← Return to Index Archived October 26, 2026
The Lead — Oct 26
HUMAN X INTELLIGENT · MADALENA COSTA

Will AI replace product managers? Why shipping 10x faster doesn't mean 10x more winners | Rich Mironov

As AI accelerates software development, product leader Rich Mironov argues that faster code will not create more buyers or bigger markets. The real work shifts toward customer discovery, commercial judgment and deciding which problems are worth solving.

33m / October 26, 2026 /aiproductbusiness / Transcript sourced from openai
All episodes from Human x Intelligent →·Podcast website →·Listen on Apple Podcasts →

The Story

Madalena Costa opens with a figure from Rich Mironov's writing. Most new products already fail. If AI lets a team ship ten times as many, Mironov calculates the failure rate climbs to around 96 percent instead of producing ten times the winners. His reasoning is plain economics. A market for salon or gym software doesn't grow because code ships faster. If four of ten products used to find paying customers, now a hundred products are fighting over those same four wins, and prices drop. He separates generating code faster from generating money faster, and says almost nobody wants to hear it.

Mironov admits he isn't building products these days and hears about the work secondhand. Coding is getting faster, by amounts that depend on context. A dating app can tolerate sloppiness a medical device can't. But the bottleneck has moved. He says a designer can produce 40 prototypes in an hour, but no user will review 40 prototypes, and nobody wants ten new features a week. He'd accept a 10x gain on fixing known bugs. He won't accept it for new products.

Designers lose the pixel work but keep their taste. Mironov separates "plausible" output, which looks believable to someone who doesn't know better, from good design that actually meets a need. He compares the moment to the arrival of compilers, when engineers moved on to bigger problems. If every company asks the same large language models (LLMs) where the next market is, he argues, they'll all get the same answer and crowd into the same failure. He is most protective of live user interviews. With a product manager, designer, and engineer on one call, someone can chase an odd remark like "we never run those reports on Thursdays." A scripted bot just moves to the next question.

He describes the product manager's job as a barbell. The middle, meaning roadmaps, specs, and sprint management, shrinks. The ends grow. At the front is whether something should be built at all. At the back is the pricing, channels, and messaging that turn code into revenue. He won't call them rules, but he suggests matching discovery effort to risk, having a human try every feature before it ships, and keeping humans accountable for results.

For 2031, he points to the 2008 rush to build mobile apps and expects AI to become as unremarkable as running on a phone. His homework for listeners is to explain in under 25 words why someone would pay for your product, and to take someone from finance to lunch. He closes with his wife handing out 38 pairs of eclipse glasses to strangers by the Atlantic.

Main Themes

The central argument is that speeding up one part of a system exposes the slow parts everywhere else. Users, buyers, and budgets don't grow with token output, so the scarce work becomes deciding what deserves to exist and how it will make money. Building things nobody buys, which Mironov calls "product waste," becomes the main risk.

Beneath that is a case for human judgment as something earned through past failures. Models produce the most likely answer from existing data. New markets come from noticing the odd detail that doesn't fit. His closing advice comes back to the same habit of listening with humility, whether to a user, a finance colleague, or a stranger on a beach.

Do I give up my judgment about what’s interesting? I do not, because if I’m not walking down the street in the mind of my user, we’re generating slop. — From the episode

Full Transcript

Source: openai 33m runtime

Welcome to HumanX Intelligent. Welcome to HumanX Intelligent, a podcast about what it means to be human in the age of AI. From all of us living through it, I am Madalena Costa, and every Monday I bring you a conversation about how AI is changing our lives, our work, and the way we relate to each other. AI is changing how quickly we can build software, but building something faster does not mean that is answering the question of whether anyone needs it. So today we are exploring what happens to product work when writing code becomes easier and why deciding what to build still matters. My guest is Rich Mironov, a product management leader, advisor, and author. I will make sure to share all the links below, but in this conversation we talk about what makes a product worth building, the differences between faster engineering and business results, and the human judgment behind product decisions. So let's get into it. Hello, Rich. Welcome to the HumanX Intelligent Show. Thank you so much for being here. I want to start with somewhere specific, actually, because you've spent this year putting your thinking out in public. I met you during Productized, not only that, but also on your own blog. So let's start with a number. You published yourself. Most new products already fail, something like six out of ten. And if AI lets a team build ten times more products in the same time, the naive math says ten times more winners. But you did the actual math and got a 96% failure rate instead. Walk me through how building faster makes things worse. Sure. And it doesn't make things worse from an engineering point of view. It just makes things worse from an economic point of view, right? So let's imagine we have an existing market, right? We're selling software to manage beauty salons or personal gyms or health spas, right? That market doesn't get any bigger because we can ship faster, right? In fact, if we and all our competitors launch ten times as many products into an existing market, there can't be ten times as much money in there. In there to buy those 10 times as many products, right? In fact, with more products, prices are going to come down, right? So instead of asking how many do we win, I ask how many do we lose. So if today we're finding that four of our 10 products succeed, and by succeed, I don't mean that we ship them. I mean that we found a market, made money, found customers, and generated revenue, right? So if four of our 10 products found market fit and found revenue and found success, and we ship 10 times as many products, we're shipping 100, we haven't created any conditions that make more than four of those successful, right? We still have the same health spas or personal gyms, whatever, looking for software to manage their day, but we have 10 times as many products fighting for those same four wins, right? So we moved from four out of 10 wins to four out of 100 wins. And of course, everybody's going to say, Well, I'll be there first, and I'll take share away from everybody else. Pretty short-term play, right? There won't be more winners in that market unless we can create entirely new markets, which, by the way, is hard because at least I've been building software for 40 years, and we're mostly going into the same markets and improving the things we're selling and displacing existing competitors, right? The world software budget doesn't get 10 times as big because we can ship 10 times as fast, right? The distinction I try to make here that nobody wants to listen to is the difference between can we generate code faster and can we generate money faster? Because products are more than just their code. They're the pricing and the positioning and the channels and the messaging and getting folks to try them, right? There's a lot more in the box than just can we generate more code. Yeah, and— I guess something that came to mind, that's why I also asked about the number, is that I'm sure every time you're in the room and say something in the round of that number, every time the room just stays there. It gets very quiet, or people start to yell, right? Right. And for me, it's, again, it's the distinction between being a software engineer and getting really great tools that let me build, well, maybe generate code ten times as fast. I would say we're delivering workable, you know, useful products twice as fast or three times as fast. The whole rest of the company isn't moving faster, and even more than that, our user base isn't growing that fast. So somehow we have to think not just about code generation. We've got to think about the exchange of money on the other end. If we focus, or let me come down out of the theory for a second, because it is easy to nod along the bottlenecks or the bottleneck moves, right, as an abstract idea and lose the thread of what that actually looks like inside a team, a real team. So my question to you—yeah, let's paint the current picture of what is happening. What are the products and engineering teams actually doing different right now, this year even, because of artificial intelligence? So I have to tell you, I'm not building products lately, right? So I am not on the front lines of people who are living through this and figuring it out. I'm getting it secondhand, right? Again, I'm hearing from people that they are dramatically accelerating the step in the process where we create code. And people argue, and there are tools, and is it three times faster or 30% faster or 10 times faster? I don't know, and that varies an awful lot with context and situation. And do you have existing customers? And will people die if your application goes offline, right? So, you know, if you're building an online dating app, you probably are a lot less concerned with some of the quality and testability than if you're building software for spaceships and you can't bring them back, right? Or medical devices where if your CPAP machine stops working, people die, right? So there's very different contexts here. And again, almost all the write-up is about the middle of the engineering process where we've already decided what to build, and we haven't yet shipped it to people who need it to work, right? If we expand that to what I think of as the maker team, right, design, product management, support, the other parts of the technical organization, you know, different people are getting different amounts of acceleration here. If I think about designers for a minute, I believe that good designers can generate 40 prototypes in an hour. But I can't figure out who's going to look at 40 prototypes in an hour, right? There aren't more users who are willing to spend more time doing this, right? I believe that product plus engineering plus design could add 10 small features a week to an existing product, but I don't know any users who are willing to try 10 new features a week because they've got a job to do, right? So, you know, the bottlenecks move from the raw throughput of raw engineering to everything downstream, and maybe everything upstream. Upstream for me, the most important one is we ask, is this a good thing to build? Is it good for the world? Is it good for our users? Is it good for our company economics? Can we sell it, right? Many years I've talked about product waste. Right. Engineering waste is when we don't do a good job building it. Product waste is when we build something the world isn't going to use or buy. And if we're looking downstream to the bottlenecks of users and buyers and marketing and sales, we moved over to that side of the problem, right? I think product managers are starting to spend a lot more time on the front end. What should we build? Should we build it? And on the back end, how are we going to sell it, package it, find partners, turn it into money, position it? Yeah. Because engineering is going to do a better job of building it faster. Let me take another slice of this. If the part of the problem we're worried about is, can we fix obvious bugs in the software faster and ship patches to known problems faster? I'll sign up for 10x there, okay? We look through the tickets, we find all the ones that are obvious to fix, we have our engineers turn them over to their agents, and we don't worry about shipping fixes to bugs because we know they're bugs, right? Not so much with new features, and not at all with new products, right? So I'm trying to pull the product folks out of the engineering whirlwind here so we can ask the product questions. Even with what you just said about the clients versus speed and how we can actually do the opposite of what they were looking for. Yes. How right now the goal is all acceleration, right, for most of the teams, specifically to startups. I was discussing the other day with a CTO, and he was very worried about the design part and what was my opinion on how designers would enter this area. My question to you is, based on your experience throughout all these years, how would you say that each different individual in a product team can continue showcasing their value? Sure. And I love that you use the word value. Really important. So, because it's not about throughput anymore, right? So let's think about our designers. Let's think about a senior designer who's been down the road and gotten their nose bloodied a few times and shipped them things that really didn't work, and so has taste and judgment and insight and continues to do a lot of real end-user validation of what's working and not working, right? We're not going to turn that item over to the scripts and the agents, right? I think that designer no longer has to do all these pixel-perfect color selection, layout stuff. That's gone, right? But if you had your agent spin up four or six versions, someone has to look at that and say, I know which ones of these are good, right? And I'm going to use a careful word here: plausible, which means they're believable, especially to somebody who doesn't know any different, versus good, which means it's actually going to meet the need, right? So I think lots of CEOs are busy playing with prototyping tools, and they're creating dozens of prototypes, and they're confusing those with good design because they don't actually know the difference between good design, mediocre design, and terrible design, right? And so I see the designers leveraging up. So they're going to waste a lot less time with drawings and layouts and maybe even workflows, and have the agents figure out where folks are getting stuck in the workflows. But they're going to bring taste and judgment and insight and human factors to the design problem, as they always have, right? On the engineering side, we're going to spend a lot less time generating code. I'm old enough to remember when compilers came along and I stopped writing ones and zeros and I could start writing code, right? It made code writing faster, but we all leveled up, and now everybody could write simple programs, and you didn't differentiate your product by being able to put a SQL statement in there. Because everybody could do that, right? You had to think of something bigger. And so now we get engineers who worry about scalability and security and meeting the need, right? So I think we may have smaller teams, but I think we have more interesting work because we're going to give the bottom 60% off to our agents, but not the top 40%. Where are you at when it comes to judgment between artificial intelligence? Does it have judgment? Does it not have judgment? Should it have judgment versus human judgment when it comes to creating not only products, but everything around what a product is for a company? Again, I think that's a hard question, and there's layers. Okay, right. So if we ask the question of generic technical support documents, okay, first of all, nobody ever read them, right? And second of all, nobody's going to read them in the future. So what we really want is we want our AI tools to work through all the software and figure out how it actually works and write all the steps and interpret all the questions from our users that say, Well, how do I...? Right? Because honestly, it's not very interesting, right? When the question is, What should the application do? not, What does the application do? So I would use, I do use a bunch of my AI tools to figure out what's existing in the world. Okay. And actually, to do a lot of help getting me using software that has terrible documentation, because then I can ask how it's supposed—how it works, right? But let's think of a world where all of the LLMs have all of the existing data about all the existing markets, and we're trying to launch a new product that's going to open up a new market, okay? We're going to ask all the LLMs what the new market should be. They're all going to give us the same prediction as they gave to all of our competitors, and we're all going to rush into the same market, build the same product, and they're all going to fail, right? So I think we have to ask the more interesting questions of ourselves, right? What are our users actually trying to accomplish, not what we told them to? How do they really measure success? How does the buying process really work? Of all the messages that our LLM gave us, which ones are actually going to resonate with real users, not the synthetic users that we used with all of our biases to give us the answers we wanted quickly, so we could move on and skip all the hard steps of knowing what's happening in the world, right? You've written about teams that want to automate the whole pipeline. Yes. This is perfect run-off straight for, for example, a customer ticket to live code, and about teams outsourcing their actually user interviews, like you were saying, to AI bots and quick surveys. So where do you see people quietly handling over their thinking, and they should be doing themselves? In pieces, right? So when I think of the discovery process, right, there's an awful lot of it that is no value add, right? Looking into our user base and figuring out who the specific users are for features or functions or products, sending them requests to book time with us, getting on their calendars, right? A lot of stuff. Now, I should myself write the email that's going to go out to those thousand people, because if it's wrong, nobody's going to respond, right? And I think the moment we have to have humans in the room is the moment when we're actually asking the real questions of the real users. So show me what worked. Tell me what you didn't work. Why are you unhappy, right? And I'm always saying that we should have— A product manager and a designer and an engineer on the call together, right, endlessly, because you don't get to do that interview a second time. And so when the user says something that's really very workflow-oriented, and as a product manager, I missed it, and my designer, if I'm running the interview, my designer sends me the quick text that says, Ask this question, right? We have a moment to diverge from the script, to ask the interesting stuff. Somebody says to me, Well, we never run those reports on Thursdays. Wow, never would have imagined. Okay, tell me more, right? When we script up the interview for our agents or our bots, it asks all the questions we planned. It may or may not get good answers. They may be the completely wrong questions, right? And it doesn't react in the moment to something interesting, right? The reason I want my engineer or one of my engineers on the line is somebody says something that's all about data cleanliness and scalability, and I missed it because I'm a stupid product manager, right? And my engineer, because we've been trained, doesn't blurt in, but writes it on a note and says, No, no, go dig deeper, right? Ask the question, right? Or maybe we switch chairs, right? Because discovery is about learning something we don't know. This would seem obvious, but everybody wants to automate it. And so we're going to ask all the standard questions and get all the standard answers and learn all the standard things, which is nothing, right? Now, do I want my AI tools to run through those transcripts and summarize them? Absolutely. Score them, right? Figure out which ones are interesting. Do I give up my judgment about what's interesting? I do not, right? If I'm not walking down the street in the mind of my user, we're generating slop. Before we go further, if you're enjoying this conversation, subscribe to HumanX Intelligent wherever you're watching or listening so you can join me for a new episode every Monday. And I would love to know if your team could build anything faster, how would you decide what was actually worth building? Tell me in the comments or send me a message directly. I'd love to hear how this shows up in your own life. Now, let's get back to Reach. Something that I found very interesting when using these tools when I go to interview clients is that, or users, is that the way it analyzes, it's very mathematical-based or very constructed way, while I'm thinking during the call of something that, oh, this is interesting. I was not expecting that this person said this in this way, and I write that. But when I see what the AI summarizes, it's never there, or at least not with the same strength that the user gave, at least before the current AI boom. To say to people that don't do any user interviews or that they have customer representatives going to, that have no connection, directly connection to product. To go to these interviews or to these conversations with clients, and they don't believe that anyone from product should go because it's a waste of time and they should just ship it, ship it, ship it. It's a really hard argument to win because if you've not seen the failures, if you haven't missed out on these, you don't believe they exist. It's for anybody who remembers the Dunning-Kruger effect, right? The less I know about your job, the easier your job is, right? Yeah. So we've talked about what goes wrong when people move fast. So the right question is, if we got a hundred or a hundred thousand tickets telling us to make this change, yeah, what was the problem? Not what was the solution. What was the underlying problem that caused all these people to raise this issue? And is the proposal that some of them are making the right solution, right? Judgment, thought, get our engineering and design and product folks together and we say, well, we're getting 50 complaints a day about this. What's really broken? Maybe not what they're seeing. An honest answer says, We understand the problem that you're having. We sympathize. Here are some things you might consider that are not the solution that you gave us, right? Yeah. And I think that's as humble and as polite as you can be. Now, of course, there's other issues here, but I want to zoom out now to the shape of the job itself, of product management itself, because you've actually put the name on how it's changing, and I love reading what you write, just parenthesis. But you've described the shape of a product manager's job changing what you call a barbell, and I really want people to learn more about it. Walk me through that. So when I put two pictures in the post, and it was on a sort of weightlifting theme, right? And so if we think of the product management work, maybe not today, a year ago, three years ago, as a circle, right, and we superimpose it on the picture of like a five-kilogram weight, right, the middle 60% has traditionally been with engineering, right? Roadmaps and specs and standing over them and protecting them and keeping the CEO from wandering through and helping and Scrum or Kanban, right? All the sort of software development cycle stuff, right? Now, do I think that's really much smaller this year? Not so much, but let's pretend. Let's pretend that that gets from 60% of our job to 10%, right? Because engineering is getting a lot more done and we don't have to write so many specs, right? We still have to understand what we're building, right? And so that takes a lot of the middle out of this circle, and what we're left with is the ends. So the picture of the barbell with the weights on the ends, and at the front, which we always called the fuzzy front end, it's the hard questions. Who's going to use this? How is it positioned? Will it make money? What problem are we solving? Where is the competition? Is this a good thing for us and the planet? Does our CEO end up in jail if we ship this? Right? All of the should we, is this a good idea? Do we have validation? Do people seem excited about buying this? Right? Because if we build the wrong thing, it doesn't matter how fast we build it, right? And then the back end, back to our other bottleneck, it turns out that if we just have a bunch of beautiful code in a box, nobody buys it because we need channels and messaging and a sales team and a marketing team and maybe PR and partners. There's all this stuff that turns code into money, and engineers don't see it because they're working on the code, at least the product. And so we as product folks are going to have to spend an awful lot more time. Because remember, everybody's shipping ten times as many products now, or fifty times. So how do we differentiate it, and what's our key selling proposition? And almost all that, by the way, comes from the interviews with the end users, okay? Because they tell us what's important, and they tell us what messages matter, and they tell us how they measure success, right? And if we're not channeling our end customers, then it's positioned wrong, it's priced wrong, it's packaged wrong, it's through the wrong channels, the messaging is terrible, right? And we built something beautiful. Gosh, there's nothing sadder than a beautiful product that doesn't earn its keep and gets shelved because there's no money. We get the water bottles and the little tote bags, right? But it turns out that most of the company doesn't care about those. They care about whether money comes in. Yeah, and for sure. And if the job is generally reshaping, like you're saying, more weight on both ends and less in the middle, right? I like to say that a few years ago, I was so sure that what I was saying was correct, and for the time it was correct, because we cannot think of the knowledge that we don't have right now. We have to think of the knowledge that we have now. That doesn't make it wrong, because it is right for the time that we are in. I would like to ask you, in terms of guardrails, right? Because if a product wanted to move fast with artificial intelligence without losing themselves in the process, what are the actual rules you put in place for this? Rules is a really strong word. I'm not sure I believe in rules, right? In fact, if I go way, way back in time to the early agile days, right, it turned out agile wasn't a method; it was a way of thinking, right? And it was summed up to try something. If it works, do more of it. If it doesn't work, try something else, right? So, you know, the idea that we know what the rules are, I think is reaching it. But some suggestions, right? So one of them is to say— We want to have a sort of proportional investment in discovery, right? There's some little things which we know are right, and we're just going to go ahead and do them and put them in the hopper and ship them and not worry about it because they're obviously right, and we could back them out if we needed to, right? You know, the bigger things we want to be more thoughtful about, right? If there's risk involved, if we can't undo something and it's big, then we want to spend more time upfront thinking, doing discovery, modeling it out, you know, premortems, right? What else? I think that we shouldn't ever decide that a feature works until a human's tried it. This would seem obvious, but everybody's automating their testing, right? And I'm worried that we're going to ship a lot of features that actually don't work, right? So anyway, I think humans need to be in on the important part of discovery, and I think humans need to be responsible for the outcome, right? So if we ship something and it fails, we're not allowed to blame it on our agents, right? And if we ship something and it's wildly successful in the market, we should be able to stand up and say we were the coordinating thinkers there, right? Yes. I'm thrilled by a lot of sort of product ops and DevOps stuff where we're automating the creation of tests, we're automating the mining of data, we're automating the, you know, a lot of this stuff was hard but not high or value, right? And so, you know, I see us stepping up to do the more interesting work. Do I know where this is going? I really don't. I'm pretty skeptical about whether there's any new ideas being found in the averaging and median data that comes back from LLMs because they're trying to find the thing that is most likely. Not the thing that's true. They don't have that capability. No, but there's an awful lot of information out in the world which I have trouble finding, right? I'd much rather have that. That's working much better for me than Google search ever did. You know, I've been enjoying right now the new addition to Google, like the Gemini being inside of the research. That has been helping me a lot to focus. So I agree with you. I think since this tool started to be more mainstream, it started to help to make the process of discovering the competitor matrix, competitor analysis, whatever it is, more quickly and more easily. You bet. But let's zoom out, or zoom all the way out now. You've been doing this for 40 years, which means you already lived through more than one moment where everyone was certain that the job was about to disappear. Dangerous industrial times, whatever. I want you to take me to 2031. No, no, no. I want you to take me back to the future based on what you experienced in the past. Sure, I'll take you back a few. Let's say only five years out, because now they're saying that every month there's something new. But from what you experienced in the past. Let me fast forward again to maybe 2008, okay? Launched the iPhone, launched Android, and for about a year and a half, every board of directors asked every company whether they had a mobile app and whether it was in the App Store. Nobody knew what they were supposed to do, right? And there were all these tools that came out that let you build mobile apps, right? And every product management job that was posted was for a mobile product manager or an app product manager, right? So when I look ahead to 2030, right, 2031, I believe deeply. That we're going to have figured out what the AI tools are good for, what they're not good for, right? Everyone will be using them. I mean, I'm like the last person on the planet who uses a laptop for anything instead of my phone. They will be commoditized, right? All of the AI tools vendors will mostly have been collapsed down to a few survivors, right? The price point will be near zero. Tokens won't matter anymore. We will once again discover the product managers need to figure out not how to generate code faster or how to add AI, but what their users are trying to accomplish with their tools and their applications. I think we're going to have more software engineers than we did before because we're going to open up lots of new cool stuff to do. I hope that's true of designers too. And we're going to get back to the fundamentals of, well, what are we using our AI tools and all of our other tools, and where do they go? Where do they fit? And we won't be talking about it anymore. I don't know anybody who asks me whether my app is going to run on a phone. Yeah. I mean, what are you, stupid, right? That's become an assumed thing for every application. It's got to run on a smartphone, right? Or you didn't do your homework. But it's not a differentiator. It's not value-add. I see product managers back to basics again. I see product managers on finance applications and product managers on dating applications and product managers on ecological applications, all of which use AI in the places that we know it adds value. And we don't talk about it anymore because it's not interesting. Everybody has the doomsday scenario where, you know, we saw last week two of the big companies let their hacking tools out of the boxes and they attacked other companies. So, you know, maybe this is Terminator and, you know, we're just looking at the smoke. But assuming that doesn't happen, I think this is going to look a lot more like the last few. Yeah, but let's focus for now on today. Okay. And to give the people listening one thing that they can actually try this week. Okay. I'm thinking, right? Things that we could try this week. So one, I think I put out a book called Money Stories last year, which is all about can you explain in less than 25 words why someone's going to buy your product, right? And I think every product manager should sit down and, you don't have to get the book, but, you know, you should be able to explain to the revenue side of your company whether your product makes money and how it makes money without ever talking about the tech, right? Because it's going to turn out in six more months when the CEOs are tired of overspending on tokens, this is the question they're going to ask, and we should be prepared, right? The other thing I think people should do next week that isn't anywhere on my list, but why not, right? You should find somebody in another department that you don't know much about, and you should take that person out to lunch, and you should learn what they do and interview them a little bit and find out what they care about and how they're measured and how success is for them, and get a little humility about how everybody else in the company does something that we don't do, that we think is easy and turns out not to be easy, especially in finance, right? Because the money stories are going to require that we know something about the finance of the company, and the finance folks are really smart and don't care about product. And if you have a friend over in the finance department who will help you out on something because you've asked about their kids or their favorite sports team and you've been humble and listened for a half hour the same way we listen to users, you're building a network in your company that's going to be hugely valuable, and you're going to understand whether your next product is going to fail or not. Thank you. Before we close, there's one more thing I ask every guest, and it has nothing to do with product management or product or CPO. It's what does it mean to you right now, in this moment of your life, to be more human? More human. I think of that a lot in terms of doing wonderful, doing things that bring me joy, and then doing things that either bring joy to somebody else or pay it forward, right? I'm at the place in my career, which is at the end, where I've helped pull a lot of people up the ladder behind me, and my plea to them... is to reach down the ladder and pull somebody else up behind them, right? And if you're walking down the street and somebody drops a package, you help pick them up, right? And when somebody's having trouble walking across the street, you help them walk across the street. I think for me, the human is to get out of our own way and to leave the agents behind for a little bit. And there was a solar eclipse here a couple of nights ago, right? And my wife, who's brilliant, needed to buy some of those glasses, you know, and Amazon only had them in packages of 40, and they were cheap. So we ordered a package of 40, and we went down the hill here to the Atlantic, because I live in... And she gave out 38 pairs of glasses to kids and people and whoever who didn't have them, right? And it was just like, you know, what are we doing today to bring a smile? What are we doing today to make somebody else... You know, I'm very privileged to have a lot of things, right? What can we do today with what we have? You know, my kids grown up, so it's not about kids anymore, but, you know, gosh, how do we be good neighbors and good citizens and appreciate art and read a good book and do the thing? It's perfect, and thank you for doing that. That's such a good thing to do. And thank you also for sharing that with me and sitting here today with me. I want to close something with something more practical now. Is there anything you'd like to share with the people listening and where they can find you, your writing, and Miranov Consulting? Right. I'm really easy to find. I put the links in your notes, so you might send them out later. But everything I've written for the last 25 years, because it's a long-running blog, is on my miranov.com site, and it's echoed to my Substack, and it's free. I've got a couple of books out there. The most recent one is called Money Stories. Guess what it's about? They're on Amazon, and they're cheap. And if somebody really can't afford it, drop me a note. I'll send you the bits for free. And, you know, I'm on LinkedIn and WhatsApp and whatever. If somebody needs my help for something, drop me an email, right? I mean... It's not about the money, and it hasn't been. Cool. Thank you. Thank you again so much. It was a pleasure. Thank you so much for spending this time with me on HumanX Intelligent. Today, Rich and I talked about the difference between building software and building a product people need, and the decisions behind deciding what comes next. For me, it's a reminder to ask a better question before reaching for a faster tool: what problem are we solving, and for whom? Speed is useful when we know where we are going. I would love to hear what stayed with you. If someone came to mind while you were listening, share this episode with them. You can find Rich's work in the description and follow HumanX Intelligent on Instagram and LinkedIn. And I'll see you next week for another conversation about what it means to be human in the age of AI. Until then, stay curious and stay human.