LaunchPod - Oji Udezue
===
Jeff: [00:00:00] All right, Ji. What's up man? Good to have you back to the show. How you doing?
Oji: Thanks for having me back. I feel like I'm getting up there in terms of repeats.
Jeff: It's been about a year since you've been on the show, but I saw you last week down in Austin and Dallas, so we're getting a lot of quality time lately.
Oji: Yeah, that's right. I'm not tired of your face yet,
Jeff: Not yet. Hopefully. So for, for folks join us , who weren't here last year, right? You're, you know, just to give quick background, you've been everywhere. You're product Microsoft, you were at Atlassian, Calendly type form.
You're at Twitter for a little bit. Maybe just give the, the folks at HOA, the 92nd TLDR , on who Aji is. We can go from there.
Oji: So, folks at home my name is Aju Dewe. I've been a product manager for 25 years, a product marketer for like two or three years. Dabbled in sales enablement and different parts of technology industry, corporate value creation machine. I'm mostly a product manager though, and for the last seven years I've been head of product and Chief Product Officer Atlassian Calendly type form Twitter.
And lately I've been spending my time in the global south [00:01:00] helping build capacity for building technology companies in Nigeria, in Nairobi, in Kigali in India. And I've also been working on building a venture fund to invest in the great replacement of our field intelligence B2B SaaS. So yeah, that's what I've been up to.
Pretty busy.
Jeff: And on top of all that, you've been flying around the US with me periodically talking to proc leaders , in all sorts of different cities. , So, a year ago we had you on to talk about the book that you and your wife wrote called Building Rocket Ships, about building great high performing product teams. we knew how much AI was going to disrupt this function over the ensuing year, but, you know, who could know exactly what it was gonna do?
That's what we're gonna talk about today, before we jump into it, right, we're gonna talk about the three speed problem. That you've brought about how AI has really caused, you know, speed increases across the board, but has not hit every team the same rate. Engineers, clearly they have cursor or windsurf or whatever they're using. It's changed their velocity probably a lot more than, you know, go to market teams or product teams.
But maybe first, , let's speed run building rocket ship. Let's fly through, [00:02:00] what's the, what's the shipyard model? What's sharp problems? What does that all mean?
Oji: The book is called Building Rocket Ships and it's collective 50 plus years of experience building hard hitting high growth technology companies. Some of them are the ones I've mentioned, but isn there was ahead of. New products at Time Inc.
Head of data and AI at Procore CPO at WP Engine. She's worked in hidden companies that are still high growth and doing amazing things. And so collaborating on that meant that we had a lot of perspective on all those 50 years. Now, you know, she's a product professional, product executive like me.
So we have a really good partnership , on thinking about it. And, and for those who are paying attention the book is actually pretty cool. You can get an Amazon, but there's a digital version with lots of practical content that you can use to make it part of your work life instead of just reading a book.
the main things we sought to do in Book one, was. About the fundamentals of product management, everything from how to pick a sharp problem, how to think about user experience, how to think about things like virality and design. And [00:03:00] all the way to pricing, which is like right there at the top of like specialized product management skills that you have to share with marketing and sometimes with sales.
And then book two is about how to lead teams that are really good at this stuff, right? How to think of product strategy, how to think of product execution, how to think of bringing everyone in the company together,
so we started out talking about the sharp problems. And this is the fact that we noticed that contrary to a lot of tech wisdom, the most important thing to success is picking the right problem because the right problems are forgiving and really this centers around. Optimizing the two forces of value.
You either make something like five to 10 x cheaper, or you make it five to 10 x better. Sometimes we use words like three to five x, but five to 10 X gives you the kind of ambition you should be shooting for. And so if you can find a workflow that is valuable, high frequency for your customers and you can make it five times better, you're gonna create value.
And even when you make mistakes, the market will be forgiving of you if you don't pick a sharp problem. In a high frequency quadrant. Mistakes like running out of cash for a day or a week, whatever. You're [00:04:00] just dead in the water. The other thing we talk about is shipyard teams. You have to think carefully about the mix and autonomy of the teams you used to execute.
So we identify like six skill sets, not people. Not actual individual people. Skill sets are required to do thinking and shipping and go to market. Product management, engineering. Product management is little like, why are we doing this? Is this the right thing to focus on? Product engineering, which is building design, which is part of like, how does this thing look and feel?
So when we get to solutions, user research, which is essentially mind reading data and AI and ml. We've had data for a long time. It's morphing into data engineering and product marketing and storytelling. If you can get these six skill sets, either in one person or two or six people, you can build a thousand person teams who work on interesting problems for your company.
And of course you have to remember that. Orbiting this core team is go to market you know, customer support, the skin, the skin of this big, you know, beefy Android because you have to feel the market and you know, [00:05:00] those sales teams, those. Customer support teams talk to customers so much in a way that you can't, and if you mind their intelligence, you'll become much better at building available company.
So these are the core things. We go into a lot more things, but the shipyard framework is very important that people talk about that. We give people language for it, and the sharp problems for framework is very important because it's a precursor to creating value.
And you know, these are the fundamentals. We've had 40 years of technology, you know, core software technology and maybe like 60 years of silicon, right? From Fairchild, Intel and so on and so forth. And what happens is every 10 to 15 to 20 years, we get to a new technology level.
So at the, the first level was just microchips. Okay. That was Intel. The second level was desktop computing. The next level was internet HT TP. Right. And internet protocols. And the next level has been artificial intelligence. In between, there were some mini boomlets, like social
Jeff: Mobile. Yeah,
Oji: and mobile.
Jeff: But doesn't change the fundamentals here, how
Oji: Yeah. They accelerated some things, but AI is like legit the next level. So it's like a [00:06:00] staircase. Of new technology and every time that happens, we actually, if you pay attention, recreate everything again because. There's some fundamental workflows at work and in and in leisure and leisure.
Things like dating. Or sports and betting, those are the killer apps of leisure, right?
And then the next in, in work it's like, okay, communications is huge, right? That's front desk to CEO and then there's specialized workloads like the CFO and accounting and so on and so forth.
Every time the technology level changes, we recreate the entire software world. We always do. And so. The book Shipyard Team, shop Problems, those, those are fundamentals regardless of technology level, the skills, the listing you do to the book, you can apply. Now in the book, we actually wrote a whole addendum on art field intelligence and how it impacts everything we just wrote.
We spend the last nine months actually deepening that. You know, Izzy and I see ourselves and you know the crew at Product Mind. She, us to the future of AI and software. we wanna point in the right direction of where things are going so that people can waste less [00:07:00] money, be more effective at making their teams better.
And so if you go to the podcast, you go to substack, you'll see us charting the course, and I think it's very valuable to check it out. So, but these fundamentals remain they'll never be. Out of, out of fashion. And so that's why you should ingest that. Now we can talk about the implications and we're about to talk about a three speed problem, but I just wanna make sure people understand the framing.
Jeff: on that front, I guess, you know, if you have, let's take for granted, people are, you know, operating in that framework correctly, and they, you know, they, they've got the great shipyard team. Obviously the big enhancer, like you said in the past couple years has been.
This growth of artificial intelligence. And, you know, AI comes in many flavors for, for anyone who thinks it's brand new, you know, past two or three years. You know, it didn't start with chat GPT I mean, Watson goes back to the teens and, and there's been natural not, sorry, there's been there machine learning since
Oji: Mm-hmm. Mm-hmm. Mm-hmm.
Jeff: But there's more to it than that, right?
You kinda lay out five flavors of ai. We can go into it or not,
Oji: No, no. I do, I I really want to, so,
look, if you were a civilian, keep scrolling, You can call it AI all you want, [00:08:00] right? But if you think about it, artificial intelligence mean it's a meaningless word. It doesn't mean anything. And you know, even the word we call a G, artificial intelligence is also meaningless because you most, 99.9% of you can't tell me what it means and define it.
And actually we can define artificial intelligence. So what I ask builders, software executives, product people to do. Is to use different, more compelling and clear words to think about actual intelligence. Because if you can change your language, you can change how you think about it. And so one of the things that product Mind we've done is to use the framework of new capabilities.
If we say technology level has changed. It makes sense that there are new capabilities that didn't exist before that have come about. But what are they saying? Artificial intelligence doesn't elucidate? What does that are actually clouds it and people can talk for four hours about AI and talk about different things.
So what does ai, what are the new capabilities of artificial intelligence? There are five [00:09:00] things we've identified that are net new or net new in the form based on transformer based large language models or what I call large media models that have come about in the last three, five years. The first one is content generation and transformation.
This is everything from make a marketing newsletter to generate code. To soar, to generate video and images and so on and so forth, you know, to nano banana.
Jeff: Yeah.
Oji: LLMs can understand us even imperfectly, the same way humans can understand each other, imperfectly with bad grammar and slang or whatever. And it can create things that approach very skilled people, right?
That's one. It's new. Isn't that new? We've tried this for many, many years. It's new. And so if you think about it that way, you can build new products on that capability alone. You can measure how good it is at doing that alone. Each of these things has its own metrics, which is really cool because then it's beyond just artificial intelligence.
The next one is natural language interaction. I can't tell you how long I've worked on natural language input and output, Microsoft and Atlassian. Microsoft has a long history of pen [00:10:00] computing, voice computing, acquisitions, dragon naturally speaking, blah, blah, blah, blah, blah. This is the first time it's really good because the human brain is really good that this stuff.
And so everything we did before LLMs sucked, right? Even Google's, you know, transcription was like. 90% for a very long time. You're stuck with machine learning technology, so that's net new whisper, all these things. The next one is information analysis and data synthesis. You can throw a snowflake planet worth of data to an ai.
If you have the right compute. It'll understand it and understand it without you having to clean it up as much as possible. This is also net new. The fourth one is personalization. AI and I did this at Typeform AI can literally personalize and understand 8 billion people and give them exactly what they want.
This is new. It's a little terrifying actually, because if you think you can ad industry and so on, you're salivating, you are, you know, at at Typeform, if I and the other companies, if I got away with sort of mapping people into like a four by four 16. Quadrant and say, you know, people [00:11:00] are in this quadrant, I'm gonna make the software behave like this.
People are in this quadrant. I'm gonna make, if I recognize that I'm gonna make the software behave like this. And this is the kind of thing people like me did, right? Personalization was course. With ai, you can go so granular that no two human beings will see the same response on anything, anytime, anywhere else.
So power also terrifying. Then this last one is autonomy, right? Autonomy is important because this has been a, a holy grail. When we talk about robots, robot dogs, we talk about all this sci-fi stuff. What we are really talking about is autonomy, is you feed a thing and intelligence a lot of data, and they can go off and make decisions and come back with results.
That's also net new. Say, you know, every conversation, ai. And you say these five things, or which one of these five things, and you have this conversation with your team yourself. If you're moderate to yourself or your boss or the board, you, you will have stronger conversations immediately.
That's our gift to people who are listening because I think it'll change your life.
Jeff: Hopefully everyone wrote that all down. 'cause I, I've definitely found [00:12:00] just being able to elucidate with a better language has, has helped make stronger product decisions. So I recommend it. Alright. Now the big thing, this is the, the peak behind the curtain a little bit. 'cause we've been, you know, as Aji and I said we've been around the country talking to product leaders and, talking about this three speed problem. So, you know, AI has done revolutionary, revolutionary things but, you know, the killer app, really from the beginning everyone looked at at least, I dunno, maybe I, I did and we've talked about this code, code makes just sense.
It's deterministic at some level. You can look at it, it's functional, it's easy to understand if you know the right things. It seems like AI was. Built for that. But that's only part of delivering great software products. So hence gives us, you know, the, the thing that you and s Nate have been talking about a lot lately, which is this three speed problem maybe lay it out for us.
You know, it starts, it starts a product, ends with go to market. You got engineers in the middle. What?
Oji: Yeah.
Jeff: ma'am?
Oji: Yeah. So, by the way, there's a whole keynote we've done did that industry, which is somewhere online, so people can, if they want an extended take on this they can get to it. So the AI capabilities to eat my own dog food that we're [00:13:00] really talking about is content generation. Now, content generation is one of those things on the a GI scale where AI has gotten really good, .
Sometimes better than humans. It's still kind of annoying way it composes it and m dashes and stuff like that. But it's really good because it takes like the, the whole knowledge of the internet to generate something for you, which is better than most humans can do.
But this has set up a new conflict. And the core unit of software development. Now, what is the core unit of software development? It is the squad, the team that makes stuff. We've talked a little bit about the shipyard team, which is sort of a new way to think about what that team looks like, the skills it has, how it works together, but the core.
Limiting function of our teams for the longest time, maybe 20 years, 25 years, has been the speed of engineering, right? Everything we've done from, waterfall to Agile has all been to optimize slow engineering. Even the fact that in a company, a technology company most of the head count. [00:14:00] Is developers is because of the rate of development.
So if you're following along, there are three core things that you have to do to make valuable software and then create a valuable business. The first is you have to divine markets and customers find their problems, listen to them, read their minds, synthesize a new solution to the workflow problems they have.
Design it, boom. First part,
Jeff: I I gotta be sorry to jump out. I gotta be honest. Sometimes when you say things, it, it sounds a lot simply like, oh, that's all we gotta do. But that one there is like, oh, that's all. You just have to read their minds and, and, God, it's amazing. We have anyone has ever designed anything decent, like that's a pretty high bar.
I think back to write the five, 10 x better thing. That really is it though. If you want a, a fast growing company that's going to grow and, and really, you know, be able to hyper grow like companies are looking to that is the bar like, it sounds difficult because it is, but
uh,
Oji: it is.
Well, this is, this is a soul of, you know, people keep asking what a, what a product managers do. Well that is what we do, right? We, we read customer minds. We read [00:15:00] Market Minds. We pay attention to where people have pain, and depending on the technology level, we compose a new solution that accelerates them better than it did in the past.
And because we do that correctly, when we do that correctly, people pay us and make our companies rich, right? And then we take all the externalities of software development, which is in, you know, like zero marginal cost. If we get to a million people, we make a lot more money. And that's how technology companies, this is why the richest people in the world, technology companies, they build it on the back of good product managers Now that's the first part and it's very hard and actually 90% of companies miss it.
This is important. Most companies can't actually get around building a great product. Every great product you see there are nine that try the same thing that died. So this is a huge leverage point. Well, let's not belabor it too much in the middle. Once you've figured out a solution, how its form and function should be, you build it to engineer your write code.
Now, in the past, code has been. Code has been static. You know, you build it back and you build on the front end. You write static [00:16:00] code, you make it complicated to handle scale, to handle errors, to handle security, to handle all the ways that users can screw up software. And then you get it right, hopefully.
And then on the backside of that, you make the software get into the hands of customers. So you do all kinds of things. You do storytelling, you do marketing. Sales and so on and so forth. Those are the three things that make a Val technology company. What we are saying is that the middle, which used to be the limiting factor, is over the next 10 years going to accelerate by five and then 10 x maybe more, right the other day.
I think Claude in an experimental session, held enough of its attention to remake Slack in 35 hours in one shot. They rebuilt Slack. So think about what's possible in two, three years, four, five years, right? You can build entire products. , By the way, you can't actually do that easily. You have to put rules and tell it exactly how it all these things work, and it can do it.
If you say go build Slack, it won't do it for you. You have to be very clever about how you do that. And even when it does that, it's still like. For [00:17:00] first draft, you still need real developers, but that's still incredible. So this middle is gonna accelerate very fast. Now the question is, and the why, it's called a three speed problem, is again, conception and solution build, go to market.
Those are the three speeds. This thing is gonna go fast, and what's gonna set up is a conflict if the product management and design doesn't accelerate. It is gonna log jam on the capacity to build. If the go-to market doesn't accelerate, it's gonna log jam on the capacity to build. So this is an inversion of a problem we used to have.
So over the next five to 10 years, maybe shorter, most teams in the economy, surely in advanced economy like the United States, have to figure out how to impedance match these three levers for growth. And what we talk about is how to do that and all the different ways you can do that. We definitely have to speed up our customer market divination.
We definitely have to speed up GTM because this thing is gonna go and we talk a lot about how to do that, how to think about it.
Jeff: it might not. Fully log jam building, right? Like let's say product can't keep up. The, [00:18:00] the worst case scenario is not actually potentially that engineering has nothing to do and they just sit there twiddling their thumbs. It's that they just keep doing anyway and you get all sorts of half baked, features that maybe
Oji: Yeah. But, but doing what? Let me give you a, a work If, if you have a ship with a warp speed drive.
Jeff: right.
Oji: And you're pointing at 360 degrees of space. If you go fast in the wrong direction, you, you, you've just killed
Jeff: No. That that's what I'm, that's what I'm saying is, is the worst case scenario is not that engineering stops because they're log jammed. It's that they're in a rocket ship going warp speed the wrong direction, and
Oji: Yeah. And they will never stop who, who has ever, who has ever made an engineering team stuff.
Jeff: No, I mean, I think, I think what you, you were talking about when we were in, in Austria, Dallas, you were bringing up the team where the CEO started vibe coding something and they released it and they got the CRO on board to start selling it, and the fricking thing just didn't work.
And so they were pushing this thing out into market because that's the risk of going fast without, without
Oji: Yes. No, a hundred percent. A hundred percent. No. No. It, it is, it can, it is actually like harmful. But I will say [00:19:00] this, actually even without speeding up product management, people can start to. Spend this smorgasbord of capacity teams that are definitely looking at me right now with side eyes thinking, what's smorgasbord like?
We have not seen a smorgasbord, but it will happen, trust me. If you have a good CTO, it will happen. , But even before you speed up, I, you should. There are also creative ways, all this tech that we keep yapping about. We can rewrite it faster. Just start afresh, build it on AI first.
Backend that you've been trying to go from. You know, we went from monolith to microservices and back to some micro monolith.
You
Jeff: and
forth. back and
Oji: you,
you can,
you can restart this stuff because you're gonna need a new code base anyway. That's very model aware. To do that. You can spend that capacity on. Refinement, right?
Make your product like a jewel. Every pixel looks beautiful. Every moment is a reward. You can do that. You could spend that capacity on experimentation, so not products, but making multiple iterations of something faster than you could ever in the past. So that now instead of [00:20:00] like making testing with customers, you make 10 times, you test each general customers and the one they like goes forward.
And you don't have any lag in customer adoption. There are actually creative ways to use that capacity today without increasing the speed. But eventually those things get tired. Those things get boring for engineers. And frankly, what we really need to do is make the decision making the direction finding faster, and I think that will make everyone more satisfied.
Jeff: We've talked about this, we've seen teams doing this or starting to you know, one team I talked to has picked up the practice of, whereas they used to take two weeks to do a design sprint and kind of come, you know, over two weeks they'd work with designers and product and, and other things.
And at the end of two weeks, you would have one. You know, flat file Figma prototype. You know, I use air quotes for the prototype because it is just, it is a picture. It was a nice picture. But now they're talking about what they've instead done is they've basically looked at what could we really do here?
And they will take, you know, a couple hours each to go through and kind of use, you know, [00:21:00] one of the many vibe coding tools to prototype quickly and
Oji: Mm-hmm.
Jeff: of an idea and come out with eight takes on it. And in an hour they have eight fairly functional.
Oji: Mm-hmm.
Jeff: Takes and they actually will ship that to, to a small live kind of alpha group who will test it.
They'll get feedback and so they're able to go to engineering with ideas that are, you know, they're not coding the final product. Like you said, you still need engineers here, but their speed of ideation, but also their ability to try. Riskier things and throw away the trash and find the, the nuggets of gold.
They've been able to, the CPO said like 20 x their velocity. I don't, I dunno if I buy
Oji: That
is, my, no, that is my favorite version of solving a three speed problem. I want your listeners to pay attention to the beats of that solution. 'cause it's very clever.
You listen to customers where you don't like. Go to the nth degree, right? So that's a speed up right there because you know, go going to 80, 90% of certainty is faster than going to 95, 90 9% of certainty. And then you take that knowledge and instead of like writing a PRD, you write an app and maybe in [00:22:00] certain cases you write six apps.
Right now the app is the PRD. But now you can do it in code versus write it and you will, don't tell your engineers that you'll ever ship this stuff 'cause they'll get mad at you and then they'll start fighting you and nothing will ever get done. And then watch this. You ship it. To willing Guinea pigs, right? Because they're always the willing Guinea pigs in your audience who are like, gimme, gimme, gimme, gimme, gimme. If you have a good
product,
Jeff: the newest, the fastest, the best,
Oji: right? And so in the first phase, you've now done a ship cycle. You've done a full ship cycle.
It's super fast and because you are working on volume, customers are whittling down. What works for you right now in the old way, you have to do a lot of divination. And pass it through an expensive bill process. And at the end you could still be wrong, but in this one you are working out wrongness very quickly.
And then when customers, by the way, small set of customers will not tell you everything you need to know. Like you could still have a small set of customers like something and the end fail. It happens all the time. At Twitter, we happen all the time.
You know, when you get to a million it's very different than when you got to 10.
But then. Your engineers get a [00:23:00] prototype hopefully in the same stack they use to go forward if you're smart. Right. So they don't have to throw it away completely. And then they put force behind the thing. The engineering team can then spin off a hundred different variations of that one.
Right. But it's now more in line. It's like increasing accuracy. , That's one of my favorite versions of how to. Tackle this three speed problem.
Jeff: Yeah, so there's a lot more to this. Clearly we could keep going on. I think there's the fourth speed problem, which is there's only so many hours in the day and, and you're a busy man.
Aji, thank you for coming on and kind of talking about these important topics because, you know, while we could do three more episodes and go deeper and deeper, I think this gives at least the framework to start thinking about it. JI'S on LinkedIn,
So check him out. It's a great set info though. And then, and hopefully this gives you guys the first kind of way to start thinking about this new world.
You know, it's gonna be a long adoption period, as much as everyone thinks it's going quickly. Ai, depending on news, could be longer or shorter. But you gotta start driving it. But sharp problems, never go away. Ji as always, thank you my friend, for joining. It's been a pleasure. And hopefully we can talk again soon.
Oji: Thanks, Jeff. It's been a pleasure too. Thank you. Cheers.
===
Jeff: [00:00:00] All right, Ji. What's up man? Good to have you back to the show. How you doing?
Oji: Thanks for having me back. I feel like I'm getting up there in terms of repeats.
Jeff: It's been about a year since you've been on the show, but I saw you last week down in Austin and Dallas, so we're getting a lot of quality time lately.
Oji: Yeah, that's right. I'm not tired of your face yet,
Jeff: Not yet. Hopefully. So for, for folks join us , who weren't here last year, right? You're, you know, just to give quick background, you've been everywhere. You're product Microsoft, you were at Atlassian, Calendly type form.
You're at Twitter for a little bit. Maybe just give the, the folks at HOA, the 92nd TLDR , on who Aji is. We can go from there.
Oji: So, folks at home my name is Aju Dewe. I've been a product manager for 25 years, a product marketer for like two or three years. Dabbled in sales enablement and different parts of technology industry, corporate value creation machine. I'm mostly a product manager though, and for the last seven years I've been head of product and Chief Product Officer Atlassian Calendly type form Twitter.
And lately I've been spending my time in the global south [00:01:00] helping build capacity for building technology companies in Nigeria, in Nairobi, in Kigali in India. And I've also been working on building a venture fund to invest in the great replacement of our field intelligence B2B SaaS. So yeah, that's what I've been up to.
Pretty busy.
Jeff: And on top of all that, you've been flying around the US with me periodically talking to proc leaders , in all sorts of different cities. , So, a year ago we had you on to talk about the book that you and your wife wrote called Building Rocket Ships, about building great high performing product teams. we knew how much AI was going to disrupt this function over the ensuing year, but, you know, who could know exactly what it was gonna do?
That's what we're gonna talk about today, before we jump into it, right, we're gonna talk about the three speed problem. That you've brought about how AI has really caused, you know, speed increases across the board, but has not hit every team the same rate. Engineers, clearly they have cursor or windsurf or whatever they're using. It's changed their velocity probably a lot more than, you know, go to market teams or product teams.
But maybe first, , let's speed run building rocket ship. Let's fly through, [00:02:00] what's the, what's the shipyard model? What's sharp problems? What does that all mean?
Oji: The book is called Building Rocket Ships and it's collective 50 plus years of experience building hard hitting high growth technology companies. Some of them are the ones I've mentioned, but isn there was ahead of. New products at Time Inc.
Head of data and AI at Procore CPO at WP Engine. She's worked in hidden companies that are still high growth and doing amazing things. And so collaborating on that meant that we had a lot of perspective on all those 50 years. Now, you know, she's a product professional, product executive like me.
So we have a really good partnership , on thinking about it. And, and for those who are paying attention the book is actually pretty cool. You can get an Amazon, but there's a digital version with lots of practical content that you can use to make it part of your work life instead of just reading a book.
the main things we sought to do in Book one, was. About the fundamentals of product management, everything from how to pick a sharp problem, how to think about user experience, how to think about things like virality and design. And [00:03:00] all the way to pricing, which is like right there at the top of like specialized product management skills that you have to share with marketing and sometimes with sales.
And then book two is about how to lead teams that are really good at this stuff, right? How to think of product strategy, how to think of product execution, how to think of bringing everyone in the company together,
so we started out talking about the sharp problems. And this is the fact that we noticed that contrary to a lot of tech wisdom, the most important thing to success is picking the right problem because the right problems are forgiving and really this centers around. Optimizing the two forces of value.
You either make something like five to 10 x cheaper, or you make it five to 10 x better. Sometimes we use words like three to five x, but five to 10 X gives you the kind of ambition you should be shooting for. And so if you can find a workflow that is valuable, high frequency for your customers and you can make it five times better, you're gonna create value.
And even when you make mistakes, the market will be forgiving of you if you don't pick a sharp problem. In a high frequency quadrant. Mistakes like running out of cash for a day or a week, whatever. You're [00:04:00] just dead in the water. The other thing we talk about is shipyard teams. You have to think carefully about the mix and autonomy of the teams you used to execute.
So we identify like six skill sets, not people. Not actual individual people. Skill sets are required to do thinking and shipping and go to market. Product management, engineering. Product management is little like, why are we doing this? Is this the right thing to focus on? Product engineering, which is building design, which is part of like, how does this thing look and feel?
So when we get to solutions, user research, which is essentially mind reading data and AI and ml. We've had data for a long time. It's morphing into data engineering and product marketing and storytelling. If you can get these six skill sets, either in one person or two or six people, you can build a thousand person teams who work on interesting problems for your company.
And of course you have to remember that. Orbiting this core team is go to market you know, customer support, the skin, the skin of this big, you know, beefy Android because you have to feel the market and you know, [00:05:00] those sales teams, those. Customer support teams talk to customers so much in a way that you can't, and if you mind their intelligence, you'll become much better at building available company.
So these are the core things. We go into a lot more things, but the shipyard framework is very important that people talk about that. We give people language for it, and the sharp problems for framework is very important because it's a precursor to creating value.
And you know, these are the fundamentals. We've had 40 years of technology, you know, core software technology and maybe like 60 years of silicon, right? From Fairchild, Intel and so on and so forth. And what happens is every 10 to 15 to 20 years, we get to a new technology level.
So at the, the first level was just microchips. Okay. That was Intel. The second level was desktop computing. The next level was internet HT TP. Right. And internet protocols. And the next level has been artificial intelligence. In between, there were some mini boomlets, like social
Jeff: Mobile. Yeah,
Oji: and mobile.
Jeff: But doesn't change the fundamentals here, how
Oji: Yeah. They accelerated some things, but AI is like legit the next level. So it's like a [00:06:00] staircase. Of new technology and every time that happens, we actually, if you pay attention, recreate everything again because. There's some fundamental workflows at work and in and in leisure and leisure.
Things like dating. Or sports and betting, those are the killer apps of leisure, right?
And then the next in, in work it's like, okay, communications is huge, right? That's front desk to CEO and then there's specialized workloads like the CFO and accounting and so on and so forth.
Every time the technology level changes, we recreate the entire software world. We always do. And so. The book Shipyard Team, shop Problems, those, those are fundamentals regardless of technology level, the skills, the listing you do to the book, you can apply. Now in the book, we actually wrote a whole addendum on art field intelligence and how it impacts everything we just wrote.
We spend the last nine months actually deepening that. You know, Izzy and I see ourselves and you know the crew at Product Mind. She, us to the future of AI and software. we wanna point in the right direction of where things are going so that people can waste less [00:07:00] money, be more effective at making their teams better.
And so if you go to the podcast, you go to substack, you'll see us charting the course, and I think it's very valuable to check it out. So, but these fundamentals remain they'll never be. Out of, out of fashion. And so that's why you should ingest that. Now we can talk about the implications and we're about to talk about a three speed problem, but I just wanna make sure people understand the framing.
Jeff: on that front, I guess, you know, if you have, let's take for granted, people are, you know, operating in that framework correctly, and they, you know, they, they've got the great shipyard team. Obviously the big enhancer, like you said in the past couple years has been.
This growth of artificial intelligence. And, you know, AI comes in many flavors for, for anyone who thinks it's brand new, you know, past two or three years. You know, it didn't start with chat GPT I mean, Watson goes back to the teens and, and there's been natural not, sorry, there's been there machine learning since
Oji: Mm-hmm. Mm-hmm. Mm-hmm.
Jeff: But there's more to it than that, right?
You kinda lay out five flavors of ai. We can go into it or not,
Oji: No, no. I do, I I really want to, so,
look, if you were a civilian, keep scrolling, You can call it AI all you want, [00:08:00] right? But if you think about it, artificial intelligence mean it's a meaningless word. It doesn't mean anything. And you know, even the word we call a G, artificial intelligence is also meaningless because you most, 99.9% of you can't tell me what it means and define it.
And actually we can define artificial intelligence. So what I ask builders, software executives, product people to do. Is to use different, more compelling and clear words to think about actual intelligence. Because if you can change your language, you can change how you think about it. And so one of the things that product Mind we've done is to use the framework of new capabilities.
If we say technology level has changed. It makes sense that there are new capabilities that didn't exist before that have come about. But what are they saying? Artificial intelligence doesn't elucidate? What does that are actually clouds it and people can talk for four hours about AI and talk about different things.
So what does ai, what are the new capabilities of artificial intelligence? There are five [00:09:00] things we've identified that are net new or net new in the form based on transformer based large language models or what I call large media models that have come about in the last three, five years. The first one is content generation and transformation.
This is everything from make a marketing newsletter to generate code. To soar, to generate video and images and so on and so forth, you know, to nano banana.
Jeff: Yeah.
Oji: LLMs can understand us even imperfectly, the same way humans can understand each other, imperfectly with bad grammar and slang or whatever. And it can create things that approach very skilled people, right?
That's one. It's new. Isn't that new? We've tried this for many, many years. It's new. And so if you think about it that way, you can build new products on that capability alone. You can measure how good it is at doing that alone. Each of these things has its own metrics, which is really cool because then it's beyond just artificial intelligence.
The next one is natural language interaction. I can't tell you how long I've worked on natural language input and output, Microsoft and Atlassian. Microsoft has a long history of pen [00:10:00] computing, voice computing, acquisitions, dragon naturally speaking, blah, blah, blah, blah, blah. This is the first time it's really good because the human brain is really good that this stuff.
And so everything we did before LLMs sucked, right? Even Google's, you know, transcription was like. 90% for a very long time. You're stuck with machine learning technology, so that's net new whisper, all these things. The next one is information analysis and data synthesis. You can throw a snowflake planet worth of data to an ai.
If you have the right compute. It'll understand it and understand it without you having to clean it up as much as possible. This is also net new. The fourth one is personalization. AI and I did this at Typeform AI can literally personalize and understand 8 billion people and give them exactly what they want.
This is new. It's a little terrifying actually, because if you think you can ad industry and so on, you're salivating, you are, you know, at at Typeform, if I and the other companies, if I got away with sort of mapping people into like a four by four 16. Quadrant and say, you know, people [00:11:00] are in this quadrant, I'm gonna make the software behave like this.
People are in this quadrant. I'm gonna make, if I recognize that I'm gonna make the software behave like this. And this is the kind of thing people like me did, right? Personalization was course. With ai, you can go so granular that no two human beings will see the same response on anything, anytime, anywhere else.
So power also terrifying. Then this last one is autonomy, right? Autonomy is important because this has been a, a holy grail. When we talk about robots, robot dogs, we talk about all this sci-fi stuff. What we are really talking about is autonomy, is you feed a thing and intelligence a lot of data, and they can go off and make decisions and come back with results.
That's also net new. Say, you know, every conversation, ai. And you say these five things, or which one of these five things, and you have this conversation with your team yourself. If you're moderate to yourself or your boss or the board, you, you will have stronger conversations immediately.
That's our gift to people who are listening because I think it'll change your life.
Jeff: Hopefully everyone wrote that all down. 'cause I, I've definitely found [00:12:00] just being able to elucidate with a better language has, has helped make stronger product decisions. So I recommend it. Alright. Now the big thing, this is the, the peak behind the curtain a little bit. 'cause we've been, you know, as Aji and I said we've been around the country talking to product leaders and, talking about this three speed problem. So, you know, AI has done revolutionary, revolutionary things but, you know, the killer app, really from the beginning everyone looked at at least, I dunno, maybe I, I did and we've talked about this code, code makes just sense.
It's deterministic at some level. You can look at it, it's functional, it's easy to understand if you know the right things. It seems like AI was. Built for that. But that's only part of delivering great software products. So hence gives us, you know, the, the thing that you and s Nate have been talking about a lot lately, which is this three speed problem maybe lay it out for us.
You know, it starts, it starts a product, ends with go to market. You got engineers in the middle. What?
Oji: Yeah.
Jeff: ma'am?
Oji: Yeah. So, by the way, there's a whole keynote we've done did that industry, which is somewhere online, so people can, if they want an extended take on this they can get to it. So the AI capabilities to eat my own dog food that we're [00:13:00] really talking about is content generation. Now, content generation is one of those things on the a GI scale where AI has gotten really good, .
Sometimes better than humans. It's still kind of annoying way it composes it and m dashes and stuff like that. But it's really good because it takes like the, the whole knowledge of the internet to generate something for you, which is better than most humans can do.
But this has set up a new conflict. And the core unit of software development. Now, what is the core unit of software development? It is the squad, the team that makes stuff. We've talked a little bit about the shipyard team, which is sort of a new way to think about what that team looks like, the skills it has, how it works together, but the core.
Limiting function of our teams for the longest time, maybe 20 years, 25 years, has been the speed of engineering, right? Everything we've done from, waterfall to Agile has all been to optimize slow engineering. Even the fact that in a company, a technology company most of the head count. [00:14:00] Is developers is because of the rate of development.
So if you're following along, there are three core things that you have to do to make valuable software and then create a valuable business. The first is you have to divine markets and customers find their problems, listen to them, read their minds, synthesize a new solution to the workflow problems they have.
Design it, boom. First part,
Jeff: I I gotta be sorry to jump out. I gotta be honest. Sometimes when you say things, it, it sounds a lot simply like, oh, that's all we gotta do. But that one there is like, oh, that's all. You just have to read their minds and, and, God, it's amazing. We have anyone has ever designed anything decent, like that's a pretty high bar.
I think back to write the five, 10 x better thing. That really is it though. If you want a, a fast growing company that's going to grow and, and really, you know, be able to hyper grow like companies are looking to that is the bar like, it sounds difficult because it is, but
uh,
Oji: it is.
Well, this is, this is a soul of, you know, people keep asking what a, what a product managers do. Well that is what we do, right? We, we read customer minds. We read [00:15:00] Market Minds. We pay attention to where people have pain, and depending on the technology level, we compose a new solution that accelerates them better than it did in the past.
And because we do that correctly, when we do that correctly, people pay us and make our companies rich, right? And then we take all the externalities of software development, which is in, you know, like zero marginal cost. If we get to a million people, we make a lot more money. And that's how technology companies, this is why the richest people in the world, technology companies, they build it on the back of good product managers Now that's the first part and it's very hard and actually 90% of companies miss it.
This is important. Most companies can't actually get around building a great product. Every great product you see there are nine that try the same thing that died. So this is a huge leverage point. Well, let's not belabor it too much in the middle. Once you've figured out a solution, how its form and function should be, you build it to engineer your write code.
Now, in the past, code has been. Code has been static. You know, you build it back and you build on the front end. You write static [00:16:00] code, you make it complicated to handle scale, to handle errors, to handle security, to handle all the ways that users can screw up software. And then you get it right, hopefully.
And then on the backside of that, you make the software get into the hands of customers. So you do all kinds of things. You do storytelling, you do marketing. Sales and so on and so forth. Those are the three things that make a Val technology company. What we are saying is that the middle, which used to be the limiting factor, is over the next 10 years going to accelerate by five and then 10 x maybe more, right the other day.
I think Claude in an experimental session, held enough of its attention to remake Slack in 35 hours in one shot. They rebuilt Slack. So think about what's possible in two, three years, four, five years, right? You can build entire products. , By the way, you can't actually do that easily. You have to put rules and tell it exactly how it all these things work, and it can do it.
If you say go build Slack, it won't do it for you. You have to be very clever about how you do that. And even when it does that, it's still like. For [00:17:00] first draft, you still need real developers, but that's still incredible. So this middle is gonna accelerate very fast. Now the question is, and the why, it's called a three speed problem, is again, conception and solution build, go to market.
Those are the three speeds. This thing is gonna go fast, and what's gonna set up is a conflict if the product management and design doesn't accelerate. It is gonna log jam on the capacity to build. If the go-to market doesn't accelerate, it's gonna log jam on the capacity to build. So this is an inversion of a problem we used to have.
So over the next five to 10 years, maybe shorter, most teams in the economy, surely in advanced economy like the United States, have to figure out how to impedance match these three levers for growth. And what we talk about is how to do that and all the different ways you can do that. We definitely have to speed up our customer market divination.
We definitely have to speed up GTM because this thing is gonna go and we talk a lot about how to do that, how to think about it.
Jeff: it might not. Fully log jam building, right? Like let's say product can't keep up. The, [00:18:00] the worst case scenario is not actually potentially that engineering has nothing to do and they just sit there twiddling their thumbs. It's that they just keep doing anyway and you get all sorts of half baked, features that maybe
Oji: Yeah. But, but doing what? Let me give you a, a work If, if you have a ship with a warp speed drive.
Jeff: right.
Oji: And you're pointing at 360 degrees of space. If you go fast in the wrong direction, you, you, you've just killed
Jeff: No. That that's what I'm, that's what I'm saying is, is the worst case scenario is not that engineering stops because they're log jammed. It's that they're in a rocket ship going warp speed the wrong direction, and
Oji: Yeah. And they will never stop who, who has ever, who has ever made an engineering team stuff.
Jeff: No, I mean, I think, I think what you, you were talking about when we were in, in Austria, Dallas, you were bringing up the team where the CEO started vibe coding something and they released it and they got the CRO on board to start selling it, and the fricking thing just didn't work.
And so they were pushing this thing out into market because that's the risk of going fast without, without
Oji: Yes. No, a hundred percent. A hundred percent. No. No. It, it is, it can, it is actually like harmful. But I will say [00:19:00] this, actually even without speeding up product management, people can start to. Spend this smorgasbord of capacity teams that are definitely looking at me right now with side eyes thinking, what's smorgasbord like?
We have not seen a smorgasbord, but it will happen, trust me. If you have a good CTO, it will happen. , But even before you speed up, I, you should. There are also creative ways, all this tech that we keep yapping about. We can rewrite it faster. Just start afresh, build it on AI first.
Backend that you've been trying to go from. You know, we went from monolith to microservices and back to some micro monolith.
You
Jeff: and
forth. back and
Oji: you,
you can,
you can restart this stuff because you're gonna need a new code base anyway. That's very model aware. To do that. You can spend that capacity on. Refinement, right?
Make your product like a jewel. Every pixel looks beautiful. Every moment is a reward. You can do that. You could spend that capacity on experimentation, so not products, but making multiple iterations of something faster than you could ever in the past. So that now instead of [00:20:00] like making testing with customers, you make 10 times, you test each general customers and the one they like goes forward.
And you don't have any lag in customer adoption. There are actually creative ways to use that capacity today without increasing the speed. But eventually those things get tired. Those things get boring for engineers. And frankly, what we really need to do is make the decision making the direction finding faster, and I think that will make everyone more satisfied.
Jeff: We've talked about this, we've seen teams doing this or starting to you know, one team I talked to has picked up the practice of, whereas they used to take two weeks to do a design sprint and kind of come, you know, over two weeks they'd work with designers and product and, and other things.
And at the end of two weeks, you would have one. You know, flat file Figma prototype. You know, I use air quotes for the prototype because it is just, it is a picture. It was a nice picture. But now they're talking about what they've instead done is they've basically looked at what could we really do here?
And they will take, you know, a couple hours each to go through and kind of use, you know, [00:21:00] one of the many vibe coding tools to prototype quickly and
Oji: Mm-hmm.
Jeff: of an idea and come out with eight takes on it. And in an hour they have eight fairly functional.
Oji: Mm-hmm.
Jeff: Takes and they actually will ship that to, to a small live kind of alpha group who will test it.
They'll get feedback and so they're able to go to engineering with ideas that are, you know, they're not coding the final product. Like you said, you still need engineers here, but their speed of ideation, but also their ability to try. Riskier things and throw away the trash and find the, the nuggets of gold.
They've been able to, the CPO said like 20 x their velocity. I don't, I dunno if I buy
Oji: That
is, my, no, that is my favorite version of solving a three speed problem. I want your listeners to pay attention to the beats of that solution. 'cause it's very clever.
You listen to customers where you don't like. Go to the nth degree, right? So that's a speed up right there because you know, go going to 80, 90% of certainty is faster than going to 95, 90 9% of certainty. And then you take that knowledge and instead of like writing a PRD, you write an app and maybe in [00:22:00] certain cases you write six apps.
Right now the app is the PRD. But now you can do it in code versus write it and you will, don't tell your engineers that you'll ever ship this stuff 'cause they'll get mad at you and then they'll start fighting you and nothing will ever get done. And then watch this. You ship it. To willing Guinea pigs, right? Because they're always the willing Guinea pigs in your audience who are like, gimme, gimme, gimme, gimme, gimme. If you have a good
product,
Jeff: the newest, the fastest, the best,
Oji: right? And so in the first phase, you've now done a ship cycle. You've done a full ship cycle.
It's super fast and because you are working on volume, customers are whittling down. What works for you right now in the old way, you have to do a lot of divination. And pass it through an expensive bill process. And at the end you could still be wrong, but in this one you are working out wrongness very quickly.
And then when customers, by the way, small set of customers will not tell you everything you need to know. Like you could still have a small set of customers like something and the end fail. It happens all the time. At Twitter, we happen all the time.
You know, when you get to a million it's very different than when you got to 10.
But then. Your engineers get a [00:23:00] prototype hopefully in the same stack they use to go forward if you're smart. Right. So they don't have to throw it away completely. And then they put force behind the thing. The engineering team can then spin off a hundred different variations of that one.
Right. But it's now more in line. It's like increasing accuracy. , That's one of my favorite versions of how to. Tackle this three speed problem.
Jeff: Yeah, so there's a lot more to this. Clearly we could keep going on. I think there's the fourth speed problem, which is there's only so many hours in the day and, and you're a busy man.
Aji, thank you for coming on and kind of talking about these important topics because, you know, while we could do three more episodes and go deeper and deeper, I think this gives at least the framework to start thinking about it. JI'S on LinkedIn,
So check him out. It's a great set info though. And then, and hopefully this gives you guys the first kind of way to start thinking about this new world.
You know, it's gonna be a long adoption period, as much as everyone thinks it's going quickly. Ai, depending on news, could be longer or shorter. But you gotta start driving it. But sharp problems, never go away. Ji as always, thank you my friend, for joining. It's been a pleasure. And hopefully we can talk again soon.
Oji: Thanks, Jeff. It's been a pleasure too. Thank you. Cheers.
