# AI Product Builders: Readiness, Risks, and Role Evolution

**Podcast:** All Things Product with Teresa and Petra
**Published:** 2026-05-12

## Transcript

Hi folks, this is All Things Product with Petra Wille and Teresa Torres.
And we're so happy you're here.
Petra, everybody's talking about product builders these days.
And I know you have some skepticism or maybe just a nuanced perspective as we tend to do.
Tell me about your thoughts about product builders.
Yeah, it's not so much skepticism.
It is more...
Coming from the conversations that I'm currently having with product leaders in the coachings, where we discussed this kind of org setup of the future, right?
Because it's what every product leader right now is currently thinking about is, am I well staffed for the future that is laying ahead?
So do I need people with different skill sets?
Do I need more senior folks and less junior folks?
Do I need more junior folks that are more literate in AI?
So what am I actually?
What are the scenarios that I'm planning for, right?
And one thing that we figured or one thing that I realized we're having this conversations is that it starts to surface that a lot of people are deep down in AI experimentation in these teams, but then come to a point where they're like, and I'm personally at the point as well, and I can double tap on that in a second, is do I like the work and how it is?
has changed that I'm doing right now right because it feels like being less creative and more like having an intern job shadowing you the entire day and you have to tell this intern in my case Claude Code where to go what to look for hey we have created this finalized week why have you forgotten about it all these kind of things that you tell the intern constantly and Yeah, one thing that we've realized in these conversations and coaching sessions is this, just because I can do it, just because I can now work on code as a product person, is it something that I enjoy doing?
And is it something that I have enough experience to be really good at and fluent and getting into the flow that I have gotten from, I don't know.
writing backlog items half a year ago.
I don't know, right?
Whatever the job is.
But that is the conversation that we're currently having.
Are you seeing something similar?
A little bit.
I mean, there's certainly a lot of conversations about should product managers be coding?
Should they be vibe coding?
A lot of companies are now putting AI prototyping tools in the interview process.
I think you raised two things.
And rightfully so.
Rightfully so.
Yeah.
I think you raised two things that I think are a good way to frame it is, is it fun for you?
And what was the second thing you said?
Yeah.
Are you having enough experience to really be good?
Oh, yes.
Is it fun for you?
And do you have enough experience to do it well?
And so like the thing that excites me about this product builder trend personally is it's really fun for me.
And I think like if it's not fun for other people, I think there's probably still plenty of other parts of the job to do.
And so like I wouldn't say I don't fall on either side of the argument of product managers should vibe code or product managers should not vibe code.
And I've seen both arguments.
Right.
Yeah.
I actually think it's a tool in our toolbox that we can decide who on our team is best versed in this.
has fun with it, wants to do it, wants to contribute to it.
So I think the fun piece is the most important one.
I think the second, the one around skill, I think it depends on what we're talking about.
So if we're talking about prototyping to explore ideas, I think the tools are good enough.
Anybody can do that.
If we're talking about creating production quality code, I think we're getting to the point where the tools are good enough.
For production quality designs, by the way.
So design systems, that's another.
Yes.
Okay.
This is where I'm getting at.
Like, yeah, I think we're getting to the point.
If you have a design system in place, if you have your engineering practices set up to support coding agents, if you have quality assurance, if you have automated testing, if you have a good CICD process, if you have a good code review process, like for some organizations, even.
people without engineering skills are getting to the point where they can release features.
Yes.
If they want to, if that's the way they want to work.
I think where we get into trouble is people make these like extreme arguments of all product managers should be doing this.
And like a lot of companies, their code base isn't ready for it.
Their engineering practices aren't set up for it.
I do think engineers will always be involved in How does this work at scale?
How does this work securely?
How does this work in a way that's going to be maintained?
Maintainable over time.
So yeah, like I love this idea of like, if you want to create, it's now way more accessible.
But I don't love that like people put shoulds all over it.
You should do this.
You should not do that.
Yeah.
And we recorded another episode on that one is kind of, Sometimes overestimate the expertise you are having in the field that you have not been exposed to enough.
For example, if I have conversations right now with product leaders, heading product managers that are now vibe coding a lot, I can derail them rather quickly with the question of, but has anyone checked in on non-functional requirements?
So like maintainability, enhanceability, all these kinds of things, right?
And then everybody's staring back at me.
Nobody has ever heard of these things.
And I'm like, yeah, okay, maybe you should still involve your engineers every once in a while.
And I had another extreme case where a product person was tasked to create a design system.
And this is too big of a task for a product person with no design expertise, even with the help of AI.
It's just not something that...
is coming out healthy at the other end because all this inclusion considerations that you usually have to have and even design trends are a thing and you have to read the room as in read the internet or read the app store or whatever to sense what the current trends are and all these kind of things if you know that this is a thing.
visual trends and then you still can use an AI to help you extract those and learn about what is currently trending and what is maybe looks outdated, how icons maybe change over time and that we're not using the floppy disk any longer for safe because Gen Z is not getting it.
All of that is important knowledge and AI can help you to uncover those but not if you never heard of these kind of things that icons are following a trend as well.
But to me, this is the organizational challenge, not an AI challenge.
Yes.
Because like product managers have done design since the beginning of time.
Look at how many product teams don't have a designer.
Is that good?
No.
No.
Is that what has been happening in practice forever?
Yes.
Does AI at least raise the floor?
Like a product manager with AI is probably a better designer than an unskilled product manager without AI.
Yeah.
Yeah, I know, but there is danger to it because the stuff that they deliver looks way better on the surface level, but crumbles apart really quickly if you really dig in as a team, for example.
But for executives, it looked amazing.
So they were like, okay, but it looked amazing.
Why are we now having all the discussions with the teams?
Now they're building it.
Yeah, okay, so I agree with you.
It might exacerbate some of those things or it might turn up the volume on some of those things.
But I don't...
I still think this is just a tool.
Like the symptoms we're seeing around this are because organizations undervalue design.
Organizations undervalue the engineering required to make something scale.
Organizations undervalue automated testing and the value we get from that.
And so it's really easy for an organization to look at this and say, of course, product managers can do this.
Because they don't value the things that the product manager isn't good at.
They value, we shipped a thing, we got it out the door.
Who cares that in six months it's going to be.
Yeah, I agree.
And maybe to share another observation, when you talk about what the organization values, I see a similar thing happening with what happens with the time that is freed up by the use of AI.
And you can really tell how far an organization came in their business.
product transformation.
Because there is this natural tendency to either focus on more output.
Now that we have a bit more time at hand, let's build more features.
Or they default to amazing.
We can be more innovative.
We can do more discovery.
We can even look at more opportunities and run more experiments.
And this is so interesting to see right now how many companies have this strong output-oriented muscle memory still in place.
And it's another leadership challenge that I see right now that leaders need to manage that kind of freeing up of time and what we do with it as an organization.
Yeah.
So like, I think for this type of topic, it's fun to explore both extremes.
So like on one end, we have an organization that's perfectly set up to take advantage of this technology.
So what does that look like?
They have code review processes that evaluate Is this designed for scale?
Is it designed to be secure?
Does it have automated testing?
That's automatically in place.
So when a product manager ships a feature, they get pushback.
It doesn't get accepted because it's missing these things.
Or even better, they have skills associated with their coding agents so that when the product manager builds it, the agent knows to do those things, even though the product manager doesn't know to do those things.
They have a design system in place.
that a designer designed that is used across the organization, right?
Now forget product manager.
Now you could have someone in the marketing team.
You could have somebody in the sales team.
You could have somebody in the customer success team build a feature and contribute it.
I think there's like, that's the extreme case.
I think there's a lot of positive upside that comes to that.
It raises questions like the bottleneck moves.
Now it's how do we evaluate what should actually get integrated into the product?
So when we talk about ideas can and should come from everywhere, now those ideas are coming in in the form of code.
Yeah.
But we still need the same process to evaluate what gets integrated in the product, who owns coherence.
And I still think that's a role of a product manager.
Yeah.
Yeah, ideally.
And we need somebody setting up the...
AI systems in a way that they comply to all these things that we set up.
Exactly.
So that is another major thing we will see in organizations to basically be creative.
Then I think we can look at the other extreme, which is the organization has none of this in place.
They have no design system.
They have no coding practices.
They have no review process.
They have no CICD process.
They have no automated testing.
Ideas come in from everywhere in the form of code.
Straight to hell.
And they end up with totally unmaintainable code.
Their design is, it's utter chaos, right?
Yeah.
And I think the challenge is people look at the tool and they say the tool causes this chaos.
Because they're not aware of the other end of the spectrum.
They don't realize that like the tool can be effective when the environment is right for the tool.
And I think this is on leadership, right?
Because they just say adopt AI at all costs.
Yeah, that's why I look at like, what's our organizational readiness for this?
Yeah, that's why I'm having so many of these in-depth leadership conversations right now.
And sometimes I'm forming my opinion why I have the conversations, frankly speaking, because so many things are so new and even so much conversations and viewpoints and stuff like that are really new and you need to think it through.
all of them, all at once at the same time right now.
But yeah, it has a lot to do with when new, I don't know, programming languages came up and everybody fell in love with Ruby on Rails because now finally more people could use it because it was easier to use and yada, yada, yada.
But nobody...
back then said like I put it everywhere on all the roadmaps and this will basically help us efficiency gains massive efficiency gains that would have been ridiculous but now with AI because to so many leaders and executives it is they still don't well understand what it actually does and what it actually is.
They understand it's powerful.
And they know that there is a lot of pressure from the market, basically, either from the customers asking about what are you doing with AI in the next two quarters or investors are having these questions as well.
So they need to have a take on it.
And currently, a lot of them just paint AI.
all over their slide decks and that's basically where they leave it.
And then it's on this next layer, which is the product leadership team that I am usually talking to.
They have to figure it out.
Yeah, this is where I think it's more similar to like the mobile transition, right?
Like we need mobile everything.
Okay, wait, hold on.
Let's talk about what's appropriate for mobile, right?
feels different than just a technology thing.
That's why I equate it more with mobile and the internet as opposed to microservices or I forget what the other technology you brought up was.
Ruby.
Oh, Ruby Unreal.
Just to make it really simple with the programming language.
Because I think it doesn't just change our products.
It also changes the way that we work.
Yeah, and it will change the organizations.
Organizations are going through change at like layers.
And what's interesting is the individual has adopted it really fast.
Yeah.
So the mushrooming is what organizations struggle with a lot as well.
And this is not when you're working for a bank because it was.
clear to every employee of that bank that mushrooming with AI tool and just like experiencing, experimenting with them, maybe not a clever idea in such a regulated market.
But there are so many other companies that are like not that well regulated.
And their individuals learn so fast to introduce tools without the organization even knowing.
And now the organizations have to pick up a lot.
Yeah, so I think it's...
At the individual level, it's being adopted really fast, but that's creating chaos because we don't know what it looks like at the team level.
I get that asked all the time.
Is that like, okay, I'm using cloud code.
How do I share with my team?
How do I collaborate with my team?
And then we certainly don't know how to handle it at the organizational level.
And that's just on how the organization works.
We haven't even gotten to how does it impact the roadmap and what you should be building.
Exactly.
And that's exactly what I wanted to add.
There's a difference between personal efficiency gains that I want to scale throughout the organization, right?
Yeah.
That's one thing that I see happening.
But then there is the way how we all work.
So really like process impact of AI.
And then on top of it is what you're saying, impact on the actual products.
And everybody is still trying to figure out all these things.
Can you?
You don't have to re-say what you said.
No, what I basically said is like everybody's figuring stuff out on the personal level and then personally efficiency gains.
Then you want these personal efficiency gains be gains for a bigger group, so the team.
But then there is this.
entire layer of process change that needs to happen to really leverage the power of AI.
And then on top of it, there is the product layer, how it is influencing our products.
And that's different things to look at.
I actually think the product layer is just a whole different stack.
Yeah, yeah, yeah.
I get asked all the time, how does AI change discovery?
And nobody wants to hear that you still have to talk to your customers.
Like, that's still the input.
It doesn't change.
You still have to care about your customers.
You still have to talk to them.
Like, sorry, you can't just ask Claude.
It's not really going to work.
But I think, like, when you're deciding what to build, it's not about personal productivity gains.
No, you might get personal productivity gains in terms of scheduling, in terms of synthesis, right?
But like, fundamentally, there's still this exploratory, innovative process that you don't just layer productivity on top of.
I think what you layer on top of it is like how deep you can go.
And I think AI can probably help with how deep we can go.
We can now make sense of...
all the support tickets that are coming in and all the sales calls.
And that's amazing.
But it's not the full story.
And so I think that like how it should affect your roadmap, I think of it as a different stack from the organizational productivity stack.
Yeah, I totally agree.
And you need to manage both really well.
And I like what you said in the beginning, people should think in the extremes and in scenarios about both of these stacks as well.
Yeah.
Thank you, Teresa, I would say.
I was going to wrap up with, if we go back to the beginning, product builders, the myth part might be not everybody has to be generating code, but everybody can be if they want to.
And if you don't want to, there still might be work for you to do.
Just find other skills that your business cares about.
And you're excited about.
Thank you, Teresa.
Thanks, Petra.