Tony Fadell: recommendation

Transcript context

…Well, the discipline of product management sits between all of these functional roles, okay, whether that's marketing, sales, distribution, sometimes manufacturing, depends on what the product is, engineering obviously, and customer support. So when you sit between all of these things, maybe those roles shift or change, especially depending on what it is you're building and that kind of thing. But you have to interpret what's going on between all of them and stitch them all together to make this thing sing. And what we're saying is, "Oh, I can just, today in the AI world, I can just make a prompt and all of a sudden it gets spit out." And you don't know what all those little functions are. If you are not aware of each of those functions, even in an AI world of what those things are, they are very clear definitions of certain points of view for the customer and you have to consider them. And to say that they're going to get washed away and an AI is going to come up with it, it reminds me a lot of how software coding is getting done with something like Claude today. I don't know, maybe it was a month ago when the Claude source code leaked, right? Claude source code leaked. And everybody's like, "Oh my God, it leaked." And at the time, and maybe they changed it, maybe they didn't, but at the time Dario was saying, " 90 to 100% of all our codes written by Claude, and we just monitor it and watch it." And we're like, "Oh, well, that's really interesting." And then the code leaks. And then if you looked at the code, anybody who looked at the code who's a real software architect and engineer threw up, they were like, "It made what?" They're like, "This stuff is brittle." Engineers were looking at like, this should be layered in four or five, actually 12 or 15 different sub functions. This is the main loop of Anthropic's Claude. The main loop, not just something off in the... This is the main loop. And people are like, "How can you do this? This looks brittle. It's unreadable." And they're like, "Well, the AI knows it." But when you think about it and you have to maintain it, you can have an agent make code for you and it could work and it could test, but is it secure? Is it maintainable? If there's something going wrong, can you roll things back and understand what's going on? There's so many other aspects to writing code and delivering a product thing that you still need humans in the loop. And so when I think of product design, I think of software code with AI. And when you look at AI, and if it's not architecting things, and it's not segmenting things, and looking at each of those things, like I was saying, like there's software architects, and then there's software optimizers, and then there's just general coders, and then there's security reviews. and looking at each of those things, like I was saying, like there's software architects, and then there's software optimizers, and then there's just general coders, and then there's security reviews. And if you don't have those different mixture of experts around the code structuring it so that subsequent generations can get better and better and it just kind of devolves into this mass of things you don't know, you're getting short-term gain for very, very long-term laws. And that's called software debt, right, technical debt. And everybody hates technical debt. So it might be fixing something, but it most likely is giving more technical debt, especially when you're at the high level. So now how does that map to product management? So if you're a product manager and you type this in and you get some result, but you don't have a really good marketer, a really good marketing communications person, a really good salesperson, a real good channel salesperson, a really good architect, a really good manufacturing manager, all these different things in this thing that you're getting, you're not going to be able, you might be able to make the first one, but when you go to version five, six, how does that work? You're building on a really crusty foundation and people are like, "Well, my AI is going to be smarter." It's not proven to do that. What's proven is if you properly architect it and have Claude Code go into certain sub-segments, or have Claude help you build the architecture, you modify, refine it, lock it in and say, "Just work on these few things and these more limited-scoped things," yeah, you can make that work. And I think that's the way we have to consider how we use these tools in a product management capability as well: to just say it's all abstract away, it's all going to be better. And when you ask specific pointed questions and you want this thing fixed in your marketing thing or whatever, and then they're going to be like, "Well, the AI should just figure that out too." It's like, you have given up so much. To me, sure, you can write code, but there's going to be the difference between... It's like the difference between H&M and a luxury brand. You could go get certain things that look like that and copies that, but it doesn't last more than one washing or one season, and it's this, and you throw it away, and it's cheap and blah, blah, blah. Or you go to the luxury thing and you pay more and it's crafted, it's handcrafted, it's dah, dah, dah, dah, and you know it's going to be around for a while. There is this dichotomy of like fast and throwaway. So like it's called fast fashion. We got fast software. But software, if you're going to build a real company, can't be throwaway. Maybe it can, but I don't think it can be if you're really going to do this because you get just technical debt and you got to start over again. u're going to build a real company, can't be throwaway. Maybe it can, but I don't think it can be if you're really going to do this because you get just technical debt and you got to start over again. So you've got to really understand how you're using these tools, and a lot of these things that these AI coders can do, agent coders can do, can make you incredible prototypes, do more prototypes, do more of those things to help you get that informed gut to say, "We're going this direction," architect that in and then work on the sub-segments below it and all the expert systems, expert things that you need in each of those domains.…

Stored transcript either side of the excerpt. The highlighted words are the published quote; the surrounding text is unedited source, never generated.