Why Some AI Products Strike a Chord (and Others Don't)

Speaker A: Foreign. So as we record this, we're a few weeks away from the 31st anniversary of Bill Gates's Internet tidal wave memo. I don't know how many of you have read it. It's, it's probably worth a read, but I'll summarize it for you. It was basically, rumor has it, he kind of headed off into the wilderness for a couple of weeks, went off to a cabin somewhere and had to think about what this Internet thing was doing. Microsoft, of course, you know, it wasn't invented there. It was something that kind of happened to the industry. And Bill came away with this two page memo which was really a company wide call to action for Microsoft employees to rethink product management in light of a technology that, to paraphrase Bill, would result in a lot of uncertainty as we first embrace it. That will change so rapidly, it will require us to revise our strategies from time to time and, and it will require better intergroup communication than ever before. I don't know about you, but swap out the word Internet for AI or ML and the message today really still resonates, right? We're dealing with a set of technologies just like in the early days of Internet adoption where there's considerable pressure to just add AI back in the same way it was. This people would just add web or HTML to existing product. But that kind of pressure, especially from the top, from boards, from executives, can lead to some performative efforts. I, and I hopefully don't offend anyone. This, I kind of call it AI karaoke, right? You kind of recognize the music, but the result just doesn't sound right when it comes out the other side. And we see that a lot with products that are just, I suppose people are desperately trying to bolt AI onto them, but it may not always be authentic. So in this episode, I guess Marlon Davis offers up his take on how organizations can move beyond service surface level, I should say, AI initiatives and build products that truly deliver. Welcome to another episode of the AI Forecast, proudly sponsored by the incredible folks at Cloudera. Some amazing product managers there, I should say. I'm your host, Paul Muller. Join me along with leading companies and industry experts every week as we explore the past, the present and the future of data and AI in the enterprise. Well, our guest today, I mentioned him briefly, is a former NASA JPL staff, A second NASA person we've had on the podcast. It's starting to turn into a bit of a, a NASA tribute band. This, uh, podcast. Thinking about it, he's now fractional Chief Product Officer with a track record in private equity backed SaaS product management spanning nearly two decades. Um, he writes extensively and interestingly compellingly on the topic on his LinkedIn frugal product manager newsletter. Welcome to the podcast, Marlon.

Speaker B: Thanks for having me, Paul. Appreciate it.

Speaker A: It's great having you here. Excuse the long introduction, but, you know, as I was getting ready for today's episode, um, I don't know something about the topic and the idea of putting myself in the shoes of the product manager and thinking, you know, this does remind me of 30 years ago, which I am unfortunately old enough to remember, um, where this Internet thing came at us from out of nowhere and, and you were either an Internet native and you'd been thinking about this for a long time, or you had to kind of play catch up with everyone. And I do remember it created some really interesting and frankly a lot of crappy products at the time.

Speaker B: That's, uh, true.

Speaker A: So yeah, you might have a point of view on that. I don't think. I don't know if you're old enough to remember this, but we get on to that. Oh, you are. All right. All right, feel me then. You're feeling me.

Speaker B: I love this.

Speaker A: I'm even wearing my hoodie, by the way. I don't know why you wear a hoodie on this show, but I thought we're talking product management. I'm channeling my inner zucker for today's episode.

Speaker B: All right, there you go.

Speaker A: We do something, um, we, uh, call the fast four. We, uh, call it different things. I call it dessert before dinner because we put our lightning round at the start of the podcast. No one else does that. We're different by design.

Speaker B: Mhm.

Speaker A: So you ready to do this?

Speaker B: I'm ready.

Speaker A: Let's limber up. All right, first question. What would people who know you best could be anyone. Your friends, family, colleagues, enemies. What would they say is your superpower?

Speaker B: I think this one's going to be pretty cliche because I had a chance to read, uh, and I should say watch some of your other podcast guests. And mine has been simplification too. So taking complex concepts and being able to simplify them and communicate them in a way that people can execute on them.

Speaker A: Uh, you know what? I think it's product management 101 though. If you're not able to simplify, you

Speaker B: don't know what you're talking about. Yeah.

Speaker A: What did Steve Jobs used to say is at first you simplify, then you deep dig deep and add complexity when you really understand the problem, and then you simplify Again, if you can't simplify

Speaker B: it, you really don't truly understand it.

Speaker A: I love that. All right, next up, uh, the most impactful technology of all time.

Speaker B: I want to go outside of my field and say medical technology and advancements.

Speaker A: I love that.

Speaker B: If we're not around, we can't invent all the rest of this other stuff. So I'm going to go with that one for 500.

Speaker A: I love that. And you know what? You were on a, a podcast, uh, recently from the Pragmatic Institute. Is that what they call themselves?

Speaker B: That's right. Mhm.

Speaker A: That sounds pretty pragmatic to me. If you're dead, you're not innovating nothing, right?

Speaker B: You got to start there at the, at the base of the pyramid and work your way.

Speaker A: Mvp, can you fog this mirror? Beautiful. Let's go. All right, excellent. All right, next up, what would you like to see AI automate for you in your future? Just be selfish for a second.

Speaker B: I'm going to stay in the same vein. Medical advancements, apply AI to that all day long. Let's fix these diseases that we have out here. Let's improve people's quality of life. Uh, let's just park all the data center, compute on that all day long.

Speaker A: From my perspective, last but not least, you've been working with data all of your career. What are some of the best practices you've learned as someone who works with

Speaker B: data going to say? The very first one is making sure you know what your outcome is supposed to be, because that's going to be your guiding and your North Star. Everything else is getting in the weeds. And people, as they say, I think the, uh, proverbial phrase of the young kids getting lost in the sauce, you don't want to, you don't want to get to that level. You have to stay above and use your guiding light.

Speaker A: So let's jump into this thing. So, I mean, there's so much interesting stuff to talk about. I'd love to just talk about the world of product management at the moment. Business leaders really need to start to think like product managers. You know, if, you know, if you've ever read from project to product, right, like the idea that, that we've got to stop thinking like project managers, beginning a middle and an end, and start thinking like product managers across the board. So we could do a whole podcast, I think, on the changing world of product management and the need for maybe people like yourself as fractional cpo. But let's talk about what's, I guess what's shaped Your career. Tell us, give us a bit of backstory about your journey through the world of product management.

Speaker B: I've always been centered on solving problems for customers. Back when I was an engineer moving into product management and then becoming a head of product, it's just a different angle on where I was trying to solve that problem. And so as you look at my career over the years, you know, from uh, being an engineer to being a product manager to then becoming a head of product, it's pretty much been that focus on the customer and trying to understand how can we solve the issues in their businesses much better than it's being solved right now. I think there's a quote from a study by pindo that says 20% of the product features are doing all the heavy lifting for the customer. And so if we can improve on that just a little bit by understanding the customer that much better and solving their problems, that will, you know, accrue to a huge benefit for those companies that are able to adopt that and improve their product. Uh, user fit.

Speaker A: I mean, obviously there's that old adage. I think it was Henry Ford's quote, you know, if I'd have asked my customers what they wanted, they'd say a faster horse.

Speaker B: Right.

Speaker A: Which is a way of paraphrasing, I guess, that one of the challenges of being a product manager is you're simultaneously, I suppose, trying to gather data by, by talking to the users. But I suppose a lot of it also needs to be like the really interesting breakthroughs, uh, I guess are must be developed by watching what they do. Not what they say they're going to do, but how they behave. How much is the sort of, I guess, the anthropology of product management important to what you do?

Speaker B: It is huge. I was trained on, um, I'm going to give a shout out to Karen Holdsblatt. I don't think she's doing this anymore, but I got field research techniques training back in. Oh, now you're going to know how old I am. Back in the late 1990s, I've used those field techniques in order to go out and shadow and watch what customers do. And then the real superpower there is to be able to put them into models, geo models, cultural models, flow models, to understand, you know, and kind of craft that stuff into patterns so you can analyze it and make sense of it as opposed to, you know, many times we go out and we'll listen to customers and we write our notes and we come back, but they're not usable, they're not actionable. And so when I got that training that really catapulted me to go, okay, every place I go I'm going to spend the time to understand these customers and I'm going to shadow them, I'm going to watch them, I'm going to ask questions, but first and foremost I'm going to be a detective about understanding their business processes in order to solve these challenges and issues better.

Speaker A: I just get the feeling, and I'm going to get hate mail over this, that a lot of product managers I've met in the past, with all due respect to them, are uh, software developers who, who I guess from a career perspective their manager probably didn't know what to do with them. And like the next step in promotion is you go from being the head of R and D to being a product manager or the head of a development team to being a product manager. I don't know if that's actually true, but my observation is a lot of them don't have that kind of training. Is that fair or am I misunderstanding the industry as it was?

Speaker B: Most budgets don't account for enough training anyway to train product managers to do so. Our, our budgets are somewhere around 80 to 20 from the headcount to descript discretionary. And so if you take the ratio of, I don't know, let's say 100 million dollar company, software company, that's a really small budget, somewhere around maybe 1 or 2 mil for that product team and 20% of that is going towards training going out to visit customers. So there's not a lot that's going towards helping product managers and even product designers who should actually get this just coming out of school, uh, get these skills and techniques in order to translate customer conversations and job shadowing and all of those good field techniques and bring them into the organization where they're actionable and actually can be sustained and kept fresh and not become stale. Uh, you could walk into several organizations and ask them, hey, point me to your user database and show me all your Personas, journeys, you know, any sort of infograph about, you know, how their needs have changed over time and you might get maybe 5% of them to raise their hand.

Speaker A: Yeah, okay, that makes sense. Then it kind of validates, um, or at least uh, corroborates some of the data that I was seeing. So now listeners might be wondering, oh, it's all this got to do with AI and I'm kind of leading up to, and I guess I mentioned this in the intro is one of the challenges, you know, it's the Cliche is if everything. If all you've got is a hammer, everything looks like a nail. And there is a lot of pressure at the moment. Moment to start applying the AI hammer to whatever business problem nail we can find, even if it's an inauthentic one. And I want to be clear here. This is not the first time we've seen this. I mean, uh, you know, three years ago, everyone was talking about the metaverse and how we all needed to build and add metaverse capabilities, and billions of dollars got torched doing that. I think there are a lot of, you know, Fortune 50 brands that built giant premises in the metaverse that now lay fallow. Before that, it was blockchain or big data. You know, the list goes. You and I have seen it. All right.

Speaker B: That's right.

Speaker A: Um, and some of it becomes enduring and. And lasts. Right? There's, there's. There's mileage in there, but a lot of it. I call it karaoke only because it just feels like everyone's kind of singing the same words to the song, but not everyone's Taylor Swift, if you know what I mean. Right. They can't carry the tune. What's your take on AI? Because I think we're seeing at the moment a lot of promising opportunities appearing. There's a lot of excitement about applying AI to problems like medical and so on, so forth. But I guess I get the impression from reading some of your background that you feel that people are maybe missing the mark with how they're applying the technology to the discipline of product management.

Speaker B: Thank you for that feedback. That means that the message has resonated. The message is coming through. Am I right? That is 100% true. When you look at AI, just like any other technology, and specifically this one, I'm working with a client now. I'll just. I'll couch this answer in the context of something I'm working on right now. This client, the engineering team, is utilizing AI coding tools. That means we can pump out features that much more faster than we could in the past. Just in general, when you apply that to other development teams. But what hasn't changed on the front end is more clarity around the customer problems that we're trying to solve, giving product managers time and the skill set training that we just previously talked about, to be able to go out, listen to customers, and bring it back into actionable results. The key there is something actionable, not just going out and doing activity, but to come back and put into something actionable. Imagine you had a gun that you Only, you know, a revolver, you can only fire six bullets at a time to hit the target. Now you've got a gun, a machine gun, you can fire off many more bullets, but you're still missing the target. Okay, so I can fire off many more bullets, but I'm not any closer to the target. What you need is to be able to scope this and get, you know, utilizing product management to get closer to the bullseye in order to solve those customer problems. And so I think when everybody's hopping to bolting AI onto their solution, they're not really understanding the customer's business operations, the challenges, the workflow obstacles that you're dealing with to have a, uh, kind of impact, kind of a hierarchy of things that you want to go after and solve. Right? It's just, oh, uh, well, we know what the customer needs. That is incorrect. Let's go find out what the customer needs, rank those issues and knock off the most impactful ones for them so that we can aim a whole lot better at the bullseye.

Speaker A: So I have a friend who consults professionally and has for the last 30 plus years, and she's going to get angry at me for dating her now, but not dating her. Sorry that, you know, Karen, uh, and uh, um, uh, some other friends of hers, uh, work in that. In the call center world, um, consulting and improving processes in the call center world. And one of the things she said to me about this AI technology, she said, look, this was very reminiscent of when we had things like omnichannel communication come in unified voice and so on and so forth, where she said, the problem we find is that the technologists want to apply the technology, but they don't look at how the business process is working.

Speaker B: Right.

Speaker A: I would say at the moment, famously, one of the applications of AI that was almost like an obvious one for product managers to go and jam is let me jam chatbots into the customer experience. And I've had more than a dozen people reach out to me and um, friends and professional colleagues and just say, look, I've been boxed in to the point where I just needed to speak to a person because the chat bot had gotten itself in the loop. And, and these companies have basically applied AI and have pushed me away from their company and I'm going somewhere else. I'm, I'm going where the humans are because I need to speak to someone. So that's an example. But can you think of other examples where they're, where there's a disconnect between what companies think they're Building with AI and what the customers are actually getting out of it. I just picked on that one because it's, I think, one we've all experienced.

Speaker B: Yeah, I've seen those where we're going to craft different copy within the application. That's a common one. Uh, let me take a requirement and refashion this requirement within the application so it speaks, uh, a little bit more with grammar, correctly with the grammar flourishes, uh, that are needed. And then the application is AI specific or to your point, let's take unstructured data and let you answer, ask any question of it. Not answer, but ask any question of that unstructured data. Because we don't really know how you're going to use this application. So we're just going to apply the chatbot to it, like you just stated, and ask all sorts of questions of the unstructured data instead of, wait a minute, how do we help them pull the signal out of that, um, unstructured data within the particular business processes that they've been out there trying to uncover, discover, peel the COVID back on it and truly understand it and then walk that signal all the way through to a product or product feature that's going to get delivered back to that customer that they previously interviewed or talked to or job shadowed. So I'm kind of tying it back to our previous, uh, conversation around what, doing discovery and trying to understand the customer better. When you don't understand it, you just kind of reach for generalized tools that used to be search in our tools. Hey, I don't know what they need to do. Let's throw a search box up there and they can just type in whatever they want and find what they're looking for. Now it's, hey, let's just throw a chatbot up there to your point and let's just figure out whatever it is. I think also people don't have a good understanding of the technology. And so when you look at software, not to get too wonky and technical, you know, programming languages are deterministic. X y equals z. When you're dealing with gen AI and LLMs, it is not always x y equals z because on the back end it's looking at the probabilistic outcome and choosing from those. And so these are all weighted options that it can choose from and it'll throw it out to you, which is why if you run the same thing three or four or five, six different times, you get slightly different answers on the back end here. So people need to really understand the technology they're using and say, okay, that's not appropriate for here because the answer needs to be one to one or very concrete versus oh, we're looking for patterns. This is a perfect application for LLMs. We're looking for patterns that are previously too difficult to try and figure out.

Speaker A: Maybe taking, uh, the opposite view. So we talked about where the big disconnects are. You mentioned value and results, and that being a really big part of how you kind of grade your own scorecard. Is there a good test for whether or not an AI feature is actually generating value versus making you feel good? Because you've slapped the latest buzzword on a product?

Speaker B: Two ways. If the quality of the output is significantly better or if the process that you previously were using is significantly shorter and more efficient. A tool called an ETL tool, which is extract, transform and load. It's pretty simple. Take a CSV file, it's a structured format. You're going to extract the data from it and then you're going to put it into a database. Another structured format, you can create a mapping tool that says, take, uh, data from here, put it there. But if there was any sort of deviation in that CVS CS CSV file, now that no longer works. But I can take an LLM and say, hey, I'm just randomly looking for this information. This is what you need to populate on the back end. And I can send through the prompt and the skill and say, this is what you're looking for. This is the data structure that I want. It can do it. It can handle those different variations that previously we were kind of stuck with. When you look at traditional ETL tools or I had to write a mapper for every deviation that I wanted to do, well, that has improved that whole process tremendously. Now there's some tweaks on the back end here. And I know people that are really technical are going to say, you kind of glossed over this, but this conversation doesn't need to get into the weeds of that. The first one though is the quality output. And I spoke to that when I talked about the unstructured data, where you had to have way too much, you don't have enough time, you don't have enough hands to take all this unstructured data, try and put it into some sort of model for it to understand what you got and see if there are any patterns. Well, with an LLM, I can apply, slap it to it and say, hey, this is the model I'm looking for. Pass it to the LLM, pass the structure that I want to see come from it. And it can take a fairly decent, you know, swipe at it and give me something that I can visualize and see and go, okay, I kind of see the pattern in that data. Those are two areas where the quality was improved on the back end or the time to get there was, you know, was significantly shortened, which was the ETL tool where we didn't have to go through and create a map or every time we wanted to do something we could, uh, apply the same process.

Speaker A: I want to touch on the idea that the technology is not as deterministic because I think that that creates an interesting situation where the feature may work well in your, you know, in a, in the product management, uh, R D teams wind tunnel. Yeah, where they've got a lot of control over the data that it's, that it's being exposed to. It's a relatively well understood environment and everything seems to work pretty deterministically. Right. Like they're getting the outputs they want and then they put it out into the wild. And of course, you know, the users are going to do what the users are going to do. And uh, my own personal experience has been there are some of these AI enabled capabilities. Whether it's around doing things like, you know, fuzzy logic, pattern matching, um, helping me, I mean even do something simple like match my expenses to my receipts, right. Cannot work often enough that it kind of becomes less of a feature and more of a frustration. How do you, as a product manager, I guess you're nodding because I suppose you've seen some of this, maybe experience it yourself. How can product managers figure out whether or not they're actually frustrating the users? Because you could, you know, you could go, hey, success. We've released this amazing thing, it's out there, but maybe your users are not getting out of it what you think they're getting out of it. How do you figure that out?

Speaker B: So I'm actually going to say that's even easier to do now because of LLMs. Right. So you can lab a lot of things now utilizing AI coding tools where you can literally say, hey, look, let's do that, let's simulate that three way pattern, that three way invoice processing process that you just mentioned. A three way match is actually what the name of the process is, where you're going to match the invoices to, uh, your receipts to what did you actually receive so that you can match it out and say, okay, we should go pay this vendor. Traditionally we did use machine learning and that's still being used right now because that's a more deep learning kind of AI, uh, technology that's more suitable to that. But we'll keep it here with LLMs just for this conversation. So if I can create that and I can tie an, Maybe tie my API account to the back end to ChatGPT or Claude and I can create a lab where I can take those three different documents and I can try and do the fuzzy math on it. I can lab this with several customers, multiple customers if I want, maybe a population of 3:30 and try and run it through, get them all to use it and say, hey, what's the impact on you? Is 60% enough? 60% match enough before we have to pull humans in? And they may say no because I still have to dedicate my time. And I'm sure, I'm still not sure about the 60 that you got there because of the error rate. Maybe if you can get to 80 they go, uh, you know what? 80 is perfect. I do believe that I don't have to go back and look at the 80, I'm just only going to focus on the 20. So you can't do 100%, but you're still bringing value into my life by removing. I get maybe 500 different invoices that I'm responsible for as an account, you know, payable specialist. And man, if you can get through 80% of those in a, uh, in a month for me, I take that. I definitely would take that.

Speaker A: Well, let's talk then about one of the other topics that comes up a lot. Kind of gives me the cringes sometimes, this notion of product market. Well, just sometimes people use phrases and you sit there thinking, you know, uh, it sounds more like management by magazine. Product, product market fit is one of those things, right? People say it a lot and I'm sometimes wondering whether their definition of product market fit is. I've developed a product, now let me keep smashing it into the market until it fits. Do you know what I mean?

Speaker B: Yeah.

Speaker A: Ah, yeah. Versus how I think about it, which is, you know, I've identified it need and um, I've crafted something that solves a problem. I've solved a problem that I've imagined exists and now let me ram it down people's throats. How do you think things like topics like product market fit or disciplines like product market fit, um, might change in the world of AI?

Speaker B: I don't think that AI makes the change or the shift so much as you have to have a strict definition as what you're saying here around product market fit. In my business I actually use a capability system that focuses on three different fits. One is product market fit, product user fit, and product um, execution fit. And so the first one, product market fit, is really focused on just product strategy. Are we looking at the right markets? Do we understand their needs? Do we have good go to market planning? All of the things up front that you really want to understand before you get into it, do we need to do a partnership? Those, those kind of fuzzy, squishy strategy things, all focused on the market. How are we getting to market and who's our market segment that we're going after? Product user fit. Sometimes people call product market fit product user fit. And those are two different things. Product market fit. I'm solving the problem for a market product user fit. I'm solving issues with how they get through their work. So that is pretty much what the PDLC process is fit for, is how do I improve the product user fit by understanding customers and discovery and then taking it and bringing it full circle to delivery and bringing that solution back to customers. That's product user fit. And a lot of people confuse that with product market fit. The third one is product execution fit, which kind of sits in the bucket. That's kind of getting some lift now in product management circles, which is product operations and trying to hire in people who can understand the processes and the systems in your organization in order to make better decisions faster. So I call them right market, Product market fit, Right problem product user fit, and right decisions, um, product execution fit. So those three are the things that I'm always evaluating these product management organizations when I'm helping companies out and say this is truly what product market fit is not. These other two things that you think could loosely be equated with this conversation. So now that we have a, uh, common definition, let me help you focus on this. And this is where AI is going to actually affect your product market fit or your product user fit or your product product execution fit in terms of making your own internal decisions much faster.

Speaker A: You started a, uh, business that focuses on this notion of being a fractional Chief Product Officer, which I think is a fascinating idea. And you specifically before the show, you're talking about your focus on working with private equity firms.

Speaker B: Mhm.

Speaker A: Give us a bit of insight into the dynamic there at the moment between the firms you're working with, um, both in terms of the, the actual innovators, the private equity firms, and what gap or problem, um, a fractional CPO can

Speaker B: help these people solve Two things. One is integrating prod, integrating AI into your product portfolio where it really is going to matter. Right. And bring some value to that business. I mean, they're all after value creation anyway. So if you're applying, if you're going to go through the motions of applying AI to your product and it doesn't actually move the needle on product user fit, you're not going to get the value out of that in terms of, uh, when it's time to sell that company. So I help them understand their customers and those business processes so that we can apply the technology of LLMs to the right place. The second thing is trying to integrate it into their decision matrix. So going back to product execution fit. How can we make better decisions? How can you get better systems by utilizing AI? How can we take this unstructured data, help you get to a point where you guys can make faster, quicker decisions within your own internal organization?

Speaker A: And talk to me about the private equity side of things because you gave me the impression before we started that there's a bit of, I don't know, there's a bit of pressure, um, or maybe I guess some expectations around results and return that's coming from that segment of the market.

Speaker B: Yeah, I don't know if you recall maybe a couple of months ago when cloud code work came out. I mean cloud, I Keep conflating the 2. Cloud code came out and not, not cloud cowork, cloud code came out. And the public market for SaaS businesses took a huge valuation hit because everybody was like, oh, SaaS is over. You can go in and you can create your own internal tools, you know. But what was missed, um, that got the market all jittery about this is that there's still a lot of benefit in these SaaS platforms that even with the internal usage of AI tools, you're not going to duplicate, you're not going to have the IT governance, you're not going to have all of those things by having disparate employees creating all these different tools utilizing this. And so really you have to kind of calm down clients and say you're still on, you're still on point. Now let's figure out again, how can we leverage AI to increase the value of your platform? Because your platform's not going away anytime soon unless you m happen to be somewhere around the harness or what do we call it, the event horizon, where you fall into the gravity well of a lot of these LLMs where you're just putting a wrapper around their LLM and you're Selling the solution, those will get gobbled up. But if you've got real IP and workflow in your uh, SaaS solution, and it is irreplaceable, and that's a key word that I like to talk about, is how replaceable is your product. Going back to that 80, 20 rule you're only doing if only 20% of your product is doing 80% of the lift, maybe we can do this internally.

Speaker A: Well, look, you know, it's an interesting concept. I, um, work with a number of, um, smaller businesses, advise a number of them and um, there's a bunch of work going on at the moment where everyone's kind of got a cop, you know, got their anthropic license and they're, you know, with Claude and they're, or their subscription, I should say, it's not a license, um, and they're burning through tokens, coming up with all sorts of interesting things. And what's interesting is I would agree with your hypothesis. And again, my sample size isn't massive, but what I'm seeing is people are saying, okay, so we've got a CRM or we've got a workflow tool, you know, we've got say something like a Monday or an Asana or we might have an industry specific ERP tool that we're working with that's part of our existing environment. And in the past people would have created workarounds, maybe done a bit of basic automation with workflow tools to try and fill the gaps or close the gaps between these systems. Um, but now what I'm finding is you've got a lot of these citizen developers who are popping up and they're getting an API key for that platform M and they're building their tool and augmenting the system. They're not replacing the system. And I would submit to you that that is further cementing the position of these platforms and it's going to make them so sticky that it's like, oh, we're thinking about replacing, I'll just pick on someone. We're thinking about replacing ServiceNow with some, some different tool or a homegrown tool and everyone's going to go, you can't do it. We've built all of these integrations around this thing. I uh, think we're actually looking at potentially hardcore lock in over the long, over the long term. So, yeah, and this isn't advice to anyone listening on what you should do about your investment portfolios, it's me actually thinking about like the mother of all technical debt. If I was Actually to express where my anxiety is coming from as I'm looking at these companies thinking if you build too much stuff around your core systems, it's going to be really hard to replace them if you ever need to.

Speaker B: It's like a bunch of micro ecosystems. And I think something that's kind of comparable is when you look at ERP financial systems. And so I spent almost four years in that world where a lot of tools, middleware tools and you know, bolt on tools, you know, they uh, they grow up around all these different ERPs to supplement with the ERPs actually can't do. This is on like a whole larger scale the way you're talking about it. Right. You know, we've got all these little microcosms of ecosystems of tools that have been like latched onto it. No, guess what? Paul left, you know, two years ago. And so this tool's just been running here. Nobody else knows how this tool runs. And we don't want the business interruption. Do not shut that down. Do not let's prop up servicenow whomever we need to to keep the business flowing.

Speaker A: It's gonna be really interesting. I mean, uh, as a product manager though, it also feels to me as though, you know, if I'm thinking about a product at the moment is make it so that these tools, whether it's Open Claw or whatever, uh, you know, have the right, expose the right interfaces so that they can interact with the gentic systems more effectively. Because it will, it will further lock your solution, your platform into, you know, organizations. Enterprise architectures agree with that too.

Speaker B: And also for them to have to climb up the ladder on the value too. So you know, yes, we want to lock in these systems, but I also my system has to create more value. Oh, of course.

Speaker A: I mean you can't be lock in without value. Forget about it. Right. People are just. That's inauthentic. No, I mean genuinely do good things. But make it so that rather than think, you know, I've got to make my product this sort of closed ecosystem. Correct, enable it so that it can interact in this broader, as uh, I say, this world of agentic plus, you know, um, citizen developer kind of concept and uh, if you're adding value, people are just going to keep using you more and more.

Speaker B: I agree with you 100. That also places a burden on the product management team to rethink uh, pricing and packaging as well, along with licensing as you said.

Speaker A: We haven't even talked about that. Yeah. How much is that coming up in conversation for you at the Moment.

Speaker B: It doesn't come up in conversations. I have to bring it up in conversations because it's a token.

Speaker A: Economics is the thing you just wrote about recently, right?

Speaker B: Yeah, yeah. And so it's not something, unless you have to pay that the ferryman on the back end, that it comes to the forefront. Otherwise you're just thinking about it and saying, let's pull this into the application and let's just go sell this. Let's just go to market and see how this roughly will work. And it's like, that's not how you do this. You've got to think about pricing and packaging up front, because now your licensing and everything else needs to follow behind it. So I usually end up bringing it to the forefront of the conversation with clients.

Speaker A: So, um, look, I mean, wide rating conversation. I'm just curious, what else is on your mind at the moment? What else haven't we covered? Do you think people should be thinking about?

Speaker B: I think we're hitting the sweet spot for where most people are. I mean, there's huge disruption going on right now in terms of the technology industry, you know, with layoffs and all these things. And I think it's more of a shifting of how these job descriptions are changing and what it is that we think we should be dealing with in the age of AI to actually speed us up. But, you know, how do we slow it down and actually utilize it and become more efficient with it, make better quality outcomes with it, instead of just, you know, kind of throwing the tools out there to the, to the masses within our organizations and saying, hey, go wild, Create something and come back with what you got. Um, I don't think that strategy actually works and is going to be as fruitful to most people as they think it will be. And so my thing is that I think people need to be a lot more contemplative and strategic with their adoption of AI, uh, and internal processes, and for sure, within their products.

Speaker A: You've got five minutes in an elevator with the CEO, he finds out that you're from the world of technology, and he or she just says to you, hey, Marlon, you know, what should I be doing with this AI stuff right now?

Speaker B: My first thing would say, how well do you understand your customer?

Speaker A: Back to the anthropology.

Speaker B: Yeah. How well do you understand you might not need to be doing anything?

Speaker A: You know what? I love that. I love that answer more than anything, because I do. Uh, you know, going back to our. The opening of today's podcast, um, it absolutely is a transformative piece of technology. Going back to the Internet 30 years ago. Um, you can't ignore it but, uh, equally, you know, expect, just setting an expectation that we want people to slap it onto everything without thinking about what the outcome is going to be, what, how it's going to benefit the user.

Speaker B: And let's keep this in mind. I already gave away. You know, my age, I still work on on Prem solutions. Uh, I still run into on premise solutions with my client base when I was ahead of product in the age of the Internet. They're, they're not, they haven't gone away, they're still there. And actually some of the tide has actually shifted a little bit more towards some on Prem solutions. Then we got mobile. Everybody was a mad rush to go out to mobile. Guess what we still have on Prem. We still have website, web applications and we have mobile. We have reactive interfaces that work with mobile. Guess what's going to happen with AI? All those three that I mentioned are going to still exist.

Speaker A: Do you remember the paperless office?

Speaker B: Yeah, I do remember that.

Speaker A: Ain't no paperless office I've ever seen in. I'm telling you, it's everywhere.

Speaker B: No, they all still exist. They all coexist. So I think people need to, you know, the hype machine is just off the chart with what's happening. But you know, for me, for my business, I have to calm people down. And there's always been that way for me, even as a head of product within the companies was. Okay, there's a firefight going on, huh? Let's slow this down and look at this in a simple way so that we can actually execute. And that's just really what needs to happen here with AI too.

Speaker A: Do you think there is. I mean, I want to touch on that a little longer. I may be leading the witness here but you know, Apple famously has kind of a last mover advantage in many of the spaces it goes into. Um, now you could argue that's a pro or a con. They're arguably at the moment quite behind on this AI, on their, on their own native AI capabilities, maybe the better way of putting it. But they've often added value whether, you know, they were, they were arguably late to the market on Internet, they were late to the market on, on MP3 players, they were late to the market on watches on phones. I mean none of those categories did they enter first.

Speaker B: Right?

Speaker A: But when they enter, it's thoughtful, um, it's user centric, you know, it's fundamentally a different kind of offering and it's informed by the mistakes that People have made that have gone before them. Is that a valid strategy for a product manager at the moment or do you need to hurry up and do something?

Speaker B: You just move from Apple, which is, uh, I think the nature of how that company works, to other product managers and other businesses. I do not think that they're strategy the markets applies to other businesses. I don't think that in any way, shape or form for one. I don't think that other businesses, as we talked about with discovery, put enough into discovery or the anthropology of understanding customers. So you can't take advantage of the second mover if you don't understand them, you know. So if there are some product management organizations out there that actually put the anthropology in, uh, then I would agree with you that other organizations could adopt the Apple strategy and implement it successfully. But most organizations aren't doing that. And that's what I have to actually help with a lot is, hey, you need to get outside your own four walls and understand what your customers are doing. It's great that we can do zoom, that we can be on and have a conversation. I can interview, you can interview me and I can interview customers and understand. Understand. But to your point, the anthropology is missing. You know, I think the Japanese term is gemba, where you go to where

Speaker A: the work is to go to the gimba. Yeah, yeah.

Speaker B: Go to where the work is to see it. And I got trained in lean manufacturing too. I actually grew up in a hard consumer good company that actually had a software piece in it. So I got exposed to that early on too. And I had to kind of shake off the process, the rigidity in the process. Once I got into software companies that were just all about software, but, but the thinking, there's some core principles in there that is still applicable, like gimba. Go to where the work is if you want to. If you want to understand it.

Speaker A: Doing your Hoshin plans, if that ever. Yeah, that resonates with you at all.

Speaker B: Yeah.

Speaker A: Uh, you know what makes me happy? I just, my heart smiled a little in as literally I was working with, um, the managing director of one of the companies I worked closely with just, uh, about six hours ago. And he said, uh, he said to me, yeah, um, because it's a phrase I use a lot with him. And he's like. He looked at me. Yeah, I went to the Gemba man. So there you go. Now it's all about that. I couldn't agree with you more. Hey, I want to wrap up with, uh, a, uh, question. One of the things I thought about when we were talking about this notion of CPO and a fractional CPO in particular is how can I have product managers and also bring in someone like a fractional cpo? Or is that people uh, going to feel like their treads being uh, you know, people that someone's treading on their turf. How does that dynamic work in a, in a modern environment?

Speaker B: And I can simplify it for you too from a fractional perspective. So it's a range of services. I may be an advisor, which I've had customers, or I was advising and coaching their product management team as well as their senior leadership on, you know, how they should look in value product management and to implement certain product management processes in their business all the way up to, hey, let's workshop. You've got a problem with your sprint management. You know, you guys are, you have story point, you know, overrun every sprint. Let's cinch that up and figure out what's happening there. So let's workshop, uh, this and get this process all set up for you.

Speaker A: So it augments rather than replaces. Sorry to interrupt you.

Speaker B: No, no, yeah, it augments and then all the way up to an interim CPO where they're looking for their next cpo, but they need somewhat somebody to keep the business moving, to keep product management operations moving. That one is arguably the most engaging from, you know, my time. I, you know, most of the time I'm much more on, lower on the ladder in terms of my commitment because interim cpl, you're giving them quite a bit of your time during the week, you know, to do that. So, so yeah, it's, it's a full range of services. It's not just naturally coming in at the top. And now all ah, the product managers are trying to figure out what's going on, you know, is my job in jeopardy? And quite the contrary, you know, uh, we probably were looking at this whole process all the way from just in an advisory capacity.

Speaker A: Look, I mean I would say again, um, you know, I think the role of enterprise architecture has never been more important. Data governance, obviously you can't make good decisions if your data's dirty. Yep. And ah, not well governed. And I would say the whole notion of product management, product design, um, that anthropology, uh, element of things in particular has never been more important. Um, uh, you know, and again, no disrespect to my software development friends, it's still an absolutely critical role.

Speaker B: Yes.

Speaker A: But there are a lot of these tools that are now available that will enable those three groups of people, you know, the data governance folks, the enterprise architects with the right services and the product management and, and design teams to come in and structure a solution, get it prototyped out to the point where you've proven value and then, um, the engineering team's job is to really make it more robust, uh, and scalable.

Speaker B: Yeah, there's still a lot for them to do on the back end there as far as the back end structures that need to be there with the product. These tools can't be to your point. I'm actually just adding to your point. I agree with everything you just said. We're not going to get to the point here, at least within this year, I don't think, unless something astronomically happens in terms of, uh, the improvement. But, um, of these tools for product managers to just go from, hey, I just, oh, I hate this term. I just vibe coded up some stuff and it is enterprise grade right out the box. That is not possible at the present moment. And so we still need our engineering buddies to, uh, architect that stuff. As you stated. I'm finding, along with the anthropology, just the handoffs now that we can do these labs, you know, to prototype a lot of these features, the handoffs are now a little kind of out of whack because these guys are utilizing AI, uh, coding tools in their sprints. Well, that puts a lot of pressure on product management to now get them the right specifications and to give it in more detail. Because you're not just now saying, okay, here's the acceptance criteria for how this feature ought to work. But now you're also, what's falling into the lap of the product managers is, okay, how is the responsiveness supposed to be there if AI is involved on the back end? If there's an LLM on the back end, what's the quality of the result that needs to come back? A good example of this is, let's say you were taking unstructured data and you said, hey, give me the top five bullet points out of this data for things that you recommend that I ought to do as a user. And it comes back with, you know, here's some five bullets and there's, there's context missing. There's, you know, you don't quite understand what it's telling you to go do. Well, now you have to understand how the prompts work. You under have to understand how skills work, you know, have some understanding of it in order to write your requirements so that the developer can go, okay, I should choose this temperature setting, I should choose this context window. Uh, you know, all of these things that they're going to have to do on the back end. Now, product managers need to be that much more detailed and rich in their descriptions to hand it over to not necessarily the. The engineer who's going to implement that, but the AI coding tool that's going to implement it first, and the engineer is going to review it. So it's kind of causing some upheaval here in terms of how these roles interact and how they work together, too. So. I agree with everything you said. I just wanted to add that whole.

Speaker A: Actually, I hadn't even thought about that insight. That's actually quite incredible. I mean, it's an interesting time. Um, interesting days ahead. We've only just begin. Begun to see the beginning of it. Really appreciate you taking the time to join us this morning from, um, uh, your neck of the woods this evening.

Speaker B: I am super honored to be here.

Speaker A: It's been great having you on. Thank you, Marlon, um, for joining us. I really do appreciate it. So, um, ladies and gentlemen, Marlon Davis, who's chief Product officer, uh, at Frugal Product Management. Is that right?

Speaker B: Manager.

Speaker A: Manager. You know what? I read manager, and I'm like, I must have got that wrong when I typed it in. I doubted myself. Uh, and frugal. Don't even ask me how to spell frugal, because if I've got product manager wrong, I've definitely going to get frugal. Uh, wrong. How do we spell frugal, Marlon?

Speaker B: F, R, O, O, G, E, L. There you go.

Speaker A: So it's not spelled correctly, which I. Which my spell checker struggle with my AI was going. You can't mean that. You must have misspelled it. You punked the AI right there, didn't you, mate?

Speaker B: On purpose? No, it's just a phonetic version of. It's frugal. It is the word frugal to be frugal as a product manager, but it is the phonetic. As close as I could get it. Because you couldn't do reverse E and get that to happen. You. You're even, uh, the, uh, URL locators don't work with that.

Speaker A: Yeah, exactly. Right. Oh, I love it. Um, uh, thank you so much for joining us.

Speaker B: Thank you. Thank you so much for having me. I really appreciate it. Thanks.

Speaker A: It's been a lot of fun. All right, well, ladies and gentlemen, that is a wrap. If you've got any questions or comments, um, for myself or Marlon, uh, or any one of the team, do hit us up in the comment section. We'd love to hear from you. You can of course hit up Cloudera, our amazing sponsor on LinkedIn using the handle Laderdera, if you've got any thoughts. Uh, as always, just a tremendous array of guests coming up. Some of them will probably also be from NASA. It seems like we've got a thing for NASA at the moment. If anyone is listening from any other space agency, we're more than welcome. We don't discriminate. So if you have enjoyed today's episode, make sure to like, share and subscribe so you don't miss another drop. Thank you again to the amazing folks at Cloudera for making this all possible or helping bring another episode AI broadcast to life. And thanks again for listening in.

Speaker B: We will see you all.