Episode highlights

00:02:10 – Recent changes in the product industry 

00:04:08 – Dealing with the changes 

00:06:55 – Changes happening in Miro 

00:11:47 – Bottlenecks in the product development 

00:14:05 – Overcoming dangers that AI brings 

00:24:12 – Ensuring research is not an afterthought 

00:29:00 – Non-negotiable skills for UX researchers in 2026

00:33:20 – Final advice 

About our guest

Deniz is a Staff Product Researcher at Miro in Berlin, exploring how product teams work and build together in the agentic era. His focus is on how people collaborate and communicate, both with each other and with AI agents, and what the product development process will look like as those workflows evolve.

He’s also an agentic builder, moving beyond uncovering insights to using AI to shape and build solutions directly. One example is the conversational intelligence and discovery tool he built, which is now used by a large share of Miro employees.

Outside of Miro, Deniz is a conference speaker, mentor, trainer, and consultant.

AI is not taking anyone's jobs, but everyone collectively is evolving their job descriptions and what it means to work together.

Deniz Kartepe, a Staff Product Researcher at Miro.

Podcast summary 

In this episode of UXR Geeks, Petra Rajkov talks with Deniz, Staff Product Researcher at Miro, about how agentic AI is changing product development and the role of UX researchers.

Deniz explains that product managers, designers and researchers can now take on parts of each other’s work, which blurs the role boundaries built during the agile era. He shares how Miro encourages this by giving everyone access to AI tools and creating communities where teams share workflows.

But according to Deniz, faster building does not remove bottlenecks, it just moves them  upstream to discovery and definition. When anyone can build anything, teams risk developing and shipping the wrong thing, so understanding what to build matters more than ever. That is where researchers come in. 

Deniz also argues that speed does not have to cost depth. Well documented research plans, notes and transcripts let AI build on past studies without sacrificing much quality. A chat based repository also lets stakeholders work with research insights faster and query findings themselves, saving Deniz time. 

Deniz sees researchers becoming “builders with research specialties,” for whom flexibility and adaptability are essential. He advises researchers to experiment by building and breaking their own workflows and embracing the changes happening in the industry. 

What this means for UX teams

For UX and product researchers 

Your value now sits in defining what should be built and aligning the team on it, because building is no longer the bottleneck. AI output is only as good as the structured plans, notes and transcripts you feed it, and you remain responsible for reviewing what it produces.

For UX designers 

You can now build working prototypes directly, so your specialty shifts to owning the insights and design quality behind what gets built, not producing the artifact. Discussing half built ideas with engineers and PMs replaces the handoff between roles.

For junior researchers and career changers 

Stay flexible, utilize AI when appropriate and don’t try to flight the change in the industry. AI is a tool that can give you great advantages and it makes learning new skills, including design and coding, much faster than it was for researchers before AI. Experiment and don’t be afraid to build and break your own workflows to get comfortable.

For product managers 

Always align with stakeholders on what you’re going to build and how before you actually start building it. Although building new products is becoming easier and cheaper, developing the wrong thing still costs time, money, personnel, and results in a lot of deflated spirits. 

Podcast transcript

[00:00:00] Petra Rajkov: Welcome to UXR Geeks, where we geek out with researchers from all around the world on topics they are passionate about. I’m your host, Petra, a senior UX researcher and a research advocate, and this podcast is brought to you by UXtweak, the UX research platform for recruiting, contacting, analyzing, and sharing insights all in one place.

Today’s guest is Deniz Kartepe, a researcher at Miro, who’s been closely observing how the way we build products is fundamentally changing. As workflows become faster, more fluid, and increasingly powered by AI, Deniz is exploring a big idea. What happens when everyone, from engineers to product managers to researchers, actually start acting as a builder?

And more importantly, where does that leave UX research? That’s exactly what we will dig into today

Hi, everyone. Welcome to the new episode of UXR Geeks. We have another amazing guest with us today, Deniz, a researcher working in Miro. Welcome, Deniz.

[00:01:26] Deniz Kartepe: Thank you, Petra. Thank you for having me. Nice to be here.

[00:01:29] Petra Rajkov: Yeah. We are looking forward to the episode, and I think we can dive right in, and I would like to start with your hot take.

So over the past years, what changes have you seen happening in the industry in how products are actually being built?

[00:01:47] Deniz Kartepe: The realization I guess hit me when checked my GitHub account that I had since 2018, which was completely idle, you know, without any action. And then I started working with our engineering teams in December to dive into these new workflows and how things were changing for engineers, right?

Because they are like the canary in the coal mine, where they first start adopting the new changes. And that was the time where Claude Code was picking up for the first time around those times within our community at least. And yeah, to be able to do foundational research for engineering products and like how engineering workflows are, were evolving, I needed to dive into it as well and better understand what’s happening there.

So get familiar with terminal, with the IDEs and yeah, just try to get into that life. I dove into it and I’ve never looked back. So all of a sudden it just like you have this timeline of your contributions on GitHub and then this complete darkness for the last 10 years, followed by this like lightning green activity each day, dozens of different commits and things that we are doing.

And quickly I realized I’m not the only one. So everyone in the EPD trifecta, like the product managers, designers, and anyone else involved, the p- product marketing managers, researchers, analysts, are all in a similar path right now. Maybe it’s in my own bubble, but I have been doing a lot of interviews around the space as well because I was also doing the product lifecycle research, talking to a lot of different companies.

So I feel like this wave is happening in multiple places all at once and it’s exciting. Fun times to be alive, I would say.

[00:03:30] Petra Rajkov: Yeah, I love your approach and your attitude because I was going to ask how did it make you feel before you started diving in to that and seeing all of the changes? Because for me, I sometimes feel anxious, but you said that there are fun times ahead.

So did you go also through this whirlwind of emotions or were you always this excited?

[00:03:51] Deniz Kartepe: Because I’m working with the engineer product teams now and that Miro really went all in on enabling everyone, right? Giving huge access to every single AI tool imaginable so that they, in sandbox environments they can try to find their own workflows that work for them, and then created these enablement communities like AI Coding Guild, AI Product Guild, AI Design Guild, news we’re sharing.

And those Slack channels are right now the busiest channels that we have. Every other hour there’s this huge post about the next big thing, and it’s so scary, so overwhelming. It’s like trying to drink from a waterfall. And I was talking with the engineering products team saying, “I will go on a two-week holiday, but I’m freaking out to come back and see the entire world change.”

I was terrified of that. I was like… Because even when I go back for a weekend and then log off for the weekend and then come back on Monday, I would be seeing all these messages in these channels because over the weekend people shared, “Oh, new model here, new product there, a new workflow there as well.”

And then you try to follow it all up and it’s impossible. But then I kind of realized, especially after that little trip that I had, I went to Southeast Asia for a bit and then came back after a break of a couple of weeks, and then when I was trying to catch up, I realized, yes, a lot of things happened.

But when you zoom out a bit and look at it through a more broader lens, the things that have happened are actually pretty followable when you do that zooming out, so you can still follow. Yes, you will miss some details, but overall if you just do not look at every single raindrop and watch the rain as a whole instead, you are still able to follow.

And then that gave me that comfort and now I feel a bit more at ease about it. Before I was freaking out because everything was happening so fast and I was like, “Oh, I want to be at the center of everything that is happening so that I don’t miss anything.” But it is impossible to not miss and it’s perfectly fine to miss.

And so that’s my approach now and I started to deal with that stress. Otherwise it was a bit overwhelming in January, especially, yeah, end of January until mid-March, like January to mid-March it was a bit intense I would say.

[00:06:00] Petra Rajkov: Yeah. So it’s good to take a step back and look at it- Yeah … from a- Yeah … another perspective.

Yeah, definitely. I can see that. And I love that you shared about what you are doing in Miro right now, that you have these channels and these places and spaces where you can learn from each other and follow what is happening and things like that. So  can you walk me through few of the processes that have changed in Miro, specifically about how the features were built before and what you are doing right now?

[00:06:32] Deniz Kartepe: Yeah, like the entire model was built on, I guess, the agile transformation, digital transformation of companies that started at late 2000s, built up these established agile processes and then created these EPD trifectas and then like the engineering manager and the engineers alongside a designer and a product manager with set duties and then this horizontally cutting functional roles that would come in to help was pretty well established.

Everybody knew what they were doing and what their responsibilities were, and they deferred a lot of critical tasks to certain roles, and that went fine. And now agentic AI turns it upside down and everyone is able to get some parts of the other roles and then run along with it to a measurable level of fidelity where you can discuss those ideas that are already half built to have like concrete discussions around it, what is going to be built, what you think should be built and why, and then you can build it and show it and just explain the logic behind it.

So all of a sudden these… first of all, everyone is like super fast now. I’m able to do my research at a 5x speed. Now my question is, am I going to be doing five times more of the same thing that I have been doing, or I will use that time to expand horizontally to other functions as well, so I have a more holistic approach to product development workflows.

The same thing goes for everyone. A developer can do a tremendous desk research. Uh, a product manager can do a very high fidelity prototype by coding it. Uh, a designer can do the same. But then all of these roles that are kind of like these builder roles with specialties then can get together and be a single unit that uses their specialties, but always contribute in the same kind of approach, which is continuously thinking about the actual thing that is going to be built.

No more abstraction layers where I do the research, serve the research and hope that it goes into the minds of everyone or a designer just designing and putting it out there or engineer, blah, blah. So everyone kind of following the same syntax, but specialties allow coders to make sure that it is enterprise ready, for example, because maybe you do not trust the Y coded parts that much.

Or product manager is the last DRI to make sure that it’s communicated well. The researcher or designer are responsible for the insights that feed into it or the designs that actually are going to be materialized. But it is just a part of their work now. It’s how I’m seeing it and it’s exciting. It really speeds processes up and the learnings that we had over the last decade are not holding it up.

If I do the researches the same way I used to do three years ago, I would be the biggest bottleneck in the universe, and same goes for everyone else. And Then what we realize is the more you go at this, like solving these bottlenecks, I am five times faster now, and then I am doing the other things. I code as well, I contribute to design as well, I create the PRD as well.

It means that some bottlenecks get solved. That was what realized as well when I was talking to dozens of engineers. I also shared a bit about it in some talks already. But when a bottleneck disappears, it doesn’t mean that an entire workflow is now working smoothly. The moment you’ve fixed one part of the plumbing, bottleneck moves either upstream or downstream.

So for example, if coding was the bottleneck where people had to use a less capable co-pilot, et cetera, to write code, but they would still manually write the code, coding was the bottleneck of the entire product development life cycle. But the moment we say, “Okay, AI is gonna make that so fast,” then if that is so fast, something has to become the bottleneck.

It is always the weakest link, and then it becomes the alignment that is a bit upstream or code discovery, like knowing what to code, the definition phase. And the same goes for every single role. So the bottlenecks never disappear. The moment a bottleneck goes away, it transfers the identity of being a bottleneck to the other one that is the weakest in the chain now, and then you have to go and look at what can be done there.

And that has honestly been my challenge, my Everest to kind of summit in my researches that I’m doing right now, which is, yeah, where the world is going, and in that place, what does product development life cycle looks like, and then what are the problems and bottlenecks and pain points there so that, for example- Oh, I think we’re in-

Miro can build the next things that can support and ease people’s processes there.

[00:11:23] Petra Rajkov: So what bottlenecks do you see right now? You mentioned that the research is faster. The communication is easier because everyone has different roles and you can talk and discuss together.

[00:11:35] Deniz Kartepe: Yeah.

[00:11:36] Petra Rajkov: Where did the bottleneck move?

[00:11:38] Deniz Kartepe: Upstream. The absolutely upstream. So if we are thinking about it as a river that is running downstream, building something was the bottleneck, and now floodgates are open. We have built a standalone research tool in two weeks with another UX researcher in the team, like Michael, and it is internally already being used, and it is great.

Uh, it, it is unthinkable that we could have done that a year ago. So if that building phase, downstream building phase is solved, now automatically we come upstream where if everyone is able to build anything, they are probably going to be building the wrong thing.

[00:12:13] Petra Rajkov: Yeah.

[00:12:13] Deniz Kartepe: Or just build it bad. So knowing what to build and how to go through that definition phase Is becoming the bottleneck.

Of course, when we call like discovery, define, deliver, like big three steps of product development life cycle, it is too abstract to define the bottlenecks. It is the little steps within each. If you were to think of them like three, not diamonds, but three steps, they are actually made up of multiple substeps. But most of the pain points that are emerging now, the new bottlenecks, are in the definition and discovery phase. So making sure that you understand the signals that you’re getting, huge part. Aligning on what you are going to build and how before you actually start building it, becoming a bottleneck.

And new opportunities that arise there because the more pronounced a bottleneck is, the more opportunities there are, the more room to play for companies. Because yes, for years in discover, define, deliver was a huge bottleneck, right? And look at all these companies that has gotten unicorn status, tens of billions of dollars of investments, and being ubiquitous in all these companies, GitHub, Jira, blah, blah.

It was such a big bottleneck that billions were just blown into it. Now that is easing up, and that automatically makes other parts a bit more visible in terms of the issues that they have.

[00:13:42] Petra Rajkov: Yeah. And there are a few dangers, and I love that you mentioned already one, that a lot of products are now built faster, but maybe they’re building it wrong.

They are building the wrong features because it’s based on assumptions. A lot of help with AI, which is biased, it can hallucinate, and so on. So do you have any strategies to overcome this danger? And then I will talk about another danger that I see.

[00:14:06] Deniz Kartepe: Yeah. Yeah, it’s if you are planning, uh, a trip, uh, to, uh- I don’t know, Patagonia, you will be preparing four weeks in advance, right?

To make sure that you’re ready for it. If you– Are you in Austria right now?

[00:14:22] Petra Rajkov: I’m in Bratislava, in Slovakia.

[00:14:24] Deniz Kartepe: In Slovakia, okay. So if you’re planning a trip to Brno, probably you’ll be like, “Okay, I can just– It’s a day trip. I don’t even have to think about it.” You don’t have to think about what you are going to pack, blah, blah.

So the problem there is now this. When building becomes that easy, no one really actually thinks. If it was taking so much money and effort before, people actually focused on making sure that discovery and definition phases were actually nicely covered because it was expensive to try and build that thing.

But now it is this collective tragedy, let’s say, where we are creating these so many solutions, so many products that do not necessarily a-address anything, and then they create this entire swamp where like in the field of UX, we were co-talking a lot about like banner blindness, right? And that was because there was this accumulation of unnecessary and uninteresting ads in every single user interface that we were seeing when we were doing eye tracking tests.

People would automatically recognize what might be an ad and then skip it, and that was called banner blindness. Now, in our entire digital ecosystem sphere, we have so many slop that there are products that have just been built because they are easy to be built that you have to ignore or look out for that maybe a new definition is building as we speak about it.

So a lot of products that don’t actually address anything. It was just so easy to build, so someone with a morning idea just built it. Especially when it comes to doing it as a work, I understand the developers doing it, it brings a lot of challenges, risks, and it can really take a company into a false direction for a long time, which will definitely have costs.

So just because it is easy To build, you can directly say, “Ah, highest paid person’s opinion. Let’s build it because we want to impress them.” Go on a six-month odyssey only to realize what you built is not actually what was needed.

[00:16:33] Petra Rajkov: Mm-hmm.

[00:16:33] Deniz Kartepe: And yes, building it was easy, but still it costs time, money, personnel, and a lot of deflated spirits.

The thing that we can do there is if building is easy now, let’s make sure that we build the absolute best thing that we can build by understanding what we need to build and how.

[00:16:52] Petra Rajkov: So understanding the problem right at the beginning, yeah … of the whole product development cycle, identifying the main questions and things like that, it’s more crucial than ever because then everything is just sped up, right?

[00:17:06] Deniz Kartepe: Yeah, yeah. For example, we have a big research team. When you’re working in a company like Miro, it is extremely easy to reach out to users because I have worked in half a dozen companies in my life and I, like, none of them had the satisfaction scores from customers as high as Miro, right? So when you’re sending out an email, you can directly…

People are like, “Yes, I will do it even for free.” “No, we want to pay you.” So that it’s so easy to start getting structured feedback in 60-minute in-depth interviews that are free-flowing, nothing with a certain topic. Just tell me about your workflows. That’s roughly you did for product development life cycle, engineering product workflows, researches, where we’re like, “How is your work evolving right now?

What tools were you using a year ago, and what tools are you using now? How has that shift been? What are the main drivers behind it?” And so on, and then start diving into it because what you want to build, especially if you’re a productivity tool in the enterprise ecosystem, is to understand what is your role to play in relation to the status quo that has been established, the playing field that is available for all the tools because this change is huge.

A lot of tools are becoming obsolete. A lot of tools that didn’t exist a couple years ago are ubiquitous, and if that entire pie is being replaced with a new pie, you need to understand how much of it you can look out for. And that all results in a lot of strategic discussions stemming from what you’re hearing is happening.

That is the most critical part that builders with research specialty, so the UX researchers can aid the teams with. And then you have to automatically start using- The new powers that have been bestowed upon you with AI. So it is not only the speed. I have always been a very tidy researcher in a way, because like I study human cognitive interaction as a focus area in the communication faculty, and my first bosses were my thesis consultant and my dean of the faculty.

They created this UX research and information architecture consultancy company, and if you’re working with information architecture specialists, foldering, naming conventions, that celebrates the absolute criticality. So I could never shake that off, and I have been always very tidy in terms of the- it stays with you.

Yeah. And then so for years I was like, “I have the perfect nomenclature,” like the naming conventions, foldering systems, and people are like, “You’re spending too much time on that. What is it gonna be useful for?” I’m like, “It’s useful for my mind, and I know that it is not the most efficient thing to do, but once it…

you need it, it is helpful.” That was my only explanation, and people were like, “Okay, you do you.” And all of a sudden, the moment Claude Code arrived and it was like, “I can understand your context from your foldering systems,” I was like, “Let’s go.” And now all of a sudden, I am able to tap into all these past researches so fast to create the basis of a new research plan, and I do a research and I combine it with the past findings, and I’ll create a repository.

Uh, my stakeholders are just amazed by the velocity and the depth of the research that you can provide. And then it becomes, for example, new ways of consumption. I haven’t been writing reports a lot. So what I do instead is I upload my research plan as the full context and all the past documentation that I had about the research onto NotebookLM.

Then I start dow- uploading all the anonymized transcripts as well, and then give like this 3,000, 10,000 character prompt explaining what I want it to generate. Then I share it with the entirety of the team and the leadership where they can go in because as I said, these are like foundational researches, right?

So when you ask an engineer, “How is your workflow evolving?” And they talk about whatever, but you did that 50 times, you can be like, “Okay, this is a repository for you,” where it’s a chat-based interface. You go in and you ask your questions, then 20 different PMs, 50 different designers, the entire leadership team can come up with different questions in their mind, and they don’t even have to read a 70-page report or a executive summary that doesn’t touch upon anything substantial.

But instead go in, type their questions, get some answers. It always links them to if you want to actually validate this information or learn more about it, reach out to Dennis, then boom, perfectly fine step established. Or make your anonymized Claude code repo- repositories that you created on your folders available to them.

So everyone is using Claude code, they can just parse it and go through it as well. So all these new forms of Self-serve usage of research enables teams, particularly in that discovery and definition phases, and that is so empowering, and that is how it is supposed to be. Then once you’re able to do that, you become that builder with the research specialty because every other moment that you save, you can go into improving yourself in the other workflows, which for me has been coding.

The more time I am able to save from my research duties, I try to establish new workflows that involve the coding part of it as well, so that I can build the things for myself or speed up processes even further.

[00:22:16] Petra Rajkov: We’ll be right back after a short break with a commercial message from our sponsors. Hello, UXR Geeks.

This is Petra speaking. You might already know that this podcast is brought to you by UXtweak, a truly all-in-one UX research tool. UXtweak handles everything from concept testing to usability checks, both moderated and unmoderated. It also takes care of the analysis with no messy spreadsheets, no frustrating dashboards, or endless reports, just clear insights ready to share.

So if you are curious, go to uxtweak.com website and start for free. No credit card and no strings attached.

There is that second danger that you touched on, that a lot of researchers are… What’s the worry in the industry that with this faster system, there could be a danger to sacrificing the integrity of research or to not having so much depth in research. And you have talked about your specific example, that you have this foldering system and you were able to feed the machine with all this research that you have done before.

Is there anything else that the researchers can do or the product teams can do to ensure that the research is not an afterthought or it’s not just something superficial?

[00:23:49] Deniz Kartepe: I think similar challenges were being felt by developers, right? Very much they were like, “You cannot trust AI-generated code,” and so on.

And the voices started to slowly die down as the new models came up. This was a very- Rapid advancement. So like by, I mean a million context window Opus on Claude code, you are able to get pretty good, decent results. I say this as someone that is coming from a pretty heavy background, like information architecture and like hardcore usability testing before it was even called UX research.

Run by academics, we had like pretty rigorous approach to how research was done. For example, I would never have imagined that a research interview could have been done online, because for me it was a big no. And then COVID hit. I was like, “Of course, it’s gonna be in a lab-based setting. What do you mean I will dial in and then talk to users?”

But then that was happening, and then the similar change is happening now. So yeah, developers were saying it’s not trustable, and then all of a sudden it got so good that it could be trustable. And plus, as a researcher, you need to read all the outputs. So no matter what AI generates, I have done every single interview myself, and I think I’m a pretty good note-taker and I have a good memory.

So I can take a look at what it is doing, and I have seen this evolution over time. Like I have been doing this for three, four years now, especially the AI supported workflows, and it got really, really good. And now with one output generated from it, speeding up my workflows by five hours, then maybe I can spend one hour to make sure that it is the right thing.

But nevertheless, it like depth-wise, quality-wise becomes tremendous. Especially I think it is like garbage in, garbage out. I think the value that my previous workflows of having extremely well-documented documentation of the research plan, every single meeting that I had with my stakeholders, I would also store it, my opinion pieces about the matter.

So any research folder I have has seven different subfolders. It’s not only session transcripts or session notes or highlights, but rather my past thoughts on the matter, my meeting notes, a very detailed research plan, and even more detailed research plan background about how we brought it to be. So if they’re all very nicely structured and the AI is able to parse them correctly, the quality is still there and you are still the end person who is doing the last check and the last edits on it.

So I would say from my perspective at least, maybe because it is because of the guardrails that I had, the quality of the output that I’m generating hasn’t really dropped. In fact, I think it got better because I basically have a very capable assistant that is able to give me at least a version 0.9 of something that I can bring to 1.0 myself And the real worry that I have there is like garbage in, garbage out.

It’s about knowing how to prompt right, knowing how to feed context right, and always keeping an eye on what is being generated. But probably with the speed and development of things, we are not going to be going back from this. This is here to stay and it’s extremely empowering. And also what I’m curious about is, like when I started more than a decade ago, it wasn’t a very hype job.

There wasn’t that many people doing it, but it evolved alongside these processes anyway. We actually like the person, the people who started on it before this full AI or sped up era, I wonder how many we are. Like now we are looking at like graduate programs for Miro, for example, people who are starting to– starting their careers only now, they are fully AI native anyway.

Mm-hmm. And is there going to be a huge divide between how two different upbringings, backgrounds, timelines are doing things completely differently? Because it’s going to be a huge gap if we sustain it.

[00:27:40] Petra Rajkov: Yeah. And it’s hard to predict what’s going to happen next because it’s just so dynamic- Yeah, yeah … and all the changes, and you are just reacting what’s coming next.

Yeah. But you have built all this on your expertise, years of practicing research and- Yeah … doing it, and now you have these new researchers coming, and they are studying this, and they are using it from the beginning with AI. And you said that you use AI as an assistant. So where’s that balance to what are the things that what you have to do on your own and where AI can assist you?

Yeah. And you set up the research plan, but you do those interviews. You… When you are doing the analysis, you know what to look out for, what is good enough and what is not good enough. And then you can experiment with the formats that reach the stakeholders, so you make sure that the right insights reach the right people.

So where I’m going with this, what I’m wondering about, how is it changing what the researchers need to know? What are the skills that they need to have, and what is, like, non-negotiable for them?

[00:28:49] Deniz Kartepe: I think it is extreme flexibility and adaptability. Like, the change is constant. Everything is changing, and everything is changing at this velocity that is breakneck speed.

And you gotta be able to keep up with it because if we do not adapt, it’s the new ecosystem of doing work is not going to be very tolerant of people who stay behind. So we constantly have to reimagine, look, like, a couple months ahead about what our role can be because otherwise… Yeah, like, there was a small hype about AI is going to take our jobs.

And you’re like, well, AI is not taking anyone’s jobs, but everyone collectively is evolving their job descriptions and what it means to work together. And people work in nodes. You are not a single unit that is completely isolated, in- independent from the rest. So it is going to be- Absolutely critical for us to be able to change how we are doing this.

We are already changing it. We just have to embrace. Therefore, flexibility and adaptability are absolute musts. And now when it comes to, like, new generation learning, the best example is me learning to code. I haven’t done it before. I tried dabbling with it. It was too high a wall for me at my working time, and I just didn’t have the energy to become a coder 10 years ago, eight years ago, six years ago.

I tried, and then I was like, “Okay, I just don’t have the head space for this because I need to do my actual work.” Then now in the last two weeks, so basically what I did was… That’s the reason why I think it is applicable because then I am the graduate person, right? Because I am just jumping into something without no knowledge.

Just the audacity to be comfortable with breaking things and rapidly learning about them. And then we, we were building this internal research discovery tool for Miro to research it, as I’ve said. Michael has some background in coding. I don’t have any at all. But what I did was I first built a skill on Claude, which was called Learn.

It was a Socratic coding tutor. So after learning a million tokens on changing a database schema, I would be like, “Start that workflow,” and then it would turn into the role of a teacher where it would first explain what we have just observed, what we did together, and then it would be like, “Okay, there is now…

When we were creating this schema, we used something called Zrod, and it was doing this.” And I would show a code block. This is the before, and now we fixed it into this, the after. Why do you think we did that? It would start with that question, and then it would allow me to type in Claude code, like in terminal.

So it becomes like this quizzing teacher that goes deeper and deeper into these flows to teach you what you just did. So that was a skill that I built for myself so that I can speed up my learning, and it has been tremendously impactful. If you want, you are so enabled right now to onboard yourself into different verticals in the domain that you are working in.

And I think that same applies for the people who are just starting. So that’s the reason why if someone is learning research right now, I have to run maybe over 1,000 in-person face-to-face interviews to come to where I was, but I don’t think It is the same situation right now for someone starting. We had internet compared to people like who started 20 years ago to have access to any tutorial or any book or any source that we needed, but still we needed to find it, we need to consume it, then get back to work.

Right now, any work you do is perfectly conceptualized with the help of an AI that can give you in-moment feedback about what you are doing, and always be available very contextually to provide you the information that you need to improve. So also that adaptation onboarding process is super fast as well.

[00:32:44] Petra Rajkov: Amazing. Thank you for that. And if we circle back just really quick to our initial talk about the product development lifecycle changing and how the engineer’s role is changing, are there any other insights, uh, that you have learned that we haven’t touched on?

[00:33:01] Deniz Kartepe: My engineer, head of engineering workflows, Tash, an amazing guy, he also used that in a talk, which was a very strong statement coming from a couple of users that then we evolved into a slogan, which was these are technical architects, the people who are responsible for building the technical architecture of companies.

They were saying like there is this 52 to 1 ratio. So for 52 weeks we are aligning, and now we are coding for one week now. All that alignment, all those weeks of understanding each other, understanding the limitations, possibilities, and creating that consensus for just one week of coding.

So what is a shift in the entire world where this has turned Upside down, on its head, where what matters now is how we collaborate, how we communicate, how we create a consensus, how do we make sure that we understand each other and the domain that we are operating in is becoming so valuable, and that brings a whole set of new skill sets, new expectations from roles, and that is how the product development life cycle is evolving.

I think that is the most critical takeaway, because once we accept that as the thing that is happening, then we have to completely rethink what services these enterprise B2B SaaS companies are providing, uh, to their users and what they look like. Another part, of course, is going to be Miro was a very collaborative tool, right?

By nature. It was a collaborative whiteboard, it turned into innovation workspace, and I was always working in the communications and presence sub stream, where I was responsible for how our collaboration tools, collaboration suite worked. And I was so focused on human to human collaboration and the need for a 2D spatial contextualization, like this canvas space where you get together, put things, and then collaborate, communicate in between and build something together.

And that was human to human collaboration. Now, there’s a new interesting paradigm. There are multiple types of collaboration. It is the human to human collaboration, then you have the human to agent collaboration, then you have human to agent collaboration. It is like in computer games there is this concept of PVPVE, so player versus environment engine.

So it is us and a computer as well. So now there is human to agent collaboration. Then the next one is also agent to agent collaboration. So my agent talking to your agent and getting something done for us. So all these new paradigms are coming into play, so then we completely rethink what communication means, what collaboration means, and if this paradigm is just being established, what does it look like?

Because that is probably going to be like the first version of a car that was built, like Ford Model T And then we want to see what roads are going to be required for the Toyota Corolla of 2010, you know? Because it’s going to be happening so fast, and a lot of indie developers are building some solutions, but probably enterprise B2B companies will be scrambling in this space to establish what is right, how it is going to work, et cetera.

And then coming back to it, that is where researchers come in. That is where they can still build, but, uh, as I’ve said, builders with research specialties, they are still researchers. And they will be so critical in defining that problem and solution space together with those other builders in making sure that we are getting the right thing.

Otherwise, we come back to that initial problem, which is a lot of people will just build things and throw it out there. It, it just won’t be the right thing because it is so cheap and easy and fast to do things bad.

[00:36:42] Petra Rajkov: So researchers are here to stay, but our role is evolving and one in the product is becoming a builder, and there are these more skills that we can learn, not need, but we can learn. Right now the world is our oyster, and we can explore other specialties. So if, just to wrap it up, if there was one thing that you would recommend our listeners to start doing even tomorrow, how they can change their day-to-day work, what would it be?

[00:37:11] Deniz Kartepe: Unbox and build and break things. I’m not talking about build a product. Build up a workflow for yourself, uh, with the help of AI, what it can look like, how it can be improved, and then break it, and then build it again. And get comfortable with breaking, making mistakes, it not working, because then the realization becomes Trying something and failing is so cheap and doable right now because of the, all the capabilities that AI is giving us.

You can think of 10 different concepts for how you can run research within your company, experiment with them, fail with them, get feedback on them with the help of AI and with the help of your colleagues. You’re just able to do that so fast, for so cheap, for such a low cost of money, effort, time, cognitive load, that then the more you do that, the more you break, the more you fail, the more you trial, the more you get comfortable with it.

And the moment you can get comfortable with it and pass a certain threshold, it turns on a switch in your brain that allows you to keep on experimenting, keep on being flexible, adaptable, and then be more comfortable with this new reality. Because there is this new paradigm right now in how we are doing work that is being established throughout the world.

And of course, it feels a bit daunting and stressful at first to make that jump, but the more you do this, that, ah, there’s just a little ladder there that allows you to ease into it. And I think, yeah, to be flexible and adaptable, giving yourself the room to play, experiment, and learn very rapidly, while also creating value, because maybe the thing that you will just experiment within your first day is going to be something that’s just going to be the next big thing for you, or for your company, or for the whole world.

I don’t know. But it is exciting, and it is really enabling, and art of the possible, like the moment you jump into it, it’s just exciting. You get really comfortable with it real fast, and then the entire thing becomes so much fun. We also need to learn how to enjoy this new paradigm as researchers, you know?

[00:39:15] Petra Rajkov: Enjoy the process, yeah.

[00:39:17] Deniz Kartepe: Yeah, yeah.

[00:39:17] Petra Rajkov: Itshould be fun. Yeah. So now is the time to experiment. Thank you, Deniz, for sharing that, for coming- Thanks … to the show, and good luck in the future. We cannot wait to see what’s next.

[00:39:27] Deniz Kartepe: Yeah, exactly. A lot of things are coming up really fast, and I cannot wait to share them in the near future.

[00:39:37] Petra Rajkov: Thank you for listening to UXR Geeks. If you enjoyed this episode, please follow our podcast and share it with your friends and colleagues. Your support is really what keeps us going. 

If you have any tips on fantastic speakers from across the globe, feedback, or any questions, we would love to hear from you, so reach out to geekspodcast@uxtweak.com.

Special thanks goes to my colleagues, to our podcast producer, Ekaterina Novikova, our social media specialist, Daria Krasovskaya, and our audio specialist, Melissa Danisova.

And to all of you, thank you for tuning in.

💡 This podcast was brought to you by UXtweak, an all-in-one UX research tool.