PRD – Panel – How To Build What AI Can’t: Differentiating in the Age of AI

Hannah Clark: [00:00:00] Hello everyone. We’ll just wait for a moment while everybody to get into the room. It gets settled. Hope everyone’s got a piping hot coffee or a very cold one, depending on your preference. Um, all right. We’ve got a packed show today, so we’re gonna get going. Uh, so welcome everybody to the latest in our community event series.

Uh, so every time we do these, they grow a little bit, they get a little more fun, they get a little bit more engaging. So thank you guys for coming with us on this journey for being here. Um, and we’re very happy that you were able to make time for us today. If you don’t know me, my name is Hannah Clark. I am the executive editor for the product manager and the host of the Product Manager podcast.

And today’s session we’ll be focusing on how we can build what AI can’t, how can we lean into our unique human strengths in order to, uh, really pop up our prod, our, our products and our teams and a increasingly generative world. Um, so we’ve got an awesome lineup of experts today who’s gonna, [00:01:00] who are gonna be speaking to a number of different topics that are connected to this main topic.

Um, so I’ll give them a little intro for you. First, we’ve got, uh, Margaret Ann Seger, or Ma as she’s often known to friends. Uh, ma if you wanna say hello.

Margaret-Ann Seger: Hello. Excited to be here.

Hannah Clark: So Ma leads product and design at Stat Sig, uh, which is a unified platform for, uh, powering experimentation, analytics, and session replay for fast moving product teams.

Uh, we recently did a podcast episode together, the team sounds amazing. Really recommend the episode for, uh, those interested. Uh, but ma was previously head of product at found in a product leader at Uber and Facebook, and she brings a lot of deep expertise into scaling data-driven SaaS products. Very excited to have you with us today.

Em,

Margaret-Ann Seger: um, thank you for having me. Be a good session.

Hannah Clark: Yeah. And then we also have Roman Maser, uh, joining us today. Uh, who leads Miros multi-device product strategy. Um, so Roman here is, uh, he’s shaping pro cross platform experi experiences used by millions at Miro. There. Um, he scaled products to 100 million plus users from mobile at Miro to music [00:02:00] tech, GIZ Smartt, and his crafted high impact fan experiences at one football.

Um, he is a top rated PM educator, and he thrives at the intersection of bold strategy and user insight. Roman, thank you so much for joining us. Where are you joining us from, again, RO Roman.

Roman Mazur: Hey, great to be here. I’m from Berlin, so yeah, and that was amazing intro. I need to record it somewhere. Thank

Hannah Clark: you.

Uh, well thank you for joining us and making time ’cause I know it’s quite a bit later for you than it is for probably most of us. Um, and then we’ve also got Thomas Stokes, who’s a returning guest today. Uh, we just liked him so much, we had to have him back. Uh, so Thomas, uh, he helps SaaS product leaders align research with business outcomes, uh, as the principal at Drill Bit Labs.

So he leads UX research to guide digital strategy and he also co-author his Depth, which is a newsletter that’s packed with tools and methods to help PMs drive product success through insight led decisions. He’s also a very, very smart person and, uh, just a wealth of in, uh, information and resources on ux, uh, research.

So, Thomas, glad to have you back, [00:03:00]

Thomas Stokes: Hannah. It’s definitely good to be back and I’m excited to, uh, talk a little bit more today about, you know. The age of AI and really, what the heck do we do during all of this? Right? So excited to get into that.

Hannah Clark: I feel like that is the theme of 2025. What the heck do we do?

Um, so yeah, very excited to get started in a moment. Uh, and so I’m gonna do, be doing a little bit of housekeeping before we get into the session. Uh, so while I’m doing that, if you guys, uh, in the chat wanna pipe up and, uh, let us know where you’re tuning in from. Um, I’m just gonna go through some better, uh, some boilerplate procedure.

So. First of all, this session is being recorded and it will be available shortly afterward. We may be using clips from it in our website and social channels. Your cameras and microphones are off by default, so you will not be appearing in the recording unless you are a panelist. Of course. Um, if you haven’t been to one of our live panels before, uh, basically how it works.

We’ve got some topics for our speakers to discuss. Um, we’ll be discussing them live. Uh, but uh, after about 40 minutes of this we’ll be going [00:04:00] to an open q and a. So you are invited and encouraged to give us your questions as they come to you. Uh, pop them into our q and a section and we’ll get to them towards the end of the conversation.

Um, and if there’s a timely question that we catch, Michael will be checking out the, uh, question list if we need to pop one in sooner. So we’ll try to get to them as soon as they are relevant. Um, alright, so just to set the scene a little bit, so like we said before, we are talking all about, uh, you know, this is an age.

Where more and more tools are available to help to accelerate product development cycles, to give folks who, uh, previously did not have the tools, all of the tools that they need in order to, uh, create an MVP and possibly even found their own product organization. Um, so that means that there’s a lot more opportunity and a lot more, um, competitive potential in the world, but that also means that there’s a lot of opportunity and competitive potential in the world.

And so we’re all kind of, uh, in a position where we have to find other ways to be competitive, given that we’re just gonna see a whole lot more fish in the sea. [00:05:00] Um, so that’s kind of the, the world that we’re living in. And we’re gonna start by talking about the role of storytelling and standing out in this world.

So, um, as generative AI enables faster development cycles and launches. Uh, how do we tell the story of our products, uh, our teams, our missions, and all of that? It’s just more critical than ever. Uh, so I’m gonna send this one off to ma to get us started here. How do you see the role of storytelling unfolding in the next year?

Or what are some great examples that you’ve seen in the space today?

Margaret-Ann Seger: Yeah, I think it’s, it’s a really good question. Um, you know, obviously the, the world is changing fast and furious, but people are still people at the end of the day and we want to associate ourselves with brands or products that we resonate with.

Um, and so, you know, given there’s just gonna be more and more products out there, creating that brand that breaks through or that story that really resonates with someone is gonna be more important than ever. Um, and so what’s interesting, or at least what I find interesting is if you look at all these AI companies out there, they actually are all kind of converging on a similar aesthetic.

I know this ’cause we did [00:06:00] a whole like, rebranding exercise as a company and we surveyed, you know, the linear and the versal and all these amazing product companies, um, all have a very similar kind of like super clean, sterile aesthetic. And that’s beautiful and cool and aspirational. But I also think like, you know, if you create that today, you’re just gonna be one of many, right?

So how do you stand out? Actually, in our experience, at least my experience at stat, um, you know, bringing people to the forefront and telling people first stories has actually really differentiated us. So, you know, a few examples to make that concrete. We actually record product, demo videos with, um, the engineer who built the feature.

So we have, uh, face and a name to put to the feature. It’s often very horrifying to the engineering question. They do not wanna be in front of the camera, but that authenticity and just like, I built this, I own this, I understand this problem. It really shines through. And I, I think people resonate with that.

Um, another example is we have our dogs on our website. We have a pet page, [00:07:00] uh, and we kind of give them fake roles and, you know, it just, it humanizes the company in a way that a lot of other companies today, you’re kind of hard pressed to find that human element. And so I think actually this is a strategic advantage for us.

Um, and it’s messy and it’s unpredictable, but it is, you know, highly resonant with folks, um, and feels relatable. The last thing I’ll add here is, you know, storytelling. Um. Often benefits from repetition, and if you can repeat something, uh, you can build awareness and recall and associate that with your brand.

Good example is we’ve done, um, this thing called Experimentation Week for a couple years in a row where we basically invented it. It’s a holiday at the company. We have t-shirts for it, uh, but we blast it on LinkedIn and have kind of a theme for each day and videos and, and takeaways for each day. And then we do a recap at the end of it, and it’s actually shockingly effective.

A lot of our customers have come back and be like, when’s the next experimentation week? Like, I’m gonna get my team on board. Um, and so you can almost create something and just do it repeatedly and will it into existence, and I think that [00:08:00] becomes a really powerful storytelling mechanism.

Hannah Clark: That’s super cool.

Um, and if any of the other panelists have some examples that kind of come to mind, I feel like this is just such an interesting, uh, topic to dive into.

Thomas Stokes: Yeah. I mean, I really like the way that you took that question because it, you took it like in a different direction than I initially was thinking because my very first thought went to.

Not just storytelling with our brand externally, but literally how we tell stories internally. Because one of the big things for me in terms of the way I think about getting people together and moving them towards some initiative, like in a way that everyone kind of understands, this is what we’re doing.

We’re all bought in, we’re ready to go, is that there’s like these two different phases that people go through. First you have to get people inspired and then reassure them, right? You have to get them to the point where they’re inspired to do something about it. And then once the nervous energy sets in, like, okay, we’re gonna do this thing, reassure them that like, yeah, we’re, we’re good to go, and storytelling is really [00:09:00] great.

And that first half where we get people on a same message about this is what we’re going to do, but it grounds in why we’re gonna do it. So I love the way that you talk about like brand wise, this is how we can leverage storytelling. But it’s also that great tool internally on how we come together. Then everybody gets really excited and then we say, okay, and.

We’ve got a plan. You let out the plan that caused the nervous energy and we’re ready to go. So I love that answer. And just wanna add on, like, there’s also all these internal benefits in terms of group cohesion, getting people together and getting them really jazzed about doing something as a product org.

Margaret-Ann Seger: I love that. Yeah. I think you can’t underestimate the power of internal storytelling. Um, I don’t know if folks have checked out ab ATI’s talk on this. She did a Lenny’s talk, um, on storytelling. It, it’s a very good one. So yes. Plus one for internal storytelling.

Hannah Clark: Love it. Well, yeah. I see so many examples of that, uh, outside of product world too.

And I think it’s like, it’s such a good reminder that [00:10:00] sometimes we really need to, a problem that is very clear, maybe to one of us isn’t necessarily as clear or the stakes aren’t quite as high to others unless we’ve done a really effective job at communicating and like what. That’s the piece that we all share that kind of helps us get on board.

It’s like something that we see present in our marketing efforts, but also in how we communicate and kind of get buy-in internally as well. Um, Roman, did you have anything to add before we move on?

Roman Mazur: Yeah, I think these are amazing points. When we look in the market, we also see that people are often purchase and buy the story behind the product.

They don’t care that much about the features or like how many AI agents it uses and so on. But the story, a really compelling one is something that distinguishes you on the market and I think this is such a valuable tool that some companies actually ritualized. So yeah, I think if we do it more in general then it will help us to build a better products.[00:11:00]

Hannah Clark: Well, thank you guys for those. This is a really engaging way to kick us off. Uh, moving on to our next question about, uh, talking a little bit bit more about users. So knowing users, it’s always been kind of a core priority for product teams. We’ve never really not cared about users, um, but now we really are taking it to another level.

So, uh, what are some of the most eff efficient tactics and strategies that you have seen for developing and leveraging feedback loops? This one is for Thomas.

Thomas Stokes: Yeah, I’ll, I’ll go over some of the big characteristics, the core characteristics of teams that I see do really well with this. ’cause one of the, where I’ve really spent my career is getting what I call insight loops, because I like it more than feedback loop feedback says like, we’re taking things people directly say and then doing them.

Insight is, or generating knowledge. So I like insight loops, but teams that use these insight loops really well have a few like core characteristics that I just wanna highlight and then I’m sure others. We’ll add onto that one [00:12:00] is that it’s truly a loop and it’s a habit. It’s not just like these onetime activities or something that is one step of a process that occasionally gets repeated.

It is really a muscle that these teams develop. With that, the second core characteristic I see is that teams that with really, really strong insight loops have well thought out and defined infrastructure for how they’re going to collect that. And that includes multiple data sources. It’s leveraging all the unstructured data that’s already out there, social listening, for example, going out and collecting data and primary methods, so that can be user researchers going out and conducting studies.

That can also be, of course, something like a, um, intercept survey on your website that starts to collect things around specifics, types of site visits. So we’ve got. Muscle infrastructure, multiple data sources. And then I guess the [00:13:00] fourth main real differentiator that I see teams that do really well with this do it across the product development lifecycle.

Uh, they do it during discovery to really drive what it is that they’re going to build when they’re actually building something. They’re constantly doing different concept tests, presenting it, using that as quick iteration feedback. And then finally around launches and even post-launch measurement, they’ve got a finger on the pulse.

They know exactly how people are reacting and how that’s changing over time. So in short, those are like kind of the four core things that I think most really, really strong user led or insight led orgs do. Um, in terms of specific stories, thoughts, reactions around that? Ma? Uh, Roman, not sure if you’ve got any that you’d like to add to that.

Margaret-Ann Seger: I’ve, I’ve never heard this framed as infrastructure before, and I actually really love that. Um, because sometimes I think orgs treat, you know, UXR [00:14:00] and, and deep customer understanding is almost like an add-on, a nice to have. Mm-hmm. And I just love saying, Hey, this is mission critical, this is infrastructure.

Um, just like our infra can’t go down, we’d have this, you know, critical infrastructure to understand how our users are engaging and what their pain points are constantly. Um, so that’s, that’s a really good framing.

Roman Mazur: Yeah. I really like the point about. Sorry, Thomas. Yeah, go ahead Roman.

Thomas Stokes: It’s all you.

Roman Mazur: Yeah.

Yeah. I really like the point about building the muscle and ba basically like research and insights becoming a part of our routine. So it’s not something that we do in once when we kick off the project and then never talk to customers again. I think it’s such an important and powerful tool, and if you repeat it on different stages of the product development, you actually gonna get super valuable insights on each of them.

So like, first you identify what to do, what to build, but then you understand how people use it, how they interact [00:15:00] with it, what is working, what is not working. So yeah, just integrating it, uh, into the development product life cycle is a huge, huge benefit for the teams.

Thomas Stokes: Yeah, I love that. And just like you said, the infrastructure and the habit, building the muscle, they’re one and the same, right?

Building the infrastructure allows us to quickly and continuously get different information, whether that’s discovery about what we should build, or insights about how people are actually using it. Having the infrastructure in place lets you actually do it. You turn on and off the whole time. You’re just kind of rebuilding and restarting a process rather than having a continual action.

Hannah Clark: Um, cool. We’ll, we’ll switch gears a little bit and talk a little bit about, uh, team structure and design. Uh, ’cause I wanna talk a little bit about, uh, just how the way that we structure our teams and how, uh, we work together can really impact and offer like some kind of a competitive edge. Um, so [00:16:00] this one is for Roman.

So what are some of the key opportunities that leaders should be considering when we’re deploying teams right now at this stage in the game?

Roman Mazur: Yeah, the thing that they’re gonna tell right now might be a shock, but AI actually doesn’t create amazing product people do, right? So, uh, it’s something that we utilize in order to create an amazing products that inspire people and help them to complete the task that they have at hand.

When we look in general and the stages of product development, I think where AI is the most helpful is actually build it. But the very initial steps of defining the product, defining the problem. These are the things where we can actually utilize the, um, you know, the power for the power and the magic of collaboration and working in the teams.

And this is where we, people, I think, are the best suited for this task because we can go out there, talk to customers, empathize towards customers, uh, understanding what’s going on, what’s the underlying problems and things like that. [00:17:00] So all in all, I think that, uh, the collaboration is an essential part and we need to build our processes around that.

So, uh, one test for your teams to understand if you’re doing it right or wrong. Probably would be to ask a few questions. Like, for example, the question, uh, around the roadmap. If your team built a roadmap and never talked to any other team, most likely you’re doing it wrong. Or like you designed the feature product manager designed the feature just, uh, themself and never talked to anyone else.

That’s also a signal of something is not working. The opposite example of that and how the things might be done right, is when you create the process which actually empowers teams and help them to work together, and it’s not. Enough to just say, Hey, team, go and be empowered. Right? You know, like it’s written in the books and in articles.

You need to create a process about that. You need sometimes to bump product teams [00:18:00] and go to market teams and, and maybe legal teams and maybe localization teams and things like that. So when they collaborate together, this creates a much better results, much better products that actually then, uh, appeal to the customers in, in, in a much more compelling way.

So if we want to take a few, I think, uh, things that we want to improve or to do today, or start and do today in our teams, I would say yes. It’s not super pen fancy, but. Try to create the process which nurtures and which empowers teams to collaborate together, not to work on silos, give them enough autonomy so they’re comfortable with making decisions, they’re comfortable with making mistakes.

This is also very important one, but ensure that we ask these questions about, you know, what’s going on on the marketing side, what’s your plan? Um, what, what languages [00:19:00] our customers talk and things like that. Asking the questions early on will create a truly amazing products and reinforce the collaboration among the teams and different functions.

Hannah Clark: Yeah. Oh, sorry, go ahead.

Margaret-Ann Seger: Well, I was gonna, I was gonna hop in, I think an interesting, um, point on collaboration. ’cause I think, you know, you mentioned silos. It’s very easy for teams to get siloed in their own little words, worlds. And I think remote work doesn’t help with that necessarily. Um, but. One thing we’ve done, which is a little chaotic, like high chaotic energy index.

But, uh, we kind of let teams form organically in a way. So we’ll do a lot of tiger teams or like war rooms. We’ll have like a very, you know, specific thing we wanna get done. And then we’ll say, who are the best people from across the organization to do this? And we’ll pull them together and do a war room on it with like, exit criteria.

Or, you know, recently we kicked off a Growth Tiger team to focus on some of our self-serve, um, activation blockers. [00:20:00] And these are all just people that are passionate about working on this. And they come from different teams and it’s actually great ’cause then you get folks working together who would never have previously collaborated and now they have a relationship.

And so when they go back to their respective teams, there’s like a little bit more connective tissue built, um, which is really cool. So, you know, you can do this with hackathons where people mix from different teams together. Like there’s just these ways to kind of organically stir the pot a bit that I think help.

Um. Just build company-wide bonds and, and reduce the risk of people getting super highlighted into their old, their own worlds.

Hannah Clark: That’s super cool. And I really love this idea of connective tissue. Sorry. It looks like Thomas might have something to interject there.

Thomas Stokes: Yeah. I ma I really love, like working in little pods.

I love setting up war rooms around specific initiatives. And one thing that I’ve observed, I was curious if you’ve encountered this and if you’ve been able to kind of rectify the issue. As a org grows, I’ve, in [00:21:00] my career many times, joined like a smaller company that then grows setting those up, have gotten more Miller difficult, each stop along the way.

I’m just like, how the heck do we, because there’s a lot of value in that kind of way that people come together, but then it seems like setting them up and making sure that they work well gets harder as the org grows. Um, not sure if anybody’s got tips around how to navigate those. Growing organizations that get a little bit harder to, in a very quick, very flexible way.

Stand up, something like that.

Margaret-Ann Seger: Yeah. It, it, it definitely is harder the bigger you get. Um, I think like you need to have leadership that really believes in them. Like they, it really needs to be something that’s valued at the top. Um, that’s like most things in, in company building, right? But if your leaders value it, they will create space for it.

They’ll suggest it. And then the other thing is like, speaking of space, I actually think physical space helps. So being able to have a conference room that is like, Hey, this, you know, is the physical war [00:22:00] room that people are gonna be working in. I remember, um, and this was a long time ago example, but we had, um.

A fraud issue in China when I was working on Uber’s international growth, and we set up a China fraud room, and these poor folks were like literally stationed in this room for I think, like two months, and they were there like 24 7. And you just kind of, people would swing by, they’d be like, how’s it going?

Can we bring you food? How’s morale? And like, it kind of became a company awareness thing, just having these people in this room that was like visible to the whole company. Um, so I think, I think little things like that you can do, and it, it helps reinforce the value to folks who maybe are new to this war room or tiger team type concept.

Roman Mazur: Yeah. I guess after a couple months in this war room, people became friends for their entire life.

Margaret-Ann Seger: They are actually, I, there was a reunion in SF a couple weeks ago with China fraud folks. So sit in a room with someone for two months and you’re, you’re best friends. Yeah.

Hannah Clark: That’s [00:23:00] really cool. Okay, so I kind of wanna bring together elements of this question in the last question.

’cause, you know, when we talk about teams and 20, 25 teams are not just human teams now, and I know we’re, no, we’re talking about what AI can’t build, but that includes how do we kind of fold in more human elements into the way that we are leveraging AI in our processes? So when we think about, like, um, using AI agents as thought partners or, or even teammates, um, I’m curious, and maybe this is a question for Thomas, maybe someone else wants to jump in, but how do you view customer empathy and context into, um, the AI agents that are kind of supporting the team?

Thomas Stokes: Yeah, I’ll, I’ll start with the fact that context is king, right? No system just automatically understands users out of the box, right? Just like no person that joins, uh, whether it’s like an organization or a smaller team within an organization immediately understands those users, right? So [00:24:00] the question becomes just like someone onboarding, how do we get them real user context and info?

How do we get them up to speed? And one thing that I’ve found pretty successful, um, over the past few months is using kind of what I call like closed systems rather than out of the box large models. So a poor example would be, alright, chat to bt, come up with, um, like personas for me that that’s generally not great.

It’s relatively superficial. It’s based on just like conjecture. What’s been actually pretty useful is using a system that you can actually power with that context, with the context of any information that you have. User research, analytics, wherever it comes from, even product support tickets. There’s a few, um, like custom tools that do this.

Like one of my friends has a startup that all about uploading your sales calls to see what users are asking for. I’ve [00:25:00] also used, um, notebook, lm, uh, where like we’re a Google workspace organization, so we’ve got notebook, lm, and after we do a bit of discovery, we always upload it into a workspace. And that tool actually is really good at passing what we call the motorcycle test.

So you ask it like, Hey, what do our users say about motorcycles? And we don’t do work for Harley Davidson, so it doesn’t, our users don’t say anything about motorcycles usually. And so it’s this great test that we do that tests like, okay, are you coming up with a random conjecture? Are you just pleasing us or.

Actually understand the customers, and it’s just using the information that we have, but can synthesize it and get people up to speed in a way that is so much faster than what we could have done before. So those types of closed systems have been really useful.

Hannah Clark: Awesome. Does anyone have any, um, [00:26:00] anything they wanted to jump in with before we move on?

All right. I guess we’ll just get on, uh, to shift to gears a little bit and talk a little bit about distribution, which is, uh, as a person with a, a marketing background, this is kind of a favorite topic of mind. Um, so now that we are kind of luck, we’re kind of walking into a different competitive environment here, we’re sort of, um, thinking about a time when there’s a gonna be slash is, uh, influx of new products, a lot of new ideas.

Um. Varying degrees of quality. But what we really wanna be focusing on, uh, right now is like, how do we really master distribution and really stand out? Uh, you know, we have a great product, but how do we get in front of the right users? Um, so I’m gonna, uh, throw this one to ma. What are some of the ways that we can strengthen our go to market strategies so we can position ourselves in front of the right people?

Margaret-Ann Seger: Sure. So. I think, um, I think of this as a multi-step type thing, right? And the [00:27:00] first step is not rocket science, but I think it’s important to call out so that it doesn’t get lost, um, in, you know, this kind of advanced new era. But you actually need to deeply understand your customer pain point and what messaging resonates with your customer.

And this cannot be underestimated. This is the thing that the LLMs can’t do as well. This is the real human art, because it is more art than science. Um, but building deep empathy for your customer, like truly understanding what makes them tick. You know, not just their opinion on motorcycles, but their opinion on 10 other different things that might influence their product decision.

Um, you know, I think that it’s also nuanced, right? It’s something might be your conversation opener, right? But not what actually seals the deal on buying your product specifically. So good example is we see, you know, talking about experimentation in the aspirational. Desire to experiment more as being a great conversation opener for us at Stat Zg, but that’s not what ultimately sells our product.

What sells it is that you get this like broader platform where everyone on the team can use data together, and once they get [00:28:00] in there and start playing with it, customers are like, oh, okay, I get it. Um, but if we were to try to describe that no one wants to buy a platform, so I think it’s the, the art of understanding the pain point and understanding the product you have, and then how to position that, um, relative to that pain point.

So that’s the first step. Second step is actually also probably quite hard for LLMs, but maybe more doable. Um, but this is kind of, you need to crack some clever distribution mechanism because this is ultimately what companies live or die by. Um, you know, if you have a great product but you don’t know how to get it up to the market and sell it, you’re, it’s not gonna go very far.

Um, and the barrier to entry to build great products is going down, but that barrier to entry to sell is actually going up. So it’s kind of an inverse relationship. Um, so, you know, couple examples that I think of when I think of great distribution. There’s, you know, of course the companies that have built-in advantages, right?

Like Microsoft teams. They kind of got a free, free distribution channel there. That’s awesome. Facebook can launch anything on their platform and [00:29:00] immediately reach multiple billions of users. You just have these like super powerful, strong built-in distribution channels, um, or celebrity led brand plays, right?

Like Skims the Kardashians have. Great brands. They have, you know, millions of Instagram followers. You could drop any product on the market and it would instantly fly off the shelves. Um, but then there’s folks that I think, you know, and these are the ones I, I aspire to, that I like really admire, who kind of think about more clever distribution strategies.

Um, I’ve been like weirdly obsessed with Wiz recently. Don’t ask why, uh, but they are just like wild. They have this totally crazy brand for the security space. Like if you look at security, it’s all like red, black, hardcore, you know, enterprise and Wizz is like fun and looks like, you know, a a B2C modern product.

Um, they had like a Wizard of Oz themed booth at security conferences. They like, they, they just have a very unconventional brand and um, strategy. And I think that’s, that’s just so different from the rest of the players in the market that they [00:30:00] stand out. Another one is Vanta. And I’m, I’m picking like two companies that are in very unsexy industries, but I don’t know if you’ve, uh, driven on the 1 0 1 in San Francisco into the city, but Vanta has had a billboard right at like the entrance to the city that says compliance that doesn’t sock too much, which is just like incredibly clever.

They’ve had that for years and it just, I, I, I bet they can’t even quantify how much brand value that has built for them. It’s the location, it’s the, you know, the cleverness of it. Um, you know, side note, we did billboards. Billboards are alive and well in 2025 apparently. Um, so some of these things are just, you kind of need to figure out what’s your angle and really go after that.

Uh, if you don’t have some of the built-in distribution that some of the other companies who have big established platforms have. Um, so that’s kind of how I break it down is like deeply understand the customer pain point and your messaging and how to position your product to resonate with that. And then what’s your clever distribution take?

I’m curious what the rest of the group thinks.

Hannah Clark: Yeah, me too. [00:31:00] That, those are really great examples.

Thomas Stokes: I’ve been thinking about brand a lot lately in terms of standing out to ma. I’m glad you’re talking about this. Um, my education was in cognitive psychology. I actually did my dissertation on how memorable something is based on how much it stands out from other things within a similar set. So like very close to my heart.

And I don’t know, I think there’s like this memorabilia aspect to a lot of the examples that you mentioned, but beyond just the fact that like people remember it, like that’s obviously important. It’s important to have brand awareness at some level. I memory doesn’t just happen by random processes.

Cognitively speaking, memory only happens through attention. Like attention is the mechanism through which we develop memories. [00:32:00] The attention that it actually brings you, that develops the memory, does a lot, uh, to actually get people to engage with what you’re putting out there. So it’s not just that they recall it later, they actually engage much more in terms of deliberate cognitive processing as well as implicit processing of the messages you put out there.

So it’s not just like, oh, it stands out. They actually really end up dedicating more thought, both deliberate like cognitive aware thoughts as well as implicit like kind of subconscious thoughts to that brand. So I’m glad that you brought that one up because it’s been something, it’s been a little bit of a passion project of mine for a while, but, um, I don’t know doing that in a way that’s very careful and isn’t off putting like.

That I think is the key because we can all say like, oh, let’s just be different and do this. But how do you do it in a way such that it’s really elegantly done, it’s different and a refreshing, [00:33:00] but not like, ooh, what’s going on here type of way. That seems to be the kicker.

Hannah Clark: Ah, that’s a good observation.

I’ve definitely seen some off-putting examples where it’s like, ah, that’s, that’s getting attention. That isn’t the right kind of attention. Um, but yeah. Uh, did, uh, uh, Roman, I’m not sure if you had anything to add to that or, uh, Margaret in if you wanted to respond.

Margaret-Ann Seger: I just wanted to say, I, I’ve like so impressed that you have that background and now I can understand why these brands that are different resonate.

Like, it’s cool to understand the mechanics in your brain behind that.

Thomas Stokes: The brain is like the coolest thing to study how people like take in and process this information, what’ll capture their attention and how it all just flows. So I’m glad you agree.

Hannah Clark: Right. It’s also’s also. Interesting

Roman Mazur: how, sorry, sorry. It’s also interesting how, so we, we were talking about different, differentiating the own market.

So this amazing examples of the brands which are [00:34:00] from a quite boring industries like securities and so, but how much different do you want be actually like where is this sweet spot between too much and people just, you know, entertain with your ads and commercial and so on, but they don’t really pay attention to the product Or like, and between the being boring and like why as, as everyone else.

So it’s interesting how to strike this balance and how many iterations you need to do for that. Maybe anyone has a bad example of how the things might go, you know south when a brand tries to different, differentiate too much and goes into the woods and lost somewhere there.

Hannah Clark: Uh, I will say, uh, because this is something that I’ve observed a lot, uh, working, uh, like I, I early in my career was in marketing agencies and we, uh, would often pick up projects from yeah, very dry industries.

Um, I don’t wanna name any in case anyone here works for them and will be offended. Uh, but one thing that I [00:35:00] saw frequently is that, um, I kind of, to your point, Roman, there is kind of a nuance there of showing that you are a serious player, um, especially in something like securities. Um, but also keeping in mind that everyone, even the people who work there aren’t inherently boring people.

We all still need and, and kind of, kind of operate on the same mechanisms that Thomas was describing. Um, so it’s kind of like, you know, how do you just kinda leverage some of those kind of signals that. Humans find, uh, that kind of trigger attention in memory and retention and, uh, allow people to digest information.

I think that this is really, I relevant to things like, uh, landing pages around product features and that kind of thing because how often have you seen a product feature page where it’s just loaded with meaningless, um, you know, buzzwordy kind of language that really obscures and sometimes completely makes it for an opaque [00:36:00] message around what the product actually does and what value it actually brings.

Um, a lot of the time I feel like it’s just a lot of, uh. There’s a lot of smart words that really, um, make for an unclear message. So I think there’s like a, a lot of beauty and clarity, uh, and the use of color. I, I mean this is all just UX design 1 0 1, but yeah, I think, I think kind of thinking about those things is important.

But to kind of, the other thing that you’d mentioned there, Roman, that I thought was kind of interesting, it’s kind of an example again, a little bit off the cuff, very random, but there is, um, a chain of hotels in my, uh, area that has been doing these video series in which their employees dress up, like the characters from the Disney’s frozen movies.

They’re very unhinged. Uh, but this is kind of the thing where it’s like they’ve gotten this viral attention, but it is doing nothing for the brand. And I kind of begs the question, you know, when you’re kind of looking at it from the perspective of marketing and branding and that kind of thing, [00:37:00] it’s like, what was, if the in, if the intention here was just to, um, get a lot of eyeballs on this brand, it has succeeded.

But to the that point, you know, that’s not necessarily a KPI for success because it doesn’t really speak to the pain point. Like the people who are booking these hotels aren’t trying to book birthday parties there. They’re trying to, you know, do business travel and they’ve got other jobs to be done. So it’s like the, if does the play align with the, the market that you’re trying to reach?

Does the, uh, attention that you’re getting align with the, the, you know, psychographics of people that you’re trying to, to really connect with? Anyway, just some kind of scattered thoughts on that topic. Um,

Roman Mazur: yeah,

Hannah Clark: I, what’s that?

Roman Mazur: No, no, it’s amazing point. You know, you need to know your customer, right? Mm-hmm.

If it’s frozen fans, which is totally fine, then let go this route.

Hannah Clark: Yeah, I, I should post the video somewhere. It’s pretty insane. But, [00:38:00] um, anyway, we should move along as we’re getting kind of close to, uh, q and a time. Uh, I do wanna quickly address another question here that we had on the list, which I think is really interesting, which is, um, the evolution of Uxr in the age of ai.

Uh, Thomas, I know you’ve got some thoughts on this matter. Um, I think there’s a lot of controversy in this space as well as people kind of try to use generative AI tools in order to conduct research. There’s a lot of. Differences of an opinion about the, uh, the efficacy of those and the accuracy. Uh, but yeah, like how, in your view, you know, you’re working very close to this field, um, what do you kind of see as some of the major developments and, uh, even risks that are happening in this space?

Thomas Stokes: Yeah. When it comes to using AI to do user research, we’re very early days. Still. A lot of it’s being driven by like individuals tinkering with things and testing out different stuff, and then sharing it with their community, and then someone else tries, it completely disagrees. So we’re in this kind of like free [00:39:00] shakeout phase.

It’s not really, like, I can say a whole lot of like, we’ve really learned this. I, I can’t. But I can tell you that one thing on the other end of actually doing u like UX and user research on ai, there’s this workflow change that’s happened where, you know, in a traditional user-centered process. You would start with the user need.

And we’ve talked about it all throughout today’s session, right? It’s so important to start with it. And then you develop solutions based off the user need. However, now these days a lot of teams are being told the solution you’re going to start using is artificial intelligence. And so people are starting with a solution then trying to find the user need and the actual user problem that’s gonna solve, which is a very different way of working.

Um, and not quite what you would typically see in a very user-centered process. Um, like I think Google recently said to all their employees, any of their [00:40:00] experiences with more than, what is it? A million active users will have AI features integrated in by the end of the year or something like that.

You’ll have to check me on like the accuracy of that. I think the numbers might switch around, but generally people are being told, you will have part of your work be looking at how you’re gonna implement AI into your product. And so. That presents a huge challenge for UX researchers who are more used to discovering the need than working towards a solution.

And what we’ve been doing at Drillbit Labs, we call the matchmaking framework, it’s kind of a fun way of solving this new problem that we’re all encountering. And it’s exactly like playing matchmaker. So we start with like on one end, step one, what’s AI good at? We identify a list of just general capabilities, things that it seems to be pretty good at doing.

The second we get our users jobs to be done, so we just line up like, okay, what are our users trying to do? The third step, it’s basically [00:41:00] speed dating. You take the things that AI’s good at, you take the jobs be done and you see like, oh, how are these playing together? Do they get along? Is there maybe a match here?

And then the fourth step is, uh, kind of, you know, you’re the matchmaker at the end of the speed dating event. You rate the different matches that can be on things like, uh, frequency, how often is a user trying to do this thing? And will AI solve some sort of like general workload issue. You can do things like obviously scoring implementation difficulty or just the world value that pairing AI to this job to be done with bring.

So it’s this fun process of going through and identifying really quickly a bunch of different ways that AI can solve users, jobs be done, and then kind of just scoring it. It’s a fun workshop session. I think a lot of people get into the idea of like matchmaking like that, but it’s a very fun look at a problem that a lot of people, I think have these days.

Um, anyway, that, that I think is the biggest difference right now, is just a lot of people are being asked to [00:42:00] find a problem for a solution that’s being prescribed to them.

Hannah Clark: Yeah, that’s, that’s a big one. And I feel like that’s, uh, historically been kind of a, an area that we don’t wanna find ourselves in, in products.

So it’s interesting to kind of find a way, well, you know, this is an area that there’s a lot of opportunity in, so how do we kind of take advantage of that the best we can? Um, well, thank you for sharing that, Thomas. Uh, I don’t know if anyone had any questions about that, uh, the framework or if there was any other things to add before we move on to, uh, q and a.

Margaret-Ann Seger: I’ll just, I’ll, I’ll share that I struggle with this one a little bit because I agree generally that you should never start with a solution, but I also feel like sometimes my team’s a little resistant. To trying to think about how to incorporate AI and I have to seed it and kind of nudge and push a little bit there.

Um, and so I think it’s like a very fine line of you still wanna get [00:43:00] plant that seed and have folks remember that this is a tool in their toolkit, and that like, we are building in a different product building era without being prescriptive about use ai, like incorporate ai. Um, so it’s, I think it’s nuanced.

Hannah Clark: Mm-hmm. Yeah. It’s, it’s nuanced and I think it’s also, you know, there’s a case to be made that building a skillset now and kind of learning how to incorporate AI in ways that right now seem a little bit counterintuitive, builds that muscle that you can iterate on in future products and future features.

Um, like I’ve seen before, especially early on when, uh, AI features became popular, it seemed like, oh, I don’t really know that they’ve really found a good use case for this yet. But then over time you can see those same products kind of, um, maturing their strategy and the ai, uh, works better. It makes more sense, and they’ve obviously gotten a better grasp of how to kind of leverage the technology.

So I think that there is a kind of a case to be made of like kind of, um, doing a crappy first draft, so to [00:44:00] speak, and kind of like learning how to use the technology better and refining on those things. Um, so yeah, I appreciate you adding that in me. Uh, any, any, uh, other remarks on Uxr evolution in the age of ai before we move on to q and a?

Okay, well we are about 15 minutes out from the end of the session. So before we get into the questions, first of all, a reminder to submit your questions because, uh, we love to see them, we love to answer them. Um, there’s a q and a blocks there. You can ask your questions anonymously or you can feel free to pace them into the chat there.

Um, so get thinking about those. But in the meantime, um, I know that there’s also probably a lot of folks who might be peeling away to their next meeting and, and et cetera. So I did wanna take a moment to just thank you for joining today if you weren’t able to stay over the whole session. Um, and if you’re enjoying this session.

We would also love to see you at our next event, which is a workshop, uh, with Dr. Nancy Lee will be next week or not next week, excuse me, next month. And that one was going to be a workshop. It’s [00:45:00] more of a HandsOn approach. Um, she’ll be walking us through her framework for creating and testing AI product hypotheses.

Uh, so registration’s gonna be open for that on our website soon. So you can SVP using the link that we posted in the chat. Um, we’ll send more updates as we get closer to the event. Um, and of course, subscribe to our newsletter as well for more information on what’s going on in the event space here at the product manager.

Um, and before anyone takes off today, we also want you to know that we absolutely love feedback. Uh, we love the insight loop. Thank you, Thomas, of, uh, being able to collect more, uh, your reactions, comments, uh, constructive feedback. We love to hear it all. So please, if you have a moment, uh, to let us know what you thought about today’s event, if there’s anything you’d like to see, uh, in the future or any ways that we can improve, uh, the experience, please let us know.

Um, be vicious. We can take it. Really love to hear it. Um, okay, so we’ll get to, oh, and one last thing too. Uh, you can also follow all of our speakers today on LinkedIn. Um, so Roman, [00:46:00] Margaret and and Thomas are all on LinkedIn. Um, uh, yes, Michael has just got their, uh, URLs there for you if you’re interested in following their work from here on in.

Anyway, back to q and a. So we’ll start with, um, some that we were submitted in advance. Uh, so one is what are some AI tools and software that you can re recommend that you use in your daily work? In other words, how do you use AI to amplify your own human inputs? Um, does anyone, and this one, sorry, it was submitted by Benjamin.

Does anyone wanna jump on that question?

Thomas Stokes: I’ll give two that basically anybody can use. Uh, I mentioned Notebook, LM earlier. If your organization that you work with uses like Gmail, Google Drive, all that, you probably have notebook lm It’s actually really, really great at producing briefing docs. So if you’ve got a lot of information that you’ve gotta go through, you can dump it in there, produce a briefing doc really quickly, and [00:47:00] I find that gets you up to speed pretty well also.

I’ve been into fathom these days in terms of sales calls. Uh, we add our fathom note taker to it, and it does a pretty good job of actually taking our calls and putting it straight into a format that we use for our CRM. So, uh, those are two that I throw out there that, uh, have been pretty useful for the past few months for.

Roman Mazur: I’m very biased here ’cause Miro has quite advanced AI capabilities and this is such an amazing, um, thing that you can actually use your canvas as a prompt. So imagine you, you use your canvas as like a print dump. You have so many things there and you can actually feed it into AI model and have a reasonable answers or, you know, like start with a, solve this problem of plant page and so on.

But other than that, I, I do really appreciate, uh, course or for prototyping, just making [00:48:00] a quick clickable prototypes, iOS, android apps that you can then distribute and test your ideas early on. So that’s something that I found really helpful.

Margaret-Ann Seger: Two, that we use or that I kind of engage with day to day. Um, one is, uh, we actually. Have built kind of a, we, we call it Stat bot. Um, but it is like leveraging an AI startup called Scout to basically build, um, a customer support triage bot so that when customer support questions come in, they automatically triage to, uh, the relevant teams feedback channels, and then the on-call that day gets ping looped in and reminded at a certain cadence if you’re not following up.

Um, so it just manages all the day-to-day workflow for all of our. Uh, just customer support inbound, which is super helpful ’cause we as product teams feel that. Um, the, the other thing that’s interesting about that is we’ve actually started impro improving it and optimizing it using Statsy. So we built kind of an offline, um, [00:49:00] model eval feature where we can kind of take a model and a prompt and a set of parameters and test a couple different versions against like a set of eval questions and how it will perform on those evals.

Um, and so we’ve been testing stat bot and improving stat bot to make sure that like when we, you know, add a prompt change to catch a corner case, we’re not inadvertently regressing something else in the model. Um, and so that’s been really helpful. That actually is in production for us. Um, and it also bonus helps us dog food, which is nice.

Um, the other tool we use is actually Devon ai. Um, so this is something that looks at our docs and our code base and can help us understand, not just, um, ’cause our, our stat bot is trained on our docs. It. Is usually up to date, but often our docs are not up to date. And so it can look at the code base and say, okay, the docs say this, but the code is actually written this way.

You may wanna reconcile this. Um, and as PMs, we’ve been testing basically having it consume our product updates and automatically, um, update our docs [00:50:00] accordingly, uh, based on what’s written in the code and what the product updates say. So we’re just kind of trying to find ways to automate these like little, you know, paper cutty time intensive processes that as PMs we do a lot of, but are not particularly specialized or high value.

Hannah Clark: Well, I did also wanted to, uh, add to this question as well. Um, a big, uh, turning point for me in terms of using AI for my work, uh, actually came as a result of a conversation that we had on the podcast with Tall iv. Um, he did a, we did an episode together on creating your AI copilot. Um, it was directed at product managers, but I feel like the, um, the advice that he gave in the show was so applicable to just about any.

Job. Um, and, you know, whether you’re a product leader or a senior product manager or even, you know, if it’s your, you’re very green to the job. Um, very, very flexible advice on kind of training. Uh, your kind of favorite L-L-M-I-I use Claude, but there’s a whole lot you can [00:51:00] do depending on your, your favorite, you know, interface, um, to just kind of develop, uh, a very personalized experience that can kind of support and understand, uh, your specific KPIs, your specific workflows, uh, and just kind of ways to integrate, I guess, you know, AI agents at the most pure, uh, form.

And, uh, it’s, it’s really been useful to kind of have that additional firepower, um, especially for more templated, um, more consistent work. But, uh, what I found really fascinating about some of his advice was being able to use it to, so, so to help navigate, uh, stakeholder dynamics as well. And by kind of integrating, uh, more context around.

Your, uh, work environment, who you’re working for, who you work with, their kind of idiosyncrasies, being able to kind of input some of the things, um, that you notice are important within your team, and that the, uh, the agent will start to make recommendations according to, you know, oh, you know, this person has a tendency to look at this, so make [00:52:00] sure that you are, you know, double checking this or that thing, or that you’re updating this in advance.

You know, it, it’s just, uh, it’s been really helpful for kind of, um, bringing all of these pieces of work together. So I, uh, we’ll post the link to that, that podcast there. ’cause I found it’s been really helpful in a number of different areas of, of my work, uh, for the past six, eight months, so. Very cool. Um, so this question’s from Debbie.

She, she admit that it might be, uh, oh, sorry. This is, sorry, I’m, I’m just looking at the, what’s written here. Okay. Is there some examples of how AI helps with infrastructure projects? This okay. This might be a broad question. Does anyone wanna take that one on?

Margaret-Ann Seger: I mean, this is not a detailed answer, but I do think there’s a whole, uh, world of like causal inference that will eventually get much better [00:53:00] with more AI advancements. So understanding like if something’s breaking or down or an alert is firing, what’s causing that? Um, you know, what is the service in question that’s down?

Where is the upstream outage? Um, helping you like quickly pinpoint and isolate and debug things should be much more of a kind of like automated, uh, process In a world where AI is really good and trained on all your previous outages and all your sev reports and all the like context of how your system works, um, at a broader level.

Thomas Stokes: Well done, ma. I’ve got no notes.

Margaret-Ann Seger: Well, that, that’s actually something worth thinking about, but it’s a very hard problem, so I, I want it solved, but, uh, yeah, it’s, it’s tough. I think a lot of people are thinking about that.

Hannah Clark: Um, alright, well we’re running short on time, so I’ll, I’ll move right along. Uh, how do you decide what not to build when AI makes shipping features so cheap and easy? I love this question because I often have kind [00:54:00] of high conversations around, you know, is, is just beca, excuse me, is just because you can, uh, the reason that you should, uh, and I think there’s a lot of nuance around how do we make those kinds of decisions.

I feel like maybe this is what good one for Thomas, but anyone, uh, is welcome to jump in on this one.

Thomas Stokes: Yeah. For so long it was like shipping was hard and time consuming, so. There was even more emphasis on like making sure you’re building the right thing. But now, because it’s so easy, it seems like the emphasis on building the right thing is less, which shouldn’t be the case because you, you can still really damage your product by shipping the wrong things and it’s actually really costly to do that.

Just because it’s easy to ship doesn’t mean we should, and like there’s a few stage gates if we wanna think about it like that. Obviously we don’t want to just ship something as we can. It has to meet a [00:55:00] few, pass a few stages. One gate guardian would be user value. Then there’s business value. Then obviously if we can ship it feasibilities there.

But is it at the very last kind of gate, like is this even something that we really feel confident in, like us as people, do we feel good about this outcome potentially of watching this thing? I don’t know. It just seems like there should be these stage gates of like. Things that we should hit still, even though it’s so easy to just push something out there.

And I don’t know, I, in a weird way, I think it should make people even more like thoughtful about that process. ’cause it is so easy. But you’re right, sometimes that doesn’t seem like what’s happening right now.

Hannah Clark: Hmm. Yeah. I think in a kind of a related challenge is, um, when there’s pressure from higher leadership to, you know, build things quickly in ship and kind of lean on AI for faster cycles [00:56:00] and higher volume, um, how do you kind of find your voice to kind of push back if you don’t think it’s the right thing for the user?

So that’s maybe a whole other event unless someone has any thoughts on that. Um, all right, let’s move along so we can kind of conserve. I want to shove as many of these as possible into the last, uh, few minutes. Uh, how do you differentiate when your competitors are also trying to tell compelling human stories?

I, I’ll let you guys go ahead, but I have thoughts on this.

Margaret-Ann Seger: Wait, let’s hear your thoughts. You can’t tease the thoughts and then not tell us.

Hannah Clark: Well, I’m not the panelist, but, uh, I, okay. I, I personally think that there is nothing, there’s no finite limit on how many human stories we can consume and appreciate in the same way that, you know, the, the market for music we’re maybe competing for attention, but human stories are unique, um, and compelling and connect with people for a reason.

Like we, [00:57:00] I don’t think that we ever get to a point when we’re out in the world where we’re like, you know what? I’ve met enough people. I’ve heard enough stories. I’m closed for business. I’m not open to receiving more. I’m not open to connecting with more. I’m not open to having my worldview broadened.

Like I think that as human beings, we’re always looking for those things and always able to connect and find inspiration, um, and value in those things. So I don’t think that we have to worry about, um, competing for the most human story. I think that we just have to be focused on what’s authentic and, uh, I think if we lead with that, then the right people will find us.

But maybe that’s a very optimistic worldview to have. I don’t know if anyone has thoughts.

Margaret-Ann Seger: I, I think that’s a great take. I’m glad, I’m glad we asked. But I think the key word was authentic. Like, people can smell when something’s not authentic from [00:58:00] a mile away. Um. So I think you really need to, like, if you’re gonna lean into the human aspect, it needs to be authentic. There’s no other way.

Hannah Clark: Yeah.

And I think that people, yeah, I, I totally agree with you, ee. I think that people really can tell, no matter how well we kind of couch, uh, like purely gen AI content and like, you know, make it sound more human, make it sound more people can tell if it’s not. Uh, and it has a lot to do with your intonation.

It has to do with, um, knowing a problem very in, uh, inwardly, like just being able to communicate the feelings adequately of, of experiencing an issue yourself. Um, and knowing the, you know, all of the context around what the relief to that problem looks like and feels like. Um. Yeah, I think that those kinds of things seem very, um, obvious to those of us who kind of steeped in those, those day-to-day realities.

But being able to tell that effectively and, and tell it in a way that, [00:59:00] uh, really speaks to your own passion about those things is it’s really the differentiator that I think rises above.

Um, we are at time. I’m sorry, I’m, I didn’t mean to eat up so much, uh, space on that little, uh, soapbox there, but I just wanted to thank, uh, our panelist, uh, Roman Ma Thomas. Thank you guys so much for volunteering your time to be with us today and share with us value, uh, everybody who attended this session.

Thank you so much for being here. Michael, uh, our community producer, thank you so much again for orchestrating this event. Um, we will have the event available as a recording and subsequently a podcast on the Product Manager podcast in a few weeks. Um, thank you guys all for being here. Please again, uh, make sure to fill out your feedback form if you have anything that you’d like to add or any comments or questions or anything for us or our panelists.

And I’d like to wish all of you a very wonderful day and a great week.