AI didn't break your company. It just exposed it. | Hugo Froes

Is AI fixing your company or just exposing what was already broken?

Discover how AI is accelerating organizational dysfunction, reshaping product teams and redefining what it means to build consciously in this conversation with product operations leader Hugo Froes.

In this episode of Human x Intelligent, Madalena Costa sits down with Hugo Froes, product operations and transformation expert formerly of OLX and Farfetch, to explore one of the most uncomfortable truths in tech right now: AI isn't creating your organization's problems. It's just making them impossible to ignore.

From broken processes to bloated team structures, Hugo shares a ground-level perspective on what's actually happening inside product organizations and what feels genuinely different and dangerous about this moment.
If you're a product designer, product manager, founder, or team lead navigating AI pressure from leadership, Part I will help you understand what's really going wrong and why the urgency you're feeling is part of the problem.

In this episode, you'll learn:
- Why AI accelerates dysfunction instead of fixing it
- Why 'AI-first' is the wrong question for most organizations
- The real cost of reducing product teams to a minimum
- Why the pressure to adopt AI is functioning as a dark pattern
- How AI is creating silos inside product teams
- Why 80% of AI initiatives are failing and what they have in common
- The hidden cost of AI at enterprise scale
- Why conscious adoption beats fast adoption every time
- Why this matters

As AI accelerates execution, the real differentiator shifts toward:
- organizational clarity
- conscious decision making
- depth over speed
- systems thinking
- human judgment

Key ideas explored:
- AI as an accelerant: AI doesn't fix broken processes; it exposes them faster and at greater scale
- The urgency trap: the pressure to go AI-first is itself a pattern worth scrutinizing
- The cost reality: most teams have no real notion of what AI costs at enterprise scale
- Quality collapse: as shipping gets easier, the percentage of truly valuable products may actually shrink
- Conscious adoption: the organizations winning with AI are the ones being deliberate, not reactive

Links
Website: humanxintelligent.com
Join the conversation: https://forms.gle/qdnd3pMnr6KBDCA1A
LinkedIn: @hugofroes
Instagram: @thehugofroes
LinkedIn: @human-x-intelligent
Instagram: @humanxintelligent
LinkedIn: @madalenafigueirasdacosta
Instagram: @designwithmaddie

// Human x Intelligent explores how humans and AI design, build and collaborate in intelligent systems //

Subscribe for more on:
- AI product design
- organizational design
- human-AI collaboration
- future of work
- agentic systems
- product leadership
- UX and systems thinking

Transcript

SPEAKER_00 0:04

Over the past decade, he has led product operations and transformation initiatives in companies like OL Shees and Farfetch, focusing on diagnosing the structural bottlenecks and prevent product teams from executing effectively. Welcome to Humanex Intelligence.

SPEAKER_02 0:20

Thank you very much for having me.

SPEAKER_00 0:23

So right now it feels like every company in the world is declaring like itself like AI first, AI first, AI first, VR AI first. Please VCs give us money. So bars are asking for AI strategies as well. Like product teams are experienced a lot with tools, like experience experiment. And leadership is under a lot of pressure for sure to show that they're not falling behind. But from the work you do to helping organizations with design how their product teams operate, the challenges might not actually be technological, as we also talked in the first time that we talked. So let me start with a big question to set the scenario for this ecosystem that we'll be creating today. Is the real challenge with artificial intelligence actually technology, or is it something much deeper inside how organizations are structured and make decisions?

SPEAKER_01 1:14

The simple question is yes, it is about the organizations, right? I think what happens is the problem. So AI is getting a shortcut, right? AI is giving a shortcut to everything, right? So I I think it's a mix of the two things, right? So obviously the adoption of technology, it's hard. It's a very complex thing. And we what we have is there is a challenge, which is we're not building up the capabilities that people have in dealing with technology, right? So we assume that people have to learn how to play with technology in their free time. When they do it, you know, you go figure it out. And by the way, you have to be an adopter of AI, right? The other aspect is the organizations, what happens is what we're doing is we're exactly exasperating problems that exist, right? So before we used to add a layer of like a solution layer, right, above a problem. And it's like, oh, it's kind of working, but then there's another problem. Let's add another layer. Let's add and we add those layers, right? What happens now is it's suddenly you go through 50 layers in a couple of days, right? And you just keep adding. So the problem is if your processes are broken, if your organization is not working well, if you haven't structured the way you think about how you develop problems, AI is not going to solve that for you, right? Also, reducing your teams. A lot of people are reducing their teams. Oh, I only need one. For example, I I loved someone shared this on LinkedIn a couple of days ago, which was you only need one product manager and maybe an engineer, right? You look and you say, okay, great. What if one of them goes on vacation or one of them is on sick leave? Everything falls on the other person, right? It's like all context goes out the window if suddenly that team can't be working for that day, right? So and and also there's the whole context that you build around, you know, the humans, the customer, the the you know, the product we're trying to build and all that stuff, all this intentional way of working. And AI does not solve that, unfortunately, right? And unfortunately, AI is creating silos, it's creating things. Think it's easy to build something like I remember when I was starting off as a designer and I used to have a manager who used to come up to me and say, Hugo, I need you to do this. It's just five minutes. It's like, okay, I can do that one. Then he comes to me a couple of days later and he says, Oh, Hugo, I just need this. And it's like, it's an extra five minutes. And you're like, that's 10 minutes already, right? And it's like, oh, and here's another one. And one day I turn around and I said, you know what, if I accumulate a lot of five minutes, it does come out to a full week.

SPEAKER_02 3:32

Right.

SPEAKER_01 3:33

And I think it's the same thing now, where they think, oh, no, it's super easy. So let's just throw and you know, this I'm going around in in a lot of aspects, but I think the main takeaway here is that AI just accelerates the problems, and we have to be very conscious and take a step and consciously look at how we're adopting it and how we're using it as best as possible, right?

SPEAKER_00 3:52

100%. I saw the other day a video on YouTube that I thought it was very interesting because it says now companies are realizing that they actually need people to work in their companies, so they're back again, which is very nice, very interesting because at the time, specifically last year, which was when it this started to grow a lot and people started to get laid off, laid off, everyone was saying, like, you're gonna regret it, you're gonna regret it. And the company was like, No, we are not, no, we are not, and now they're actually regretting it a lot. And sometimes the pressure comes from leadership. I think most of the times it comes from leadership that we need to implement. Even if now it's just like you were saying, a product manager and an engineer, there still comes a lot of pressure for them. And other other teams that didn't lay off and actually have bigger teams, like six designers, let's say, uh, I don't know, like ten engineers, they have two product managers, even them have a lot of pressure. And sometimes from competitors, even and mostly from investors right now for companies that actually have investors. What does that pressure actually look like inside product organizations today?

SPEAKER_01 5:03

The problem is that everybody has to be on the cutting edge and everybody's scared of falling behind. Right? And I even heard an incredible story the other day, and I don't know which company it is or anything like that, but I heard that one of these there was a company that was the number one in their field, right? And billions worth, right? Big huge company. They came turn around and they said, our strategy does not include AI this year because we don't need it. Because of that, they got dropped down to number two, and so they got lost billions just because they didn't have AI in their strategy, right? And you look and you think, well, this is how we're defining it. It's not that AI is going to bring more value or more impact. It's just because they don't have AI there, right? And I think this is what's happening. Everybody's feeling this pressure, and you know, and I think like LinkedIn, all these other systems we have actually really incentivize for us to feel the pressure like we're falling behind because everybody's posting about how they're doing an incredible thing, right? If we go into Instagram, we go onto any social network, it's all saying, oh, you should be doing this. You know, if you use these five problems, you're gonna solve, oh, just put this in the comments. You know what they're gonna do. They're trying to sell you something, right? But it's like this whole, this is this whole scheme around convincing people how to do stuff, right? And and that you have to invest and you have to do. And it's this urgency thing. And if we think about it, it's actually a dark pattern, right? We're incentivizing to exist even more, which is final seats are already, right? And we're doing this all around AI. So everybody feels really pressured to do it, right? Even within teams, right? And I see this with the team I'm working with now. We're not rushing behind it and trying to run. What happens is we're doing it where, you know, there's there's specific areas. If someone does something, they share it with the other people and everything. You don't feel like you always have to rush behind everything, but at the same time, the collective is thinking about it, structuring it, right, and and working together. So it's it's complicated, but teams are pressuring, right? And so almost all companies have to have it as part of their strategy. You know, they have to show how AI and and and the problem is that I think it's I can't remember the exact number, but it's between somewhere in the 80%, right? Where 80% of the initiatives with AI are actually failing. Why? Because people are just rushing in. Uh and and actually, I love to use a story that I don't know if you remember, I mean, this is talking a long year quite a while back, Photoshop started coming out with filters, right? And in the beginning, what happened was everything you saw online, all the designs you used to see, it was 100 filters, gradients, shadows, drop shadows, whatever, front shadows. It was like multicolored and all that. Then what happens is we went to the minimalist, right? And so we went to the extreme of let's go minimal, you know, even skeworphism went out the window. Let's reduce it as much as possible. Then people realized, hold on, it's not as great. It's how we so now we're trying to we're finding that balance, right? And I think with AI, it's the exact same thing. We're going to this extreme where everybody feels they have to do it, they have to. There'll be a time, and and I've heard some rumors coming from, you know, in the the the US in Silicon Valley about saying, oh, we're human started, right? You know, we're human first and this kind of thing. Because they're trying to fight against the extreme of hold on, let's not just go crazy AI. And I think we're trying to find that balance, but we feel pressure. And we're in that transition phase, which is scary.

SPEAKER_00 8:17

Yeah, because we can't we can't just build something with artificial intelligence without really meaning what we should or shouldn't do. Like what is the the goal with this? And I guess we kind of forget speci specifically when we want to deliver as fast as possible to show value. We live in a soci as us that we live in a society where we are most of us, or even all of us, always want to show others what we are capable of and how we are capable of. And even on LinkedIn, like everyone is sharing what they did good only, most of the time, because the way others perceive is like, oh, this person is perfect for this because everything good happens, and it becomes like a bad cycle, and then it goes to jobs with validation and validation. And I'm curious to know also from your side in terms of processes, what would you say to companies and teams specifically? How they should handle this new way of doing things? How how can they build processes that are healthy and not generating burnouts? Because this also generates a lot of burnout.

SPEAKER_01 9:27

Yeah, there's a there's multiple answers to that question. I'll try and see if I can break it down its parts. So number one, right, is the way we work, right? And the way we work right now, let's be honest. There's tons of work that we we've been doing over the past 10, 20, 30 years that we should totally throw at AI because it is very predictable, it's very easy to do, and we have been wasting our time doing it. Let's be honest. The human mind has been wasting their time, and it's actually kind of crazy how people are fighting against some of that more mundane tasks. Right? It's like, if I can offload it, I will totally offload it, right? I want my time to focus on the things I should be focusing on, right? So I think in our processes, in the way we work, we have to look at those key moments where we can add it in, but at the same time not removing, for example, you know, you can do ideation with an AI next to you, right? The problem is number one, you put yourself in a silo, right? Number two, what you do is you're not sitting in a team actually thinking about it, understanding it, working together, reaching a conclusion together, right? And there's power that's powerful, right? If you've ever been in a real good situation where you put teams in a room to align, right, with post-its on the wall, whatever you want to use, right? It's very powerful, right? It's very strong and it helps build up that context. And I think so we have to be careful about where we do it, but there are tons of areas where we can, right? It can simplify things. For example, if I have to do a product requirement document, I can do hot at least the first draft, put in AI. If I have to do, you know, like a one pager, if I have to do even some UX uh research stuff, right? There is some stuff you can use AI, and it's kind of interesting because, you know, I I remember I had researchers telling me, oh, you know the part I hate the most? They're doing the report, building a report, right? And you're like, well, AI can help you to an extent with that, right? Where it does the first draft for you, and you just have to obviously you have to put validation to make sure that it's the right thing because it's not perfect, it makes mistakes and all that. And great UX researchers are much better than AI can do right now. But still, it can solve tons of that stuff, right? You don't have to sift through every little thing, you don't have to transcribe everything because it helps you transcribe, right? So that's one part, the way we work. Now the other part is what we put in front of our customers is complicated because everybody seems like, oh, AI is going to solve everything, right? And it's like if we just plug it into all of our systems, it'll solve all our problems and we've got a new layer, right? The problem is if you do that, what do we have? We have 5,000 chatbots that no one can differentiate what one is different different to the other. Oh, there's a brand that's different in the corner, right? That's also not the solution. I think, and and it's one of the things, as I'm deep diving more into it, what I see is a lot of the product problems, a lot of the products, what we need and what customers need, they're not changing completely, right? There's a lot of stuff that is almost exactly the same. The difference is we can add an extra layer that can help us through some of the stuff, right? Where our customer doesn't have to look for something as much as they used to. They don't have to over-rationalize something. It can help them, it can make suggestions, right? Used before we used to have to depend on a dashboard to tell us, oh, what we can or can't do. Now there's actual systems that can turn around and say, do you know what? You can do better in your job if I provide you this. Let's talk about an admin system, for example, right? So I think it brings new new avenues and new things. It facilitates that. Like I said, you have to be very conscious about what you're doing, right? What is that problem you're solving? That hasn't hasn't changed. You still have to solve a problem, right? If you make it easier for people, in some cases, we might even make it where certain areas of in a company become redundant, right? Or they don't need as many people. You're like, that's perfectly fine. But we have to accept that, right? I don't like it, but it it sometimes happens, right? But if if our job is replaceable by AI, it means also our job was, to an extent, very close to being replaceable anyway by automation, simple automation. However, and this this is a disclaimer I like to put, which is, for example, there is an issue which we have, which is, for example, let's say, for example, and that task right before used to take them a week. And now with AI, it takes them two hours, let's say, right? Simplified, great, incredible. People don't turn around and say, now you've got the free time to have a work-life balance, to structure your day so you can think about the lease and everything. No, what they turn around and say is now you can do, instead of doing just one task in those that one week, now you can do like 50 tasks in that week. So the people are feeling overloaded, right? And and the problem is, and I don't know if you you've ever seen this, but I remember when we used to do Webmaster, right, where we had to do the front end, the back end, the the AI, UI, we had to do marketing, we had to do SEO, we had to do all this stuff. I always felt like I was context switching and I could never focus 100% on something, right? And this is the problem. People are gonna be so context switching, switching so much that we won't have depth in things. I think we'll start seeing less depth in things if people if we're not careful. And we need that depth, right? And the idea is this frees us up to actually create even better depth, right?

SPEAKER_02 14:25

ID.

SPEAKER_01 14:26

And that's not how companies are seeing it, right? It's like I'll throw this what it means, I'll throw one person in it and I'll substitute 50 people, but they're still gonna do the work of a hundred people, right?

SPEAKER_00 14:34

It's yeah, yeah. So And it's really it really comes down to like you were saying, to the cognitive load, because until now there was a lot of uh there wasn't a lot of time or space to actually discuss mental health, to actually work on the day-to-day lives, like having life after work. There's a lot of countries that don't have still life after work. And instead of taking this as an opportunity, we are doing the opposite. Like we are giving more work because we have artificial intelligence, so now we can do more. But then it comes, why are you doing more? That's not only a lot of work for the person that is trying to have work-life balance, but it's also a lot of things for the user, and there is going to be a lot of friction because there's a lot of types of different users that use tools different ways, and that's a lot. So, uh, how do you think we can balance the first of all the part of how that can can we balance work-life balance and still have uh this great new innovation and that gets very interesting, that is artificial intelligence. But in the in the other way, how can we use this moment to actually look into what we are building and release it in a way that our users will not feel the need to leave because it's too much for them?

SPEAKER_01 15:57

It's so there's two points here, right? One is it's it's a vicious circle because what happens is the more easy it becomes for us to release new products, right, which is happening right now, the less value you can get from any of the products you release, right? It starts reducing the amount of value you can get. So what companies do is they feel they have to hustle more to make more money, so they accelerate with with how folks are doing it, right? And like you were saying, what they do is they don't turn around and say, well, you know what, I could have the five people on my team working a half day and producing double the work, right? What I turn around and say is now I can have those five people producing the work of 50 people, right? That that is that is one aspect. I think what happens is the companies that will start having success and that will be healthy in this new paradigm that's coming up is the ones that consciously look at this and say, the ones that again, it's it's that type of thing where you're in the crowd, right? You're seeing tons of movement and everything like that. And that one person who just steps up on a chair, looks around and says, Oh, I want to go in that direction, I'll go calmly in that direction, right? I'm not gonna get lost in the craziness. And here's the same thing. People who will sit down and they'll look and they'll say, let's take a step back, let's take it calmly, let's not go behind the rush because AI, definitely go with it. Emerging technologies, right? Because AI is not the end of it. We'll still keep seeing new things coming up. But you do it consciously, you do it calmly. And I actually think it's people who produce, you know, one or two really good solutions rather than 50 horrible solutions, right? So that hasn't changed. It it's exactly the same you could say 20 years ago, right? It's the companies who are doing this consciously, right? They're doing it enough. Um, and and it there's a point here. I think also AI is bringing up something which is there's a lot of companies who in the previous, and and this is why we're seeing a lot of the layoffs is previously, and I've complained about this very often, is companies had the concept of if I have a problem, I'm gonna throw people at it, right? And while there's money coming in, we can throw more people at it, right? And the problem is there's an S-curve, right? Which is you're super efficient the more people you add in the beginning, then it comes to a part where it plateaus, and then the more people you add, it actually starts decreasing, right? And the problem is in many of these big corporate organizations, you go in and you see a lot of waste. And AI has made that glaringly obvious, right? In some cases, and you turn around and like, for example, you look and you say, do teams need to be as big as they are? In most cases, no. Right? People have proven that. I like to use, I like to refer the book from the folks from Basecamp, the rework book. I love the way they look at things, which is we build the most lean team possible, then we add people only if needed. Only if needed, not because we've got new funding or because we've got we've, you know, because I want to open up a new area, or because of this. You turn around and said, you know, you want to open up a new area, maybe two or three people could do it. You don't have to have suddenly this new squad of 20 to 15 people or something like that, right? And I think this is the problem. We just always assume it's like, let's throw people at it, let's grow, let's grow. And then suddenly you turn around and you say, well, with AI, if half of that manual work that they were doing previously goes out the window, you suddenly have people who don't have anything to do, right? Or have a lot less to do. And so I think it's also exasperated that to an extent, and we have to be careful about this. So I think it's a mix of things that we're going, like I said, transition phase right now, and it's it's scary. It's and I think it's gonna get scarier before more comfortable. But hopefully we start seeing the comfortable coming in soon because Yeah.

SPEAKER_00 19:24

And I guess something that I've been reading a lot, it's about the many companies are experimenting with AI across multiple teams, teams at the same time, which can also lead to duplicated initiatives, disconnected tools, fragmented efforts, which I think it goes hand in hand with what you are saying right now.

SPEAKER_02 19:42

But nothing's changed. It's suspicious 10 years ago. Right? So, yeah, exactly.

SPEAKER_01 19:49

But the difference is they put it out a product all in two weeks, right? And one sprint new product out, and you're like, oh no, it's it's much worse. It's actually you the potential for getting worse is even higher than it was before.

SPEAKER_00 20:04

And it becomes quick. I read something about this. Now that with artificial intelligence, we get to see the companies that actually are doing the work for the user, not the money, but also the companies that will end quicker than the other ones, because they have no processes, they have no actually manageable team, they it's just showing, like you were saying, it's showing the people that actually do it and the people that we're trying or let's say try do it to do it, but the problem is when people start looking at the actual cost of AI for them, you know, in a lot of cases, because for example, if you're building a small little startup product, perfectly fine, you're using a couple of thousand tokens and things like that, right?

SPEAKER_01 20:46

But suddenly, if you're using an enterprise grade level product, right, that's going through thousands of people, the amount of tokens you're gonna be using constantly, I don't think they have a notion of the cost that'll be coming through.

SPEAKER_00 20:57

Yes, a friend of mine was telling me about this, was telling me that the company lets her in once, like it says, you need to work to use a lot of like NA10, for example, you need to use NA10 to do your work and have LLMs uh summarize things and do things, do the process like that, the workflow like that. And then she was like, Okay, fine, I will do it. Like she was very stressed because there was no like direction, there was no actual goal. They just said, like, you need to do this, there's no way you really need to do this. And she started building the workflow, but because of compliance and other teams that had other rules and other tools connected, she could not do her work because the system didn't wouldn't let it. So that comes again with legacy systems, internal policies, like even operational constraints. So that's another thing that it's interesting. So the leadership, or even like like you were saying, like everyone on top is saying to people that are working that they need to use this, but none of them are teaching how to. To use it because they also don't know how to use it. And then it goes around and around and around and it becomes FOMO from everyone. And like it breaks and it breaks and people will. Yeah. Go ahead.

SPEAKER_01 22:13

No, and a great point you're making there, right? For example, the amount of people that know how to use GitHub versioning, branches, that kind of stuff, right? It's not everybody. So tons of product managers have never done it. Designers have never done it. Researchers might not have never done it, right? And so the thing is you turn around and say, okay, even if we teach them all how to do the versioning in the GitHub and all that stuff, right? The problem is that it takes just one designer who doesn't know how they do it, working with AI or one PM or one researcher, whoever, right? Who doesn't know what they're doing, and they go in there and they say, Oh, Claude, put together a code for me. And now I'm gonna put it live on the site, right? And then suddenly you on the platform and everything goes down and nobody knows. And so everybody has to sift through all the rubbish trying to find what's going on, right? So that's the thing. I think it's like, oh, it's easier for me to code. Yes. So is it ready to go live? No, I wouldn't like it. I wouldn't take it coding live, right? But you know, it seems that's the problem, right? Just like you're talking to Claude or you're talking to Chat GPT or one of these systems, right? When you're talking in the chat, they convince you, they it's easy for them to convince you that they are, you know, um professionals, that they know what they're talking about, right? They do it really, really well. When you write code, right, and if you've done it with Claude, you spend just an afternoon and it creates something for you, lovable, whatever you want to call it, whatever you want to use. They create something. And you look at it and it's like, this is an incredible system. Looks incredible. Even if you've asked it to do, you know, for example, uh, at one time I tried implementing with Vercell and right, so it's a host system there, and then a database in um Superbase. And you look and you say, it's working perfectly fine. It's incredible, it's doing amazing stuff. Makes you seem it makes it seem like it's ready to go live, right? And I think there's a lot of harm that's coming from there because we're gonna get so much crap on there. Um, you know, and it's gonna be because even previously, even the way it was before, I think we were already getting tons and tons of crap, right? Tons of terrible, terrible. I used to throw this number out, and when I used to give talks, and I used to say oh, I would guess it's about 80 to 90 percent of what's out there in terms of applications, web applications and everything, it's crap, right? Now I think it's gonna get closer to the 100%, right? It's like well, maybe have two or three percent of things that are really, really good. Uh again, pure assumption. Me this is me guessing, but yeah.

SPEAKER_00 24:32

This has been the first half of my conversation with Ugu. If you enjoyed it, make sure to like and subscribe to the channel and tune in for part two coming on this Thursday. In the meantime, feel free to message Ugu on LinkedIn. See you then!