yeah. Yeah.
9:52 Brad So, I mean, the LLMs were based on just ripping off everybody else's intellectual property. Now, you know, it's coming for the next wave, you know, but.
10:04 Scott it's coming for everybody well it reminds me of this book i have i don't know where it is actually i do know where it is i gave it to one of my son's friends you know my kids are got one in college when i was college so they're into all sorts of really heady ideas and this book was The title of this book was called Fully Automated Luxury Communism.
10:24 Brad Wait a second. You recommended that book to me.
10:28 Scott did i i
10:29 Brad Yes.
10:29 Scott probably did because That's it's it's
10:31 Brad not your son's
10:31 Scott very provocative no
10:33 Brad book. That's your book that you recommended
10:34 Scott i gave
10:34 Brad me like
10:35 Scott i gave
10:35 Brad two years
10:35 Scott the book
10:35 Brad ago.
10:35 Scott to what i i gave the book to one of my son's friends because he was all into these different ideas and it's this book talked about it how these you know things are everything's becoming free right
10:47 Brad Yep.
10:47 Scott and um okay so we're off track okay i want to bring up so so we're in this creative destruction of our industry i think it's going to get It's going to get better, but it's a little rough right now. I do remember the thing that I lost my train of thought about the pandemic.
11:02 Brad Okay, tell me
11:03 Scott One thing that was better five years ago was remote work. And what I mean by it was better was suddenly I got to be at home and it felt really good. But I will say five years later, I miss being with people.
11:21 Brad That's interesting.
11:22 Scott I wish I could find some quantum way to feel to have both.
11:26 Brad It's funny that you say that because I had already been working in a fully remote world for about seven years before the pandemic. And so when the pandemic, it was funny, actually, the first year of the pandemic, I actually took like a year long sabbatical. I didn't know the pandemic was happening. It's like I start, I left my CTO role at the VR company that I was at. in december and then
11:51 Scott Mm
11:51 Brad started
11:51 Scott -hmm.
11:51 Brad and then it was like you know the first couple cases started coming in like February
11:57 Scott Mm-hmm.
11:57 Brad right and
11:58 Scott Right.
11:59 Brad so then that year I was out in the mountains away from I was near people because I was with my mountaineering friends but I was away from the bulk of people you know it was a perfectly good place to be out in the open air you know when a air air born virus is out there you know as opposed to being locked into room so I had spent a lot of time with a limited number of people during the pandemic.
12:23 Scott Mm-hmm.
12:24 Brad But for like seven years before that, I had lived in this remote world. And so I was already pretty comfortable with it. And, you know, there's elements of what we experience online where I already just kind of knew about it.
12:40 Scott And I had done a... Company from 20, let's see, 2007 to 2012 that was mostly remote. We did have an office in Seattle where many people. Well, the engineering team, that's what the deal was. The engineering team was remote. So our engineering team was spread all over the country, but the rest of the company worked in the office. And so I'd had some experience with it, and I mostly went to the office at that time, but
13:12 Brad Is that easier?
13:13 Scott I had experience. That was Cheeseburger. And so that was, I had experienced the tools and what it was like to work in that rhythm of always being on a video call, that kind of thing.
13:22 Brad Yeah,
13:22 Scott I hadn't had the solitude of just getting up and going to work in my pajamas.
13:27 Brad yeah. But yeah, to your point. Well, so now we're like really far afield from the pragmatic engineer's assessment of the
13:37 Scott Well,
13:37 Brad industry,
13:38 Scott let's get
13:38 Brad 2026.
13:39 Scott into it. What's he have to say about it? Should I give people a little bit of a summary as we go through
13:43 Brad Yeah,
13:43 Scott the whole
13:44 Brad why don't
13:44 Scott thing?
13:44 Brad you hit the summary points and then maybe kind of walk through them and decide which ones you want to pull the threads on.
13:52 Scott Well, I had probably Gemini. In fact, I know it was Gemini did this because I fed the video. to Claude and said summarize this and Claude's and it was on YouTube and Claude said I can't and I thought well that's interesting and I gave it to Gemini and Gemini did it lickety split which this is a different topic but it made me wonder is like huh does Google have a competitive advantage in the model business because they can start cutting off all this data that other people need but nevertheless okay he talked about what's changed what's not changed what What broke and what's next? And here's what I think, how about we approach it this way. We'll go through each section. I'll list the two or three things that the pragmatic engineer.
14:45 Scott uh listed and then we can pick one to discuss or just go for it how's
14:49 Brad Yeah,
14:49 Scott that sound what
14:49 Brad that sounds great.
14:50 Scott changed well we know what changed agentic software factories obviously agentic code production and the shift to open models is what's changed in the last year anything that stands out to you
15:05 Brad Yeah, I mean, I mean, one of the things that he kind of the points that he makes in the in that section on agentic software. like software factories as he basically kind of says like in these you know he kind of surveys big big tech he surveys a bunch of the frontier model companies themselves he you know he kind of is embedded in places like uber and linear and some big big software companies like that and he's like makes the point that like the vast majority of those companies like no one's reading the code anymore and so like that's what he talks about in the agentic software factory section it's just like yep people aren't reading the code and it's all about getting humans out of the way of the code that's a pretty big change it's a pretty big change yeah
15:45 Scott Yeah, that reminds me of a message you sent me today, actually, from Linus Torvalds. I would imagine that thing's flying around the internet.
15:56 Brad I bet it is. It's
15:57 Scott It
15:57 Brad got to
15:57 Scott flew
15:58 Brad be right.
15:58 Scott around from you to me where I guess somebody that Linus works with.
16:03 Brad I was assuming it was somebody like on the, you know, the Linux core mods team or something was like. beating somebody else up for submitting an obviously AI-generated code or something.
16:17 Scott Right, right. To which Linus said, I don't understand this whole idiotic concept of don't use AI for X. he pulled no punches which i loved i love that he pulled no punches i the interesting thing in this so yes people aren't reading code here's the interesting thing to me in this section it's a little bit more of a business thing than an engineering thing although i think it has implications for cto's because especially of small and medium-sized companies All of the big players are moving to open models. I just read today or yesterday, Meta's usage of Anthropic is down by 50% because they're all seeing the cost and they're all realizing that it's going to be cheaper for them to build their own models, especially if they're kind of smallish language models for doing programming,
17:06 Scott right? And they have their own huge code bases that they can train it on. And that's interesting because then you get like, I feel like where I am, I don't have the resources. to go build our own or go manage our own open weight model right we can go find an open source one and do it but like getting all of our engineers that's like a whole other system we'll have to be responsible for and like what we do in companies the size of ours is you know we outsource to SaaS services maybe or as much as this stuff as we can because we don't have the resources to do it so I think that's an interest that's like it's the thing that changed this year which I think is going to be a thing that continues continues to change in the coming year.
17:47 Brad Well, and I think that if we go back a couple months, I was telling you how I was experimenting with my own local inference. I actually just the newest, well, not the newest, but Mac Studios are now purchasable again from Apple with real amounts of memory in them. So I've got a Mac Studio with 256 gigabytes of memory on its way here. here i plan to run uh open weights model on that one and the benchmarks show that that thing that the the latest sort of like open weight models that can fit in that frame of memory perform as well as like opus 4.8 and i don't know if you remember but back in january of 2026 like 4.8 comes out and everyone was like oh my god like we're done like no more programmers you know and so
18:38 Brad So the reality is that, yeah, the frontier model keeps moving forward, but the open weight models that you can run on your own infrastructure, you don't have to necessarily, you know, optimize them for your own needs. Like they're as good as the frontier was a year ago. And what is the big change that's happened in the frontier? Like one of the reasons I think that we're shifting to open weight models is not. not because the people like meta are saying like oh we can put our own engineering team on making an optimized version that we don't have to then pay anthropic i don't think that's really what that's about i think what it's really about is that other the thing that the frontier models are good at now is like long running unsupervised tasks okay but if you already were using sdd and you already were using plan mode and you were already kind of like
19:30 Brad You know, in your agentic software factory, you had kind of broke, you were breaking things down anyway. You don't need long running tasks. You can get the code quality that you need out of a, out of a, you know. a 4.8 class model for the coding parts and so why pay the why pay the premium and if you and even if you don't want to run it locally just go use open router because open routers got the big you know the latest you know
19:59 Scott The latest models, the latest open weight models, I'm not even going to name them here because it doesn't matter because by the time this episode comes out, there'll be three more open models. But the latest open models are just nipping on the heels of the Frontier models. And so that's what I think the simple answer is. So I would say to the CTOs out there, hey, get one of your get one of your teams that, you know, that you trust the most that sort of like really gets it that's fully AI pilled and tell them like, hey, why don't you shift over to using OpenRouter and OpenCode or Pi or one of the other non-cloud code harnesses? do a couple sprints do a couple sprints where you use those models um because they're a tenth the price they're a tenth the price of of the anthropic and the open ai models and so and have them run a couple sprints with it and i bet you they're going to come back and basically report back that there's like no significant uh performance difference with respect to their ability to build software faster than they can humanly code review it
21:04 Brad Yeah. Here's a question. action of going and trying the different thing there's a life cycle engineering teams go through with AI right they start with a vibe coding and they get into plan mode and they start to put some more structure around what they're doing in terms of maybe shared skills across the organization and putting different things you know just start to put some scaffolding around it all right And my question for you is my when I heard you talking, I was thinking to myself, yeah, that's exactly what we should do. And then I thought to myself, now let's get all the scaffolding in place good first. Like we're not we're not totally I feel like we're not totally as buttoned up as I want to be. And then go do that. I guess and then go do the cost saving maneuver.
21:51 Brad Like let's go do let's go do the productivity enhancing maneuvers first and then the cost saving maneuvers second. Would you do that or would you just or you just do them all at once and it doesn't matter?
21:59 Scott I think it just depends on how much you're spending on tokens right now. Like if you,
22:03 Brad Yeah.
22:03 Scott like... I think that if you're not feeling the, I mean, okay, there's two reasons to go, there's two reasons to go open models, well, three reasons to go open model in my mind. One reason is like the frontier model is just charging too much for their tokens. And so you just want to save. So if you're, if you're not feeling that bite, then don't worry about that piece of it.
22:21 Brad Mm
22:21 Scott Okay.
22:21 Brad -hmm.
22:22 Scott Second reason is there, I mean, the frontier labs have demonstrated that they are. uh looking at everything everybody's doing and next thing you know the thing that those people are doing frontier labs are then releasing themselves right so
22:37 Brad Mm-hmm.
22:37 Scott i'm not going to necessarily go so far as to say that they're stealing everybody's ip but yeah they're not to be trusted with all your api calls like they're just not like they've demonstrated that they're not they're not trustworthy stewards of of stuff so call that data sovereignty or at least call it like those guys aren't you know they're not your ally right
22:59 Brad Mm
22:59 Scott the
22:59 Brad -hmm.
22:59 Scott third
23:00 Brad Yeah.
23:01 Scott reason is a variant of that second reason which is hey if you really want data sovereignty like you don't want anybody looking at your at the stuff you're pushing through the models then set up on-prem inference set up
23:15 Brad Mm
23:16 Scott local inference
23:16 Brad -hmm.
23:16 Scott or you know and and by the way i'm certain that like you know that's what jensen's plan is like nvidia is going to start rolling in racks to fortune 500 companies with nvidia chips and open weight models and he's just going to be selling like refrigerators of on-prem inference and those
23:37 Brad Yes.
23:37 Scott are all going to be open weight So that's that's my prediction that for 2027, that's we're all going to be hearing about, you know, NVIDIA just selling like on prem inference.
23:50 Brad Yeah, you know, there's an expression in the real estate business that all value flows to the land. And I think maybe the analogy in our business is that all value flows to NVIDIA.
24:04 Scott Yeah, yeah.
24:06 Brad All
24:07 Scott So there you go.
24:08 Brad right.
24:08 Scott I think we've tackled the what changed. That's a pretty good summary of what's changed.
24:12 Brad Okay, let's talk about what has not changed. Core software principles. Traditional software engineering fundamentals, system design, resilience, security, edge case validation. These are all still things that need to be done. I don't know if they need to be done by humans, but they all are things that need to be done well.
24:36 Scott Yeah, this is by the way, I know we talk about all the things that what has changed before we kind of drill down in them, but. You know I've been saying for a long time, this is all about if you've already got a great SDLC, it's going to be a K-shaped economy. Like the people that get the fundamentals, they're the ones that are going to just take off with AI.
24:55 Brad That's right. So that's one thing he talks about is core software principles have not changed. Organizational friction has also not changed. Team alignment, clear product requirements. strategic execution remain human challenges not ai challenges and what i have to say to that is ain't that the truth brother i mean it's
25:19 Scott well i mean it's all like just because you can build something really fast doesn't mean you know what you want to build and
25:26 Brad true or that you agree on what that is yeah
25:28 Scott you agree on what that is yeah totally It's funny, by the way, I'll say like on that one, I've started like correcting anybody who used the word hallucination because
25:39 Brad Mm-hmm, mm
25:39 Scott I'm just like, no, the model didn't hallucinate that the model filled it in the model filled in.
25:45 Brad -hmm.
25:46 Scott You're just calling it hallucination because it used to be that you as the engineer or you as the product manager or you as the program manager used to fill it in when the spec wasn't clear, when the VRD wasn't clear or didn't exist, right? Or the PRD said a bunch of words that... half the team didn't really understand what they were because it was like some niche product market or something like that and you filled it in all right well the model's just filling it in it's not hallucinating it's doing the same thing you did
26:18 Brad Okay, this actually reminds me of something. It's not in the what's not changed category. It's actually in the what's changed category, and it didn't get talked about in this talk. Or maybe it should be what's changing. I'm going to just call it org structure and team structure.
26:38 Scott Yeah.
26:39 Brad You know, like. I'm now starting to think like the ideal team is going to be something like a product manager and a designer and two engineers and
26:49 Scott Yeah.
26:49 Brad like get out of their way and that's something we're starting to experiment with you know it used to be the case that the way to get a project to go faster This is an interesting Conway's law kind of thing. Conway's law, Parkinson's law stuff here, right? Like on the one hand, it used to be, and what's the other one about the mythical man, Brooks law, right? It used to be the case to get a project to go faster, you put more people on it, right? And then you started to run into the Parkinson's law and Brooks law and Conway's law's problem. And now it's like, I think there might be this new thing. Maybe I can get my own law. Maybe this can be. can be called poor as law that'd be cool i would love to have a law named after me um but i bet there's already a law for this i think it might be turning in the next year such that maybe if it's already turned i don't know the way to get something to go faster is to put less people on it which is a crazy thing
27:44 Scott Well, we know that we know that, you know, the number of communication points in a system. make the system more complex and therefore make the system uh you know slower basically um
27:58 Brad Well, and then it begs the
28:00 Scott but
28:00 Brad question,
28:00 Scott by the way i i'm just going to say like i i i agree with you like the the how to to me this is the organizational friction what has not changed is organizational friction right and so the more people you have on it the more organizational friction you have or if you've divided up your responsibility where you're Your product managers, you know, report into a chief product officer and your engineers report up into a chief technology officer. And and if the two of them aren't aligned, you know, then the orgs end up not being aligned or something like that, you know, or, you know, two product managers working at cross purposes from each other trying to leverage the same engineering team. Right. right so that squad-based model that idea of like let's let's make these but you said something really important is like oh you got a designer you got a product manager and you got a couple engineers and then what did you say and you get out of their way right
28:53 Brad Yeah.
28:54 Scott like I mean, really, to get out of their way, you also have to kind of like give them some kind of accountability around the business outcomes or something like that. Right. So, you know, I think I think what we're going to see and, you know, I see this happening already. This is out in the zeitgeist now, organizations that are like basically talking about like kind of. creating essentially little little business units you know the squads almost become business units you know and i think there's i think i think that's i think that's a you're right that that's coming yeah
29:27 Brad But then I think it kicks in one of the other things that hasn't changed about engineering fundamentals, which is system design. Because what I've seen in the past when you do these small teams that are really independent going and doing their own things, that's how you get to the ball of mud. In fact, if you asked me to build a ball of mud as fast as possible, that's how I'd do it.
29:51 Scott Create a whole bunch of teams that don't report that aren't accountable to each other.
29:55 Brad I'd create seven teams. And I would, each team would have their own product manager and designer and engineers, and they'd be working on one monolithic piece of software. And I think that's the fastest way to get there.
30:10 Scott Here's a crazy question.
30:11 Brad So, yeah.
30:12 Scott What if what if you literally just like what if you when we talk to Greg Conklin about his experience at. Not at Dropbox, but at Groupon, right?
30:23 Brad Yeah.
30:23 Scott Didn't he tell us how basically they just acquired all these little companies and instead of trying to merge everything into the same code base, they just ran the separate little code bases in parallel and they just had lots of small teams that were just running their own little stopper stacks.
30:41 Brad I think that's right. And I think what you, I think I'm just. thinking this out loud for the first time but it's like as a now look i work on internet software right that's what i've always done is internet almost entirely consumer internet my entire career but i think the way it's going to have to go if i'm just thinking this through a quick a really strong sort of foundation platform right that has some core services that are available to Available, right? And the core services I'm thinking are stuff like account service, if there's any commerce involved, the commerce type of service, you know, transaction, you know, credit cards, stuff like that. Some core shared services, authentication, I guess that's sort of some key services. But then on top of those,
31:27 Brad yeah, like everybody's just going wild doing their own thing. And you'll see whether it's by subdomains or subdirectories, you know, like people are just doing their own thing.
31:37 Scott Okay, so what has not changed? We covered it. Like software engineering principles matter and organizational friction is still kind of the thing that AI can't solve.
31:46 Brad Well, here's what I'm thinking, Brad. I'm thinking we discussed what's changed and what has not changed, but we haven't discussed what's broken, what's next.
31:55 Scott Yeah,
31:56 Brad Maybe
31:56 Scott I
31:56 Brad we should break this into a two-parter and we should discuss those next week.
32:00 Scott think that you're right, Scott. Why don't we do that? Why don't we wrap up this episode on what's changed and what hasn't changed?
32:06 Brad What do we say every week? We say smash that subscribe button, rate, review, and tell your mom.
32:12 Scott and tell your mom and
32:13 Brad All
32:14 Scott we'll
32:14 Brad right.
32:14 Scott join you guys next week on another episode of How Many CTOs Does It Take podcast where we talk about what broke and what's next.
32:21 Brad All right. See you, Red.
32:23 Scott See ya. This episode of the How Many CTOs Does It Take podcast was produced and edited by Saffron Heptacow. Our theme song was AI generated by Brad. Please follow us on all the social media platforms at HowManyCTOsPod. You can find us on YouTube, LinkedIn, Instagram, Facebook, Threads, X, TikTok and Patreon or at our website at HowManyCTOsPod.com. Dot com. And remember to rate, review, and subscribe on the podcast platform of your choice. Thanks for listening.
