00:00.50 James Welcome back, everyone, to Merge Conflict, your weekly developer podcast. Talk about all things in the world of software development, AI, and all those goodies. I'm one of your co-hosts, James Montemagno. And with me, as always, the one, the always, the illustrious, Frank Kruger. How's it going, buddy? 00:15.06 Frank Hey, James, I am the other co-host. I've never been called illustrious before. Thank you, sir. I appreciate that. I will think for the next 30 minutes about a good adjective to apply to you, the great James. 00:25.74 Frank See, I don't have your vocabulary. 00:26.08 James There are none. Ooh. 00:29.14 James i bought um I bought something on Amazon. um 00:32.06 Frank Oh, my favorite topic. 00:33.50 James So I have a i have ah one of those smart lock thingies, you know? 00:37.39 Frank Oh, 00:37.40 James And the smart lock thingy takes four AA batteries. And I had put it on the, yeah, I know. 00:41.33 Frank it's not so smart. 00:43.82 James I had put it on the IoT t Wi-Fi and it was really struggling, like the 2.4 band. 00:48.02 Frank Oh, yeah. 00:48.86 James And I don't know, because I have like the mesh network, all this stuff. So it ran out of batteries really quick. and i was like what the heck? That's crazy. It's only been two weeks. That's unacceptable. Um, and, but I was like, I'll get some rechargeable batteries. 01:00.75 James That's what I'll do. 01:01.42 Frank Yeah. 01:01.45 James Cause then you can, you know, swap them. But then I changed it to the normal, my normal one, which is a dual band, the the five and the 2.4. 01:07.96 Frank Okay. 01:08.95 James And it's like completely fine. So I don't know but anyways, I got this, I got this, look at this cute little containers and a little container and I'm going to open it up. 01:11.72 Frank Hmm. 01:16.57 Frank Uh-huh. 01:18.45 James Oh, see, that's like, it's a, it's eight. 01:19.22 Frank Oh, it's a lot of batteries. Is it charging all 8? 8 at the same time? 01:25.14 James all eight at the same time. and It is USB-C, it's a USB-C case, but let me plug it in because it's GD. 01:27.10 Frank Nice. 01:30.62 Frank Okay. What an anachronism. Modern tech meets 1970s tech. 01:37.78 James It is 01:38.68 Frank 60s? When did AA batteries come out? Please write in. 01:41.46 James so adorable. you see that? do dode do do dode 01:44.54 Frank Oh, I love it. I love it. 01:46.14 James So it does like a little grid. 01:46.40 Frank Part of the reason i hate rechargeable batteries is all the chargers are terrible. Yeah. um If you're just listening to the podcast, he has a cute little black case and it's showing actual little color LEDs for each of the batteries. 01:58.94 James And if, and ah and if you sit in front, like some of them will turn green and it's like, do, do, do, do when it's green, it should be green. 01:59.58 Frank ah 02:04.95 James Um, it was like, don't know, it was like $15 and has like the whole kit caboodle, the whole thing in the case and does the thing, it's kind magnetic and stuff like that. 02:07.70 Frank Cute. 02:15.48 James So I haven't had to use it yet because the new one's been okay. So we'll see how long they last. But anyways. 02:19.58 Frank Oh, that's good. But um yeah, that's that's the nicest charger I've seen so far. Usually they're giant, only charge four batteries at the same time and take up all 10 sockets because it plugs into the wall and doesn't even have an extension. 02:32.51 Frank i'm trying to USB-C everything. 02:33.06 James Yeah. 02:34.31 Frank you know like i'm I'm finally living in the 2010s and updating everything to USB-C. Except i was just recently traveling And I was great because I only packed USB-C cables. 02:46.35 Frank And then every hotel and everything I went to, of course, is still USB-A. The ferries are USB-A. Everything's USB-A. So I have to remember to still bring one USB-A to USB-C cable, at least. 03:01.11 James Well, I'm like the opposite. I'm still living in the USB-A and lightning cable world. 03:04.79 Frank Okay. 03:06.32 James Heather's in the USB-C world. 03:06.65 Frank but 03:07.63 James It's like a very, we have all the cables. And, ah but Alaska Airlines, which I frequent LeFi, they recently upgraded all their planes to be USB-C. 03:10.08 Frank Mm-hmm. 03:18.51 James And like every time get on the plane, I'm like, I don't know, I can't plug anything in. 03:19.03 Frank Ah, nice. Ah. 03:22.71 James But Heather's over there like, oh, no problem for me. What's wrong with you? I'm like, need my adapter. you know. 03:28.18 Frank You know, it was so funny, too, because the last hotel I went to, two USB-A ah plugs, two USB-C plugs. i'm like, fine, we're a hotel living in the 2010s. I can use this. Plug into the USB-Cs. Nothing happens. 03:41.06 Frank Had to plug into the USB-As. 03:41.18 James Oh, no. ah 03:42.75 Frank Yeah. 03:43.63 James That's funny. ah Well, let's talk a little bit about travel. We got some lightning topic as things that we have. Frank sent me a list of goodies. But ah you and I both did some travel. you did some i did some plane travel and boat travel, and you did some motorcycle travel. 03:58.05 James And coding on the road, i actually just submitted a talk to NDC Oslo, which was... 04:06.81 Frank Nice. 04:08.31 James writing more code than I ever have without a computer. And 04:13.26 Frank Ooh. 04:15.03 James um 04:15.13 Frank Define computer. Yeah. 04:17.49 James that's also correct. 04:17.74 Frank You're doing it all with a phone tree. 04:20.02 James Yeah, I did it all on the GitHub, the toll phone. 04:20.38 Frank A telephone. phone with like Put a quarter in and start... 04:24.09 James um So the GitHub mobile app has recently updated a lot to do a lot of this new slash remote with the CLI, but also updated how ah you can work with agents. 04:34.73 James Previously, it was like do a thing, create a pull request. Now it's the full agent session, just like if you're in VS Code or you're in remote or something like that, or in the CLI or in anything like that. 04:39.87 Frank Okay. Mm-hmm. Mm-hmm. 04:45.53 James um You can open a repo, tell it to do a thing, and then you create the PR. So you can iterate just like you would inside of VS Code in this virtual environment. So I've been using it to ask like in-depth questions and things like this on it, analysis, like do refinement of in-app purchases and things like that. 05:01.62 James And you have access to all the models and branches and all these things, which is very nice. So I was pretty spoiled. I was on some Alaska and Hawaiian planes that had Starlink. 05:12.19 James i don't know if you've heard of Starlink. 05:13.08 Frank oh 05:14.07 James In fact, you're on Starlink, Frank. 05:16.79 Frank yeah you can tell by my blurriness my glitchiness and my low fidelityness no i hope you can't uh yeah i've been rocking the starlink for a while so i'm i'm curious how a plane traveling i don't even know how fast planes go let's go with nine billion miles an hour um how how Starlink works on there. 05:37.98 Frank I was tempted to put it on my motorcycle because they got the Starlink Mini now. i'm like, could I just put it, just glue it down? 05:41.59 James mini 05:45.37 Frank Could I do that? that an RV? 05:46.78 James Well, we realize that also the planes are much closer to the satellites. 05:47.40 Frank Am I an RV then? 05:53.90 Frank much. um much um What are they? 05:57.04 James About 30,000 feet. 05:57.84 Frank miles up, I think. So... 06:01.12 James So test. I got 100 megabytes down. 06:04.79 Frank Okay, ah sure. 06:06.10 James um i got one hundred megabytes down 06:10.70 Frank Jeez. Okay, pretty good. 06:11.51 James and 50 megabytes up. 06:14.71 Frank Pretty good. Pretty good. 06:15.83 James That's... 06:16.30 Frank So you're the reason no one else on the airplane could use it. You're just hogging all the bandwidth doing bandwidth tests. 06:21.91 James Everyone else is streaming Netflix and crap. I don't know. 06:24.41 Frank I know, right? 06:24.59 James ah It was free. It's all free. And I was like, that's crazy. 06:27.51 Frank Oh, cool. 06:27.63 James So I didn't actually do very much with it, but I did do a bunch of coding. 06:28.27 Frank Okay. 06:30.51 James had some pretty long six hour, multi six hour flights. 06:31.61 Frank Mm-hmm. 06:33.82 James um So I exclusively have been working on a bunch of applications, um mobile applications, some Maui apps, some Swift apps, some Mac apps. ah And I was doing tons of dog fooding because I have it set up right now where when I do a pull request and I merge, it will kick off a build, it'll upload it to TestFlight, I'll get a build on my phone. 06:54.33 James So for hours, I was just revving. I must i think I tweeted 10 second clip of the 24 hours, one, two, three, four, five, six, seven, eight, nine, 10, 11, 12, 13, 14, 07:04.92 Frank Oh, cool. 07:09.59 James 15, 16, 17, 18 different sessions, sessions, 07:13.78 Frank Oh, my God. Oh, my God. 07:14.84 James And these are like huge, like I'm saying like thousands upon thousands of lives of code changes that I'm reviewing from my phone. And that's only on one of the six hour flights basically. 07:22.33 Frank Yeah. 07:24.75 James I must have slang so much code ah exclusively from the GitHub mobile app. 07:25.54 Frank Oh, boy. 07:31.03 James It was bananas and it worked really, really well. 07:33.98 Frank Interesting. 07:35.63 James Cause I had internet obviously and and and good internet and streaming in these logs, asking it questions. 07:39.13 Frank here 07:40.81 James And I worked on all sorts of different applications. And the only real problem that I had, of course, is that everything that it's being built on is a Linux machine. So if I'm doing different builds, it's not that great. 07:50.97 Frank Oh, I know. You beat me to it. 07:54.48 James Yeah. 07:54.70 Frank That was going to be my complaint about this whole thing. Because I wasn't doing this from an airport or anything. But um just this morning, I wasn't even traveling. 08:05.11 Frank I just couldn't really drag myself to the computer this morning. Just one of those days, you know. But I had my iPad and I was watching TV. And i'm like, what if I, you know, try to not code while coding in front of the TV with my iPad? 08:17.14 James Yeah. 08:19.19 Frank And James, I kid you not, I had 20 agents going and I got like 15 features into iCircuit and five bug fixes. 08:32.44 Frank I felt like a supreme hacker because like everyone's always showing those shots of like, look, I've got 30 agents going and all that. I'm like, i I can barely follow one at a time. But I did it. 08:43.28 Frank I i went for the deep dive of um and just queuing a bunch. And i did it all from the iPad. um As the author of an iPad IDE app, I believe coding on the iPad is still possible. 08:57.38 Frank But you know, the best way to code on the iPad is to let an AI code on the iPad for you. 09:01.98 James Yeah. 09:03.19 Frank And boy, I felt productive. But um I did run into that same exact frustration. that um it would every so often create compilation errors because it couldn't build Mac and iOS software in its dumb little Linux environment. I swear, it only web developers that work on these agents? like do they Are they aware there are other operating systems than Linux? or Someone should tell them. 09:32.12 James Well, and that's why I hadn't really thought about I'd turn off all my machines at home, but if I was really smart, I would have done a slash remote on my Mac and then just had that one publicly available. 09:39.38 Frank Yeah, good point. 09:42.54 James I didn't think about it, but you would have it done that on like each repo at a time. ah But I was like, that would have been pretty clever to do in general, um but I didn't think about that. 09:47.43 Frank Mm-hmm. 09:53.18 James But the one thing I have been doing, which is, obviously we all have this set up, but I have been doing this, which is, It codes, especially if it's doing.NET code, it can compile, but if it's Swift code, it's going to have problems, ah obviously. 10:10.81 James um But what I've been doing is I'll do the PR and then I obviously run the PR CI build, which is very fast, like on a Mac build host on it. 10:21.13 Frank OK. 10:21.56 James And then if there's a failure, what you can do is just get a link to the build failure, go back to the thing and say, hey, go fix this stuff. yeah And it's pretty, so once you have the build and and like what I've mostly seen is it's struggling and I probably don't have a good instructions file, mostly struggling with Swift 5 versus Swift 6 syntax. 10:30.78 Frank Yeah. 10:42.70 James That's been the big issue. 10:43.16 Frank Swift 6 concurrency. I don't think any of the models, no human understands Swift 6 concurrency and the models certainly don't understand Swift 6 concurrency. 10:45.14 James That's exact, yep. 10:52.52 Frank I can't even say the word. 10:52.86 James Yeah. 10:54.68 Frank um But that's good. 10:54.97 James Yep. 10:56.24 Frank that mean, that's better than nothing that you have at least that loop going. And that's what I have also. So I have some GitHub actions. So I could always kick off the action and then copy and paste. 11:06.62 Frank But those are um compiler errors only. And so it kind of made me realize right away, and I think your loop is a little bit better than mine, because we're really... 11:18.84 Frank the The issue filing or the initial prompting, that's kind of become an important part of the job. Code revealing has become an important part of the job. 11:25.48 James Yep. 11:27.30 Frank And then the last part is testing. I think we're all just becoming testers because like the code is kind of cheap now. And you know AIs being AIs, they're going to cheat left and right and take shortcuts left and right. And so it's your job as a developer now is to like make sure it's not cheating and make sure it's doing a good job and to test the software. 11:48.44 Frank um You can read the code and be like, okay, that looks like it's generally going to work. But in the case of iCircuit, it has to like draw elements. So I need to actually render some elements and I need you to look at them and make sure they look good because it's a user facing application. 11:58.40 James Yeah. 12:02.23 James Yeah. 12:06.25 Frank so you need to run that kind of software. 12:09.53 James I think that I had a pretty good experience with it, but I'm not working on ginormous applications. I think working on iCircuit or something where I want the visual validation would, yeah I think I'll get there. 12:22.60 James I've been doing since I had the time, the loop that I ended up having was like working on a bunch of apps, kind of moving between them and then installing from test flight, validating, going back and forth with it. It was a good loop. 12:32.94 James It was a little bit of a, you gotta wait for test flight to, you know, slow them process or thing. 12:36.15 Frank Slow loop, yeah. So that's where you need multiple agents going, right? That's that's the trick there. 12:40.47 James Yeah. And that's the trick. So that I was working on like three three apps at the same time, bopping between them as I was testing them. And that was relatively beneficial. Obviously, if was working on a Windows app, a web app would be just fine too, but you faster probably um if I have that set up. 12:53.55 Frank Mm-hmm. Oh, yeah. 12:55.65 James So I was lucky because I had set up all that stuff ahead of time. um and But I was really, really productive on this trip without... 12:59.67 Frank Yeah. 13:05.82 James I think that this time... I didn't have any laptops or anything, but I was able to do it in a way in which I felt productive, but I didn't take away from the trip at all, right? 13:16.90 James Because you could easily take away from the trip by like being on your phone and coding. 13:18.55 Frank Yeah. 13:20.38 James But I'm like, I'm on a plane and I'm either watching a movie, sleeping, watching Netflix or coding. 13:21.11 Frank Mm-hmm. 13:25.66 James And then the rest of the trip, when we were out, I wasn't on my phone at all. I said, if I'm not on a boat or on a plane, I'm not coding, right? So that was my MO. 13:32.60 Frank Good. Yeah, yeah. 13:33.94 James Yeah. So worked out ro well. So I was impressed. 13:37.59 Frank Yeah, that's a fair point. um Yeah, you don't want it to turn into the, you know, the joke when you work for yourself is you never stop working. Some people, ah you can look at it both ways. 13:45.30 James yeah 13:46.79 Frank You never work or you never stop working. And both are true at the same time. Somehow you're in a weird superposition of work. And I could see that easily happening on your vacation. So that's actually a really good point you make. 13:57.95 Frank I like making fun of you for doing all your coding at airports, but that's a good point that you're bringing up. 14:03.35 James It works. I don't know. It's it's like ready to to go. 14:05.18 Frank Yeah. 14:07.63 James You know what I mean? 14:07.99 Frank Yeah. 14:08.74 James So I'm glad you got to experiment it as well. I think that this entire, i think we're going to see a big renaissance. It's funny because you know you were you obviously have continuous, right? Which is about coding on your phone or on your tablet. 14:21.58 James And now I really think that it's way more viable than ever to not just write code, but to really write production code and manage things differently. 14:32.31 James um 14:34.17 Frank yeah 14:34.36 James More than ever. You know and i mean? It's really kind of crazy. 14:37.69 Frank Yeah, and I got to refine my own loop because um what I'm taking advantage of right now is the agentic workflows. So I have an agentic workflow specifically that does deep research because you know I love that term. 14:51.51 James Yeah. 14:52.74 Frank And so I i could go into an agent and be like, here's the here's the issue number and have a chat with it and tell it to go look at my code. 14:53.13 James Gotta have it. 15:03.34 Frank And then it has to give me all these permissions and everything. And I have to like type out all my constraints and everything. Or what I've done is I've created agentic workflow that has... 15:14.33 Frank so I have my agents.md that gives the rough idea over the repo. But then the agentic workflow really digs into what I want out of deep research. Like, go look at these files, all this, do all this. 15:22.07 James Mm-hmm. 15:23.78 Frank And so at any point in time, i can point an issue at one of those agentic workflows. And that agentic workflow does its deep research and then puts in tons of information into that issue. 15:37.14 Frank And then I can assign the issue to another copilot agent that actually does the coding. So that's kind of my create the issue is my shot at describing the problem. Then there's a second planning kind of phase where it does its own research. And then the third time around is the actual coding. And I like that. 15:55.31 Frank That's a system that I'm enjoying. 15:55.60 James Yeah. 15:56.79 Frank That's the the Frank loop that I'm enjoying. ah The problem is you can't just add um those agentic workflows to a repo very easily. 16:08.72 Frank It's kind of a beta system right now and you got to like compile it and add like 20 different files to your thing. So if the repo is all set up for it, it's easy for me. If it's not all set up for it, then I'm back into the chat window. 16:22.74 Frank And I don't like chat windows anymore. I think they're kind of silly. I think they're anachronistic, not to overuse that word on this podcast, but... Yeah, I think we're past it. You know, chat windows were sold last year. 16:34.20 Frank um i want I want buttons. I want a button that says, here's an issue. Go do this. Like, here are the stages and go do this in each stage. I'm almost one of those people that's going to switch back to the like Kanban boards. 16:46.97 Frank And whenever whenever there's a transition from this state to this state, run an AI agent with this prompt. Then it gets into that state. Then you drag it into another state, run this agent with this prompt, blah, blah, blah. 16:56.02 James Yeah. 16:59.90 Frank I'm feeling like that's almost a better system or 17:02.71 James yeah we're at mvp Yeah, we were at and MVP Summit. I was in Harold's like workshop and maybe it was Pierce's something. And they're like, if you're not planning or or researching, you're you're not using agents right, right? And ah What's really fascinating, especially sub agents and these research and planning things work, is I've been having the most success, especially on big tasks, like big tasks. 17:25.91 Frank okay 17:26.22 James I'm talking about like, I'm deploying fleets of background, dum i'm dumb I'm autopiloting fleeting. 17:28.18 Frank yeah Yeah. Nice. Autopilot fleets. YOLO. 17:33.91 James when When you go into plan mode and you've created a plan that recommends a fleet, you know you've done something right. 17:39.86 Frank Oh, no. 17:40.79 James ah You know it's about to use a lot of tokens. ah and And we'll talk about that in a second. But you know i was doing some big ones. And yeah, it was like, I take a research, I do a big research, and then I do on that research, create a plan, and then based on the plan, do the implementation. 17:58.71 James kind of like you're saying. 17:59.06 Frank Oh, you're three steps. Okay, I'm two steps. Issue, plan, research. Those are the same step for me. And then code. 18:05.88 James Similar. What I do is I want to distill down the research into actionable steps into the plan. 18:07.45 Frank okay 18:13.42 Frank Yeah. 18:13.55 James Now, maybe that's kind of yours combined for us, but I kind of do it in. 18:15.28 Frank That's what my... 18:18.01 Frank Yeah, that's ah that's what I tell the agentic workflow, basically. 18:20.75 James Hmm. 18:20.89 Frank like Go do your deep research, but come up with a procedural plan um and what do we call it? Ending criteria, like validation criteria. How will you know when this has actually been implemented? 18:32.79 James The thing I like about the research is I store the research. I store the research into, because the research is about the thing I want to do, not the thing that I'm going to implement. So what's fascinating is I can then, after I implement it, I can say, hey, go double check that research and make that you actually did the implementation, not just to the plan, but if we missed anything in the deep research that maybe wasn't part of the plan. 18:43.35 Frank Mm-hmm. 18:56.76 Frank Yeah. 18:57.44 James It's kind of like this this revalidation loop of like, oh. 18:58.58 Frank Yeah. 19:00.68 James And a good example of something that was ah missed recently in one of my applications ah was... 19:13.11 James has something to do with in-app purchases. Oh, it had to do with, I had to i did some deep research in-app purchases, as I always do. It had to do with restore purchases with Storkit 2. In Storkit 2, they actually recommend that you don't call app.sync or whatever. It's like, it just call this other thing first. and But that was in the research. And it was it was interesting. like it It implemented in a way that made sense, but then it didn't, then I checked on the research itself because you have it. So I like to store that into a into a, into a docs, docs folder. 19:44.66 James I'm big, you know, docs folder guy. 19:46.01 Frank Oh, okay. You're one of those people. 19:47.35 James Yeah. 19:48.23 Frank Okay. 19:48.76 James I am. 19:48.91 Frank i hate I hate checking that stuff into the code. For me, that's what issues are for. 19:51.96 James I do. That's true. Oh yeah. I guess I could have put it in an issue. Like check this issue, which has all this research in it. 19:58.65 Frank Yeah. 19:59.00 James And I just, as a, as some, a gist, it just needs to be somewhere on the internet that I can tell it to go, to go look at. 19:59.86 Frank Yeah. 20:04.54 Frank I agree with that, but I think that's where you're, when all you have is a hammer, everything's a nail. um 20:10.55 James Yeah. 20:10.58 Frank and I don't think, that the repo should be the code. 20:11.48 James Yeah. 20:13.50 Frank I just, I feel that way. like Because the the plans you're coming up with are transient. They're they're coming and they're you they're coming and going, you know? 20:22.86 James Ah, ah, ah, ah, okay. So here's the interesting part. Here's the interesting part. I no longer, i used to store plans. I no longer store plans. 20:30.97 Frank um okay 20:32.10 James I store research. 20:32.18 Frank yeah 20:34.71 Frank Fine, yeah, good. Okay, I get you. 20:36.38 James i so i so it's it's it's contextual research. 20:36.75 Frank I get you. Yeah, yeah. 20:40.62 James Now, so but and not that it's it's different. 20:41.04 Frank Right. 20:43.70 James I mean, it's different, but also similar. I agree with you. 20:46.42 Frank Yeah. 20:46.54 James I don't know if it should be in the repo or not. I think that the code should be the code and the docs shouldn't be there. I agree with you on that. um Because you got to keep multiple things up to date. 20:56.57 Frank Mm-hmm. 20:56.82 James Plans get out of date. Things evolve, X, Y, Z. So kind of kind of fascinating. 20:59.41 Frank Yeah, I get you, though. It's not a plan. It's research. Yeah, that makes a lot more sense, but it can still get out of sync with the code. 21:02.66 James It's deep research. Yeah. 21:05.70 Frank Yeah. 21:06.07 James Yeah. So you might want to you might want to dump it eventually, but you might say, okay, I'm going to let this linger for a little bit until i'm until maybe I ship this thing to the app store and maybe i refine it down, which would be interesting. 21:15.72 Frank Yeah. 21:16.19 James But I mean, the interesting part of all of it like how we've worked with agents are very different. It sounds like you're almost on this phase, which is really, like you said, anti not anti-chat window, but a little bit more less chat window, more agents running and doing things. 21:31.59 James I'd say you're almost... leapfrog me in a way in that and that way. 21:36.11 Frank Ooh, yeah. 21:36.15 James Cause I still have an agentic workflow to enough cause they're not easy enough for me to create yet. um 21:41.67 Frank Mm-hmm. 21:42.42 James And I like an easy button as you know, Frank. ah 21:44.70 Frank Yeah. 21:45.75 James So I'm gonna have to, yeah. 21:46.01 Frank They turn into easy buttons though, once you have a few of them. And like, cause you can pass um an issue number to them. You can pass information to them when you invoke them. Just like the way with a GitHub action, you can just set up a set of parameters and it builds a little form and you fill out the form when you do it, you know? 21:55.06 James Okay. 22:02.09 Frank So yeah, you can come up with, they're cool. They're cool. I think the worst part of them is they're just gross to create at the moment. They're difficult and gross to create, but once they're there and you have one tuned up, they're kind of fantastic. 22:09.91 James Yeah. 22:14.46 James Good to go. like that. Um, all right, let's get topic number two. Um, well, let's stick onto the coding stuff then we'll go to Tim Apple. Um, 22:23.48 Frank Oh boy. 22:25.34 James So we did that one, Age of Agents, we're all just testers. 22:27.04 Frank Breaking our heart. 22:28.22 James We talked about that, that's funny. We are just testers. Well, actually, I was, Kent C. Dodds just ah started a new blog and a podcast and he was saying, what was what was his tweet? 22:39.79 James It was, we're all, let me find it. Cause it's a pretty interesting, it's called The Last Software Engineer was the name of his blog. 22:49.21 Frank Okay. 22:50.38 James And he said, we're all becoming, we're all becoming, 22:51.28 Frank That'll be me. 22:56.18 James in heres We're all becoming product engineers. 23:03.08 Frank Yeah, sure. Yeah. To a point, i I disagree a tiny bit because at least today, I think the AI still needs some handholding. 23:14.74 Frank I don't think that they're fully engineers yet. 23:15.06 James Yeah. 23:17.08 Frank I think in in in the engineering world, we distinguish between technicians and engineers. 23:17.40 James He said, 23:21.92 Frank These AIs are technicians. 23:24.58 James Here's what he said. I'm gonna read this thing and we'll see if we agree or disagree. 23:26.71 Frank Hmm. 23:27.02 James go Shout out to Ken C. Dodds. Epicproduct.engineer is his website. ah Not the whole blog, I'm not gonna read the whole blog, but I'll read this part, which is... um called the arrows are changing. For a long time, software engineers like archers were judged largely by their aim. 23:43.99 James We were handed targets, build this feature, fix this bug, migrate the system, improve this metric. The best engineers are the ones who could draw the bow, account for the wind and hit the target reliably like Frank Krueger. That skill still matters, but the arrows are changing. AI agents are turning implementation into something closer to a homing arrow. You still have to point it in the right general right direction and the shot can still go wrong, but the arrow is getting better at finding the way to its target. 24:09.61 James that chant That changes the scarce skill. When arrows are hard to aim, the best archer is the one with the steadiest hand. When arrows can steer themselves, more targets become reachable and the best archer is the one who knows which target is worth hitting. 24:24.75 James Ooh, that's interesting. 24:25.65 Frank Yeah. 24:25.84 James And that is the shift from software engineering as implementation to product engineering as judgment. When the cost of shooting drops, target selection matters more, not less. 24:33.47 Frank Okay. 24:37.26 James The problem is no law is law is no longer whether you can hit the target you were given. The problem is whether the target should exist, whether hitting it could damage something else, whether there is a better target nearby, and whether the apparent bullseye is even measuring the right outcome. 24:46.02 Frank Mm-hmm. 24:55.31 Frank Right. 24:56.06 James Interesting. 24:57.65 Frank Yeah, I agree with him. i and i I don't want to turn this into a debate about terminology and definitions, but I'm quickly and briefly going to turn and this into debate about technology, definitions and terminology. 25:05.11 James Yeah, yeah, yeah. 25:10.85 Frank um I think what he's describing is the difference between a programmer and a software engineer. I take the word engineer very seriously because an engineer to me is someone who thinks about the customer and cost and whether something should be built or not. 25:18.46 James Mm-hmm. 25:27.06 Frank I know like PMs and all that have tried to usurp that position. But in the olden days, the engineers designed the bridge and then the technicians put it together. 25:38.00 Frank They did the hard labor of actually building the thing. And that was always the case. The engineers would design the airplane and the technicians would build the airplane. So to me, that has always been the job of the engineer is thinking about what should be built, taking into account all the different cost factors, taking into account all the product stuff. 25:54.96 James Mm-hmm. 25:59.59 Frank So he wants to move the goalpost a little bit because we've diluted the term software engineer to include programming, just basic programming skills. And that's because colleges, for whatever reason, started teaching programming and calling it software engineering. 26:13.94 Frank But no, to me, an engineer has always had those. That was the job of the engineer. It's just software engineers that have had to do both jobs. we We had to be the technicians and we had to be the engineers. 26:26.65 Frank um So anyway, it's the worst kind of debate debating the definition of words. And those are the most useless kind of debates. So I agree with him that um our jobs are changing, or at least we're putting more emphasis on the higher level stuff. Absolutely. 100% agreed. But in my mind, we're all just becoming real engineers instead of programmers. 26:49.05 James And I think one could be said, we're in some discussion on Twitter and someone was like, oh, asking Fowler, like, how many lines of code do you write now by hand? And Fowler's like, zero. He's like, I don't write code anymore. 27:01.62 Frank Okay, but that's absurd. 27:01.78 James And I was like, And I was like, I don't, I mean, I i don't really write lines code. 27:03.70 Frank Okay. 27:07.58 James I mean, I changed one line. I moved a Boolean down. 27:09.93 Frank Mm-hmm. 27:10.18 James I don't really write very much code by hand anymore. I can, obviously, if I want to. um But I don't as much anymore, even for small changes. And maybe that'll change. 27:18.51 Frank Yeah. 27:18.94 James We'll see. But um I think that regardless if I'm writing code or if I'm prompting for a code, 27:22.09 Frank Uh. 27:26.84 James I believe that a lot of those things that Kent were talking about were about are the same. I need to decide what I should be using my keystrokes on, whether it is for a prompt or whether it is for the modifications to the code, right? 27:35.05 Frank Mm-hmm. Yeah. 27:38.62 James So it is just how I am doing the job. 27:39.03 Frank Right. Exactly. 27:40.74 James It is the tools in which I'm executing that thing in for. 27:41.88 Frank Yeah. 27:44.30 James And I was talking to someone today, I said, you know, I've been doing.NET development for two decades, right? Plus. 27:51.16 Frank Mm-hmm. 27:52.02 James And... The reason i can be really efficient, not actually only in C sharp, but in Swift, actually, it's pretty easy language and in Kotlin, that you know, these, these languages are not that different. 28:06.61 Frank all that yeah raw object-oriented languages. 28:08.42 James they're they're especially the latest Swifts, you know what I mean? They're like, they're different, but like I can read it. Like it was a lot easier. It's a lot easier now. 28:15.98 Frank Yeah. 28:16.09 James Exactly. Any object or I can, I can traverse easily, but the architecture, of the things like I'm kind of understanding and I know these different things, especially in C sharp, obviously. But I can be even more proficient because like i know i know it. 28:27.13 James like I know it. right I know what it's doing. i can review I can review it. I can test it. I can give it iterative feedback. And that's a skill that I've grown over time as I've i've been doing it. So it's like this new tool, just like when I got full line completion or IntelliCode or whatever, it's like just another tool in the toolbox in which I'm using for it. 28:46.01 Frank Yeah, and I want to be clear. 28:46.20 James But yeah. 28:47.13 Frank my My pushback with the you don't write any code thing isn't like machismo thing. Mine is purely like, sometimes it's a lot faster to go in there and fix the stupid bug than to write a thousand words and pray that the random roll roll the dice is actually going to fix it the way I want. 28:57.68 James Yeah. 29:02.58 James Yeah. Yeah. 29:05.56 Frank Like, you have to be working on some pretty boilerplate boring stuff if you never have to write code. 29:06.70 James Oh, totally. 29:11.19 Frank If you're doing anything at all novel... ah then it's going to make mistakes. And sometimes it's way faster to just go in there and fix the stupid mistake rather than write 10 paragraphs on how to fix the mistake. 29:17.99 James Yeah. 29:24.17 James No, that's true. 29:24.34 Frank At least for me, I'm not that fast of a typist. 29:26.90 James Yeah. Well, let's talk about, we don't, we won't go too deep because it's it's still just fresh and I don't know all the details. And obviously i work at Microsoft and close with the folks at GitHub, but, know, we have some changing of tides happening. 29:40.48 James We're seeing in the last ah few weeks, basically some big announcements around GitHub co-pilot and, mostly some changes to the individual plans for a little bit for signup stuff. 29:50.75 James But then um we're going to be finally, which this is, i don't think it's really controversial, but you know, there definitely is change, changes change. 29:51.56 Frank Ah, pricing. 30:00.59 James And It's funny because there have been people on the internet that have screamed for this change at at us and me even. Like, why won't you do it this way? 30:10.86 James I don't like how you bill and do things today. So there's other way. 30:13.93 Frank Mm-hmm. 30:14.74 James But basically we're moving away from request-based billing ah to usage-based billing. which is all around toke token usage and things like this, obviously inputs, outputs, and API calls and all these other things based on the models. 30:22.68 Frank Yeah. Yeah. 30:27.74 James And that's gonna be changing in a month, basically. the the The team just announced it to moving to usage-based billing with AI credits, and AI credits are used towards that, and different models have different costs associated with them. 30:31.29 Frank yeah 30:41.86 James And the pricing is the same, the $10 for the Pro or the $39 for ProPlus, and same for Business Enterprise, you get credits. You can use them throughout the month and then you have different models that you can use um obviously and you can bring your own key obviously and do all these other things. 30:59.36 James You get these AI credits and they're used across different mechanisms, I think in areas in which Copilot are used. 31:00.59 Frank it is. 31:07.08 James so I don't know if that means like the the coding agent, I'm assuming it's the coding agent online. I'm not positive, but I know that the code review agent uses 31:14.58 Frank um believe me this 31:16.09 James the coding The code review agent, I believe, uses action minutes, so it's different. Yep, 31:22.36 Frank Yeah, it's it's a little confusing because even the um agent agents, like if you assign an issue to Copilot, I believe we'll use action minutes and AI token stuff. 31:32.34 James o 31:32.86 Frank So it's a and case for a bunch of stuff. 31:33.14 James yeahp yep, yep. 31:35.80 Frank Yeah, I mean, this is, i'm I'm a person that says it's sad. I'm sad. Because um we were basically being subsidized with the request model. And the request model was wonderful, not because it made any sense. It made absolutely no sense, the request model. 31:52.34 Frank But it was, quite frankly, cheaper. And anytime something becomes more expensive, it's just sad. And we all get why. we We knew we were using a lot more tokens than we were using requests. And so it's mostly just sad. Sad that we're going be the free lunch is over. the The subsidies are over. Satya wants his money. So we got to give Satya the money. 32:18.04 Frank um 32:19.03 James Well, I think that, mark yeah, I mean, Mario outlined it to to your extent, which is not just they want money, but it has to be sustainable. 32:19.29 Frank It makes sense. Yeah. 32:27.18 James Like, you know any company, Anthropic, OpenAI, they're businesses, right? 32:30.87 Frank oh I'll push back on that because we're still not paying the actual price for these things. 32:32.52 James Like, 32:37.02 James that's true. 32:37.14 Frank even at at these yeah Even at the ridiculous prices, the prices are about to go up a lot, but they're still not covering the cost of these machines. 32:37.37 James They're still being subsidized 32:44.41 Frank Let me just tell you, everyone, go go buy some GPUs and you'll find out very quickly how expensive these stupid things Yeah. 32:45.34 James Yeah. 32:49.98 James Yeah. And I think it's just how these models and how we use them and the tools have evolved so much. Like, you know, Mario writes, it's like, you Copilot's not the same product it was a year ago or even three months ago or even last month, right? 33:01.58 James He says it's evolved as an AI editor assistant into an agentic platform, multi-step coding sessions, bigger, huger models, iterative across entire repositories, agentic coding, It's becoming the default to bring significant higher compute and inference demands. 33:01.59 Frank Yeah. 33:10.55 Frank Yeah. 33:15.64 James The biggest thing is like a quick chat message and a multi-hour autonomous coding session can cost the users the same amount today, which is obviously in the not the same cost, ah which is nice if, but what at least what it leads to is gaming of the system, right? 33:24.51 Frank Which is nice. Obviously. Yeah, for us. 33:31.16 Frank Yeah, yeah. 33:31.18 James And when, when what is there? 33:31.55 Frank Ah. 33:33.02 James So, um you know, you see these big RALF loops and all this other stuff and these things. And and the the usage-based billing is not unique, right? Prodbench, everyone else in the industry uses this. And funnily enough, the main reason that people didn't like request-based billing, that I heard at least, was there wasn't enough transparency. 33:50.15 James What is a request? How requests computed? it wasn't transparent enough. um 33:55.93 Frank No, that the transparency was this is cheap. 33:56.76 James But, you know. 33:58.53 Frank That's how you should reply to them. Don't worry, you're getting a deal. that's what youre That's what your reply should have been to them. My biggest concern actually with the usage based is this is not transparent because a token does not cost what a token costs. 34:08.23 James Hmm. 34:11.52 Frank There are many kinds of tokens now. There are pre-filled tokens, there's input tokens, output tokens. 34:13.81 James Hmm. 34:16.25 James Cash tokens. 34:17.66 Frank cash tokens and none of these agent UIs expose any of that information to you. You know, there's um a lot of these cash tokens expire after five minutes. 34:29.70 Frank I don't know about you, but um I take five minutes between ah thinking all the time and my little agent thinks. 34:30.50 James Hmm. 34:37.49 Frank And so now I'm going to feel rushed to like, oh, man, if i don't if I don't keep chatting with this thing very quickly, my token is going to go stale. 34:41.05 James Hmm. Hmm. 34:45.80 Frank And now I'm going to have to pay the input token price, not the cash token price. So my my concern is the transparency is just not there. um and And in none of these tools, this isn't even a co-pilot critique. 34:57.72 James Thank 34:59.99 Frank ah Codex, Gemini, none of these tools actually tell you what your actual costs are. You'll find out what your actual costs are if you bring your own key, because then you'll get reports from the actual providers telling you exactly what you used. 35:15.48 Frank But now it becomes way more important that the harness, as they say, the thing that's actually um dispatching the tool calls and managing the caches and everything, that harness and its efficiency is going to become very important um because it's it It's going to dictate a lot about price. 35:40.38 Frank And sadly, that's a lot of the black box stuff right now. So i I'm a little bit, there's a little bit of a rug pull here because every new feature you get from ai providers is look at all these fancy new ways you can use more tokens. Yeah. you know, longer thinking times, longer loops and all that stuff. And we just don't have a transparency a good transparency model for knowing how much anything is going to cost. 36:06.23 Frank You can send in a thing. It could cost a dollar or it could cost $30. One request. You're just not going to know. 36:12.82 James it's It's hard, yeah. And like, you know, you can't even predict that because again, that's the whole part of LLMs. there're You can't predict their output. 36:19.26 Frank Yeah. 36:20.62 James Like if you can't predict their thing. 36:22.26 Frank But this... 36:22.64 James The harness can help, obviously. 36:24.79 Frank The harness can help. And I think this is going to put some onerous on the model providers to say, hey, we have we're using 200,000 tokens versus 50,000 tokens for a vocabulary. Because when you design a model, you decide how many tokens it's going to take to represent a piece of code. That's a part of the design. 36:44.95 Frank Then um the quality of the thinking How many times have I seen um Opus just thinking in a loop? Says, I'm going to do it this way. 36:53.44 James Yeah. 36:54.43 Frank Oh, on second thought, I'm going to do it that way. On third thought, I'm going to go back to the first way. Oh, you know, on second thought, I'm going to go back. 36:58.65 James Yeah. 36:59.67 Frank to And that's just eating tokens now. So I never had to worry about that because that was just one request. 37:02.10 James Yeah. 37:04.59 Frank I'm like, you go you go in your little circles, Opus. You go have fun with that. Now I'm just going to be seeing like dollar bills. Like, okay. 37:11.70 James ah 37:11.79 Frank I hope like while it's thinking, it's just increasing a dollar bill counter and, know, It's going to change my habits. 37:17.02 James ah well 37:18.06 Frank I don't know how yet, but it's definitely going to change my habits. 37:20.67 James It's gonna change my habits I think of now, i went back to the early days where we had to we had to nudge the model a lot with like information, like adding files and adding things. 37:31.30 James And now I've kind of given up, I'm like, it'll figure it out, right? But if it has to wrap around your entire code base and do a bunch of stuff, 37:34.14 Frank okay 37:36.79 Frank Exactly. 37:37.94 James So now I'm going to be tagging more. I'm going highlight code. going do a lot more things in there. But what's interesting too is the hard part with this, and this isn't just a copilot problem, this is an industry problem, right? 37:47.77 Frank Yeah, exactly. 37:48.06 James We're talking, we're talking, you know, not just input, output, cash, but also like inside of Anthropic, there's cash input and cash write cost to it as well, which I don't even know what that means. 37:48.53 Frank Yeah. 37:55.77 Frank Yeah. Yeah. 37:57.26 James So, and then all of them are different and a lot of different prices. And we're we're talking like substantial, right? Because 5.4, which is a model I use a lot, is per million tokens input $2.50, output $15 per million tokens. 38:07.89 Frank Mm-hmm. 38:13.17 James There's a lot less output tokens usually than input tokens. um But 5.5 that just released, right? 38:17.58 Frank yeah. 38:19.99 James I know it's a bigger model. It's more powerful, versatile. It's twice the price. in a, in a, and it's $5 per million and same with the cash. 38:24.54 Frank yeah 38:28.54 James The cash goes from 25 cents to 50 cents, which is, which is very, very fascinating. And compared to like five, three codecs, which is one I use a lot is actually a dollar 75 per input and then $14 per output. 38:41.67 James So it's, you know, it's going to be a lot for us to kind of think about. i do think interestingly enough, you know, all of GitHub Copilot, CLI, VS Code, and all this stuff, are in a very fascinating situation because they not only are allowed doing you to use these frontier models, but they're also allowing you to bring your own key and also run models locally on your machine. 39:04.54 James And you and I have talked about it. 39:05.02 Frank Yeah. 39:06.02 James i actually do think that there will become a time in which I will be able to just, I mean, that it's already very easy to set up local models, but I think i think if you're like, hey, James, use, you know i don't know what model's good locally, a Minimax model or some Quen, everyone loves Quen, right? 39:21.81 Frank quaen Yeah, whatever. Yeah. Mm-hmm. 39:25.40 James I think they i think you're gonna see this. I think you're also gonna see, i think that you'll see like a rapid rise in the next several months from everybody. And you're seeing already with like cursor with composer, which is based on Quen. 39:38.58 James um um I think you're going to see all these like really super efficient models that are very good at targeted things that are giving recommendations, right? And you might end up going with planning with a different model because it was easy before in a world where, oh I'll just use Opus or I'll just use, you know, the the latest GPT-5 or whatever because I can, It's the same price or whatever, right? 39:59.70 James it doesn't matter. It doesn't matter what I'll use it for. 40:00.41 Frank Yeah. 40:01.42 James And then I'll use X-High because why not, right? Because, you know, I never did. 40:03.99 Frank Yeah, exactly. 40:04.98 James I just, which i just used... 40:05.91 Frank It's one request. Who cares? 40:07.51 James one request, and whatever. But you're going to think about it more. I don't know if that's good or bad. I don't know if it's good or bad. 40:12.31 Frank Yeah. 40:13.03 James Maybe better for the environment? I don't know. 40:15.75 Frank It's going to be more complicated, but that's it. Complexity always increases. What I think is going to happen, what it would be nice, is um the local models won't have to solve the whole problem, but maybe the local models will become the subagents that are doing the repping around the code and doing the deep research and things like that. 40:19.86 James Yeah. 40:30.46 James yeah 40:34.55 Frank And then maybe you shove all that stuff off to um Opus and such. So that's what I'm saying where I think the harnesses need to improve. I need to be able to say, like, for these tasks, use Quen running on my 3090. 40:43.98 James Yeah. Yeah. 40:49.59 Frank um but use Opus to do the orchestration and things like that. And the moment there's a cache miss, stop using Opus. You know, like I want to be able to set rules up like that. Like only do this if there's going to be a cache hit. If there's not a cache you can't do this. 41:05.27 Frank Fall back to something else. And we might have to fall back to things more like RELF loops, where we have sub-ideal models, and dumber local models, um just running multiple times over and over in a loop so that eventually they can produce the kind of code that maybe... um Gemini or Codex or Opus could have one shot in the past. So I think we're just going to have complexity and I hope the harness people will ah up their game because right now we have the unfortunate thing of a lot of these agent harnesses are written by the agent providers. 41:44.15 Frank And that's a sick, sick cycle for us to be in because the people charging for the models dictate how much we use the models. That's a bad situation. 41:52.23 James Yeah. 41:52.89 Frank So I think this is where the open harnesses like the VS codes and open codes and everyone should just be out there writing their own harnesses. They can use multiple models and ideally some mix of local models. So I think this was coming. We all knew this was coming. Satya wasn't going to subsidize all our code forever. um It's just it's sad that um the future will be more complex than what it's been. 42:20.60 James Yeah, and I read an article on on The Verge to this point. It wasn't about Copilot, but it was it was called Your AI is No Longer Subsidized. And this was like pretty much looking at the entire AI industry and looking at many products across many providers, not just model providers, but other, other industries. And you can think of it as Uber, right? Remember when Uber remember when uber started? 42:39.93 James Super subsidized. 42:40.06 Frank Oh god, so cheap. 42:40.65 James You get an Uber for like, for like free. Now I'm going to get to the airport and a hundred bucks. Right. 42:44.69 Frank Yeah. 42:45.65 James Uh, and you know, um and it's, it's not, you know, it's just like the, the difference is these models not just have gotten so much better, but the price of them have gone up and also, the context windows have gone up and like all this stuff. 43:01.72 Frank Yeah. 43:04.04 James So our expectations have risen for the for these things. 43:06.69 Frank Yeah. 43:06.72 James So it'd be really fascinating to see how it goes. i'm I'm intrigued by all of it. I'm excited to give everything a try. And ah like, you know, in general, it's not new, right? If you move to, you know, and you have Cloud Code or Codex or Cursor, they're all like the same. 43:20.01 James Everything is as this consumption consumption or usage-based model. 43:22.63 Frank Yeah. 43:24.77 James So um if anything, just follow in line and then see how it goes and, and um yeah, go from there. 43:31.46 Frank Yeah, that's it. I just want to, i'm I'm begging the industry and hoping the industry realizes that the, I'm just repeating myself, but that the harnesses have to keep up with these new pricing changes. 43:42.98 Frank Like we can't have these pricing changes and still have these crazy harnesses. 43:43.80 James And they need to op optimize. Yeah. 43:46.26 Frank Yeah, exactly. These current harnesses that are running Opus sub-agents. No, those are gone now. 43:50.96 James Yeah. 43:51.14 Frank We can't do that. 43:51.54 James Yeah. Totally. Totally agree. ah Well, let's go on to another topic. let's let's ah Let's go on to something a little bit in the wheelhouse that we're talking. we've been talking a lot about, obviously, agents and remote agents and pricing changes. 44:08.19 James Let's talk about a change in leadership with... 44:11.83 Frank Oh. 44:12.58 James ah with our good friend Tim Cook, who's no longer cooking at Apple. 44:17.24 Frank So we're replacing Tim Cook with what now? Opus? What are we replacing Tim Cook with? 44:21.85 James yeah 44:24.29 Frank No, this is sad. End of an era. um I don't know. i'm I'm a little bit excited and a little bit afraid, as all good change should provide. Yes, if you don't know, Tim Cook is not quite retiring. 44:37.32 Frank he's still going to be on the board of directors, but he's stepping away from the CEO position and putting a hardware dude in for the CEO. It's a dude. 44:46.03 James John Ternus. 44:46.75 Frank It's always a dude. I've never heard of him. I guess he's been on some of the WWDCs, but I haven't really paid attention to him, but he's been around for 20 years. Um, 44:58.23 Frank I was a little bit nervous when I first heard him because as a hardware person, I've been a little bit frustrated with the iPhone and the iPad because those devices really haven't updated in the last 10 years, 15 years. It's been a while. Like the iPad was last released in 2010. It's been 16 years of basically no change to the iPad. So I was a little bit nervous that we're getting someone conservative, but I was just reading an article and someone said the opposite, that this guy has always pushed for um bigger changes at Apple and it just hasn't happened. Maybe Tim Cook was the... um 45:35.48 Frank the slowing factor there and to the point where like we might even get some new form factors that would be amazing apple if you could you know improve upon the devices you you revolutionized the world once so you're allowed to do it a second time it's okay so um always nervous about things changing but apple's in a good place and i think they could use a little bit of a shake up and so i'm here for a little bit of a shake up how about you 46:02.59 James Well, you know, Tim's 65, right? So it's retirement age. Tim's been there for a while. 46:07.93 Frank Is that right? Going get a social security check? 46:09.23 James Just did 65. i Yeah, he's is' ready to... Yeah, I think that's what he needs. A social security check. I think Tim's just doing just fine. um Yeah, you know, when I when i type in John Ternus, I've definitely seen this gentleman on WWC or I've seen him at different Apple events, hardware events. 46:29.34 James And ah yeah, you know, There's pivotal changes in any company and CEO change is usually relatively pivotal at some regard. i don't know if I don't know. 46:46.05 James I don't know about CEOs in general. I've worked at companies where I didn't even know the CEO and then obviously, you know, i know Satya. 46:51.22 Frank Yeah. 46:53.14 James um But I think that there's really hands-on CEOs and there's other CEOs that are just like CEO, right? And and some super direction-y and some less direction-y. 46:59.59 Frank Yeah. 47:04.51 James And I don't know where Tim sat in that. Maybe somewhere in the middle, maybe more towards... division I don't know, at different times in his tenure, maybe pivoted a little bit, it's hard to say. 47:11.56 Frank No. No. 47:17.03 Frank I everyone's got their strengths. Some CEOs are business people and all their whole job is make the stock price go up and kowtow to the government and and that kind of stuff. 47:21.19 James Yep. 47:27.12 Frank I think Tim Cook was the operations CEO where he made sure that Apple produced good hardware on time, got good deals with all the people doing their manufacturing and all that stuff. 47:31.26 James Hmm. 47:39.08 Frank I think that's what Tim Cook's going to be known for is the operations guy. Um, I'm hoping the new guy is going to be ah a device guy, a little bit of a return to a Steve Jobs where he has some taste and some opinions. It'd be nice to have opinions again. 47:55.96 Frank um See also my rant about iPhones not changing in 18 years. 48:02.07 James Well, so like, yeah, is it time for the iPhone to change? Like when we say iPhone changing, it do we mean do we mean that it is a, ah another okay, so, you know, with with Nintendo, right, which I'm a big fan of, right, they always had a portable console and then they had a home console, right? 48:08.31 Frank yes 48:21.17 Frank Yeah. Mm-hmm. 48:26.23 James And sometimes they would end up having like a squishy middle console that would kind of like fit in there. not really sure. XYZ. Maybe it's like ah an iterated iteration of the existing hardware, like the Game Boy, like Micro or whatever. 48:40.55 James um And you know then they have the 3DS or the DSi or the these different ones in there. ah And you know, you see more recently they've squished it together into the switch, right? 48:52.19 James That's that's a two in one form factor. So the question really is like when it comes to Apple hardware, we obviously have an iPad. We also have a Mac. We also have a MacBook. 49:02.92 James We also have an iPhone, right? 49:03.22 Frank Mm-hmm. Yeah. Mm-hmm. 49:04.42 James We have different accessories, but is it a another another device that you're gonna have with you or is it a, um all in or is it a this or that? So what I mean is right now in the world of most devices, let's say foldable phones, which is what we're talking about, which today, usually all manufacturers are here is a phone and then here is a foldable phone. You pick what you're going for, right? 49:34.81 James And that is an option. Like the question is like, is it a bold decision to be like, this is, 49:46.81 James This is the future of iPhone. Cause, cause here, when I think of foldable phones, everything seems like a, and here's the foldable phone. It's not, here is the future of phones. 49:59.14 James Like I, I do not believe I've not seen a foldable phone. 50:00.49 Frank Mm-hmm. Cool. 50:02.10 James I've seen very, not very many people use foldable phones. I've seen them, but, um and I've had, I've had one at the service duo for, for testing and a cool device. 50:13.73 James Yeah, totally. And, But it seems like we can't get out of this form factor, this brick in our phones. The question is, are we going to have a foldable brick that is like, this is the future type of thing? 50:27.90 Frank Okay, so foldable iPhone, yeah, that's the big rumor, but it's been the big rumor for the last five years. 50:33.08 James Forever. Forever. 50:33.66 Frank So i'm I'm not even saying whether it's going to happen or not. Plus, hardware takes years to develop. So let's say he comes in in September and says, we need a foldable phone. It could take two years, or they could have started working on the foldable phone two years ago, and that would have been under Tim Cook's austic auspices. 50:49.86 Frank So i'm at this whole foldable phone, it's not even 100% what I'm talking about. I'm talking about touchscreen Macs. but Why haven't we gotten those yet? 50:57.90 James Yeah. Yeah. 50:59.70 Frank I'm talking about whatever happened to the iPod touch. You know, why can't I buy an iPhone? That's not a stupid phone, because you know what? All I get is phone calls are spam. I don't need a phone. I want a device that fits in my pocket. 51:13.14 Frank That form factor, a hunk of metal that fits in my pocket, good, fine. We can keep that. We can put different screens on it. I would take an iPhone that has a back screen and a front screen. 51:24.47 Frank Why not? It doesn't have to fold. I'm just saying there's a hunk of metal that's going to go in my pocket. That's probably going to happen because my wrist isn't big enough screens screen real estate wise. 51:33.18 James Yeah. Yeah. 51:35.13 Frank My little AirPods, no visual element. Maybe it's going to some sunglasses with some little projectors them. I don't care. But all I can say is the Mac hasn't changed in 30 years or the the the MacBook, um the laptop. 51:55.00 Frank The iPad hasn't changed since 2010. It's improved, but it hasn't changed. The iPhone hasn't changed since 2008. It's improved, but it hasn't changed. So I'm just looking for something. Yeah. I love my vision, but um it was overly expensive, too heavy, too many negatives, and they haven't improved it. 52:15.90 Frank You know, they they just haven't. 52:16.16 James Yeah. 52:17.75 Frank um So, and please don't write the update. 52:20.27 James It's true. 52:22.17 Frank So it's it's not even, yeah yeah, I hope that's my point. is It's not just that I'm looking forward to a foldable phone, though I'm totally here for it. I want to see something change, something improve, something. Yeah. 52:33.88 James I think that's a great point. iPads have been iPads. They've gotten different sizes, different things, faster Mac, same thing. And like even the Neo, the air, the pro, even the Mac mini, we've had Mac minis. 52:44.71 James We've had them forever. 52:44.93 Frank Yep. 52:46.30 James ah You know, and maybe there's the Mac studio, which is a bigger Mac mini, right? 52:46.41 Frank Yep. 52:49.14 James So it's like, okay, cool. 52:49.62 Frank Yeah. Fat Mac mini. 52:51.11 James Uh, watches. 52:52.35 Frank Mac, not so many. 52:54.17 James We've had watches for forever. 52:54.44 Frank Watches could could have been something, but they didn't put a real operating system on it. You know, renovate watchOS. Put iOS on the watch. And now we got something. 53:06.14 James TV, the Apple TV hasn't changed forever. 53:06.58 Frank You know. 53:09.56 Frank Yeah. Yeah, that's what I'm saying. You know, so um Tim Cook's done a great job from a corporate business perspective, but he has not done a great job from a hardware company perspective. 53:24.30 Frank They've been too conservative. 53:24.91 James Oh, I like, I like cool hardware. 53:25.62 Frank So I'm here for some shake up. 53:28.86 James I'm a big, like, yeah, I like, I do like 53:29.37 Frank Exactly. You love cool. You love cool hardware more than I do. You should be on board. Yeah. 53:34.47 James Dude, I love cool hardware. I love cool hardware. You know, and I think, you know, for the longest of times, I was always like really big into Kickstarter because people were doing crazy stuff. 53:39.38 Frank yeah 53:45.74 James Even when we're making fun of the the little rabbit device and the thing that you wear, there people are doing funky stuff. 53:45.88 Frank yeah yeah yeah 53:52.41 James You know what I mean? I like cool hardware. Hardware is cool. Hardware is fun because I can't make it. You know, like I'm not a hardware guy, but 53:58.60 Frank it's tough for Apple because they have to do things in such quantity. if If the product doesn't make them a billion dollars, it's a failure. So that stinks. I get it. Big companies, it's hard. But if the big companies can't do it and the little companies can't do it, then no one's going to do it. Someone's got to do it. Someone's got to take some chances. 54:16.15 James someone's got to do it ah oh well let us know what you want from apple i think any of those ideas are cool to be honest with you i'm down for a bunch of stuff yeah i think that uh i think there's a bunch of industries and even like non-phone industries i'm really i really want um 54:20.47 Frank so 54:33.11 James like arcade stuff and pinball stuff. Don't want to to evolve. there's a bunch There's a bunch of stuff i love. You know what i really want to evolve? so if there's any people at the Gajamib Bowl, wine fridges. I'll tell you why. 54:42.85 Frank Okay. 54:43.83 James Okay. 54:44.63 Frank Okay. Wow, that came out of nowhere, everyone. 54:46.74 James Okay. 54:47.99 Frank Dang. 54:49.50 James Imagine a wine fridge in which you put your wine in 54:51.82 Frank ahha 54:54.17 James And it self-identifies the wine that you put in. It can understand the ideal temperatures for that wine, the ideal age for that time, give you recommendations for the wine. 55:05.34 James And it could transmit that data either to your phone. 55:05.96 Frank That's impossible. 55:07.50 James it It's all QR codes. um 55:09.62 Frank I know, just kidding. 55:11.29 James or It's all like ah barcodes. ah But it could even like you know have some cool touch screen thing and you could swipe through it and 55:13.40 Frank Yeah. 55:20.73 James Ideally, if when you open it, you say give me give me this Pino, and then you open it, and like, here's your Pino, right? Like, i I think that, and that's cool hardware. 55:27.00 Frank Yeah. 55:30.06 James Like, that that's now that that's, now that's a good use AI. 55:31.48 Frank Yeah. 55:33.22 James um But I think that there's, all sorts of crazy things you do. Like a good example is, Frank, look at this. See how some are green now? This is the battery. 55:43.65 James I'm going back to this bed. 55:44.23 Frank That's some cool hardware you got there. 55:44.63 James Look this. Look at this. This is cool hardware. Look at that. They didn't have to put these lights here. They didn't have to do that. They didn't have, that's warm. They didn't have to do that. These batteries are charging. They didn't have to do that, but they did. 55:58.97 James And that is really fascinating. co Co-share. I'm sure that's great. No, I'm sure it's, 56:04.82 Frank I'm sure you're saying it perfectly. 56:06.62 James sure it's not gonna blow up. um But, you know, a cool hardware. Check out this. You ready? You see this is a There's a USB, gigs. 56:18.62 James you see that's that's jigs. 56:20.64 Frank that's so many gigs 56:22.81 James watch this ready i'm holding up a usb stick 56:27.16 James you see that 56:27.92 Frank oh it got bigger and smaller why does it get bigger and smaller oh so the little okay yeah yeah 56:31.54 James hide hi hides the usb Boom. You see that? That's cool hardware. 56:38.07 Frank since 56:38.84 James They didn't have to do... they So so that's that's the thing. So that's where I think, right? I think cool hardware and in cool design, you you get it and you're like, you know what? 56:47.55 Frank Yeah. Yeah. 56:51.07 James They didn't have to do that. They decided to do that. And that to me is like a magic moment. 56:54.30 Frank Yeah. 56:58.25 James Same in software, by the way. It's like, they didn't have to do that thing, but they decided to do it. 57:00.44 Frank Mm-hmm. 57:02.45 James And that is high quality 57:02.52 Frank Yeah. 57:05.14 James software high quality hardware yeah find out yeah we'll see all right everyone 57:06.68 Frank Good. And maybe we'll get that with the new CEO. Or maybe we won't. Who knows? But at least there's an opportunity now with a change of leadership. Yeah. So bye, Tim Cook. we did We enjoyed your products. But yeah, always let's look forward to the future. 57:24.57 James yes let's do it all right well let's look forward to ending this podcast and getting out of here uh let's do it for this week's merge conflict so until next week i'm james montemagno 57:29.14 Frank Okay. 57:34.36 Frank And I'm Frank Kruger. Thanks for watching and listening. 57:38.33 James Peace.