The Impact of A.I. in UX with Jascha Goltermann

This is Hatch Design Talks, where we chat with the speakers of the upcoming Hatch conference happening in Berlin and online on October 6th. In our last episode of the season, we speak with Yasha Golterman, Design Manager at Booking.com. He's joining Hatch to talk about the learnings building Booking's new AI-powered trip planner, and he's sharing his views on how the big leap in the technology might change the way design and tech teams works.

Let's begin. Hi, Yasha. Thank you so much for joining me today on this podcast that we do to tell the people that are coming to Hatch Conference or attending online. A little bit about the speakers.

How are you doing today? I'm doing very well. Thank you for having me. Obviously, I'm very excited about the conference itself, but also talking to you today.

Nice. Yeah, you're doing a lot of stuff at Hatch. You're both facilitating a session and giving a keynote. But before we start, why don't you tell me a little bit about your career?

How did you end up in Booking.com? Okay. My career spans a little over 10 years now.

So obviously it's gone through a lot of changes and different kinds of roles and different kinds of responsibilities. I'm now a design manager at Booking.com, which means that I manage a team of designers. They work on strategic efforts.

I lead a vision and high priority projects. My focus is on the way how people search and find accommodation in Booking.com. So really the heart of the product.

And the role itself is very much focused on three things, really people, product and organization. So I support designers, I support their growth and careers. I also work with tech, marketing, product and machine learning leaders to drive projects. And I work on improving the design craft itself and the design management craft themselves.

Before I joined Booking, I was in many different roles. Obviously, a lot of things have changed in the field itself. Right before I joined Booking, I was actually in an individual contributor role. So I did not really necessarily have a linear career path, I would say.

And I spent about half of the last 10 years in different design leadership roles, some of them involving managing others, just leading other designers, and obviously a lot of classical individual contributor roles in companies of different sides. And how was actually to make that transition between like individual contributor into a manager? What do you think were the challenges that you had doing that? So I think the challenge is really pretty much the same for most people who go from individual contributor to manager.

It's the shift in focus. So obviously you shift your focus a little bit from the tactical to the strategic and from details to bigger picture to planning. But I think there is a much more core part which tends to receive less focus, even though it's actually the key change, which is shifting your focus from product to people. No matter if you're a designer or a developer, or even no matter if you're an IC or a manager, the craft itself is not the difficult part.

It's the human to human interaction. So how do you, for example, have difficult conversations? How do you guide people? or how do you inspire them to go beyond their comfort zone?

So these kinds of things are more difficult to prepare for than going from tactical to strategic, right? This is something that you can learn a lot more from soaking up knowledge online or something. So the main transitional challenge, I would say, is becoming somebody who works on people instead of working on classical hands-on design deliverable. And how did you handle that personally?

Did you have any specific, did you take courses? You just learned by doing, did you have good mentors inside the company? Yes, you do need to have mentors or maybe not so much mentors, but you need to have people that you feel that you can learn something from that are in a role like that. And it doesn't necessarily have design leader.

It can also be, it can also be a, a people manager and in completely different craft that you want to draw experience from. So it's important to basically look into how are other people, managing situations and trying to learn from that. You can try to learn by reading resources or taking courses. I'm not saying I wouldn't recommend it.

It's also something that obviously looked into, especially once I was in those role, I wanted to make sure that I don miss one of the essentials of the role but it not as valuable I think as the lessons you learn from the situations that you placed in when you in those roles And obviously that means that you can only really grow into the role once you in the role but what you can do beforehand is prepare by exposing yourself to certain thinking and to certain situations So yeah, I think you have to have an understanding of yourself and how you would deal with people in situations.

You have to understand the difference between managing and leading. You have to have sort of leadership principles and obviously those will. change and those will sharpen over the course of your career then as a manager. But you have to basically know what it means for you to lead people.

And it's good to build that understanding before you're placed in a role like that. That's a good segue to your actual masterclass at Hatch. It's called Design Manager Static Kit. Is there something like things that I wish somebody had told you once you transitioned into management how did the idea for the master class come to be um yes essentially it's that for a transition into management it's not really like you cannot put a generalized roadmap for someone right there's not really like a blueprint follow these things and then you will be set as a manager also because this role itself can mean different things the role itself means different things in different places that's one challenge and there's no generalized way this is what the role is, and this is how you prepare for it.

But there's a ton of learning still, both from my personal journey, but also from being on the sort of hiring side, being on the side of growing CEOs, being on the side of growing leads, being on the side of hiring managers. So there's a ton of learnings in what are the kinds of skills? What is the kind of expertise? What are the kind of approaches that typically companies will ask this person to be experienced?

And how do you spot sort of your opportunities to grow into those skills and gain this kind of experience? That's pretty cool. Maybe you can turn into an article as well sometimes for everyone who cannot attend it. And of course, very exciting.

You're giving the first AI-related talk of the day at Hatch based on your learnings and having to concept and implement some AI power features. And this new trip planner, Booking.com has launched very recently. I'm interested in how you tackle having to work with such a challenge.

It's the first time that you had to do something like this because you've managed teams doing products, but suddenly there's this whole new paradigm of what we're actually building and we don't have much reference to the back of our own team, but also in general in the world. So how did you get it going? Luckily, it wasn't the first, let's say, project working with ML technology per se. right?

That's not something that was necessarily new because Booking is quite a mature organization set up around machine learning. So there's a machine learning hub in Tel Aviv and there's machine learning scientists and principal engineers and all these kinds of things. So working with parts of the expertise at least was not new, but then of course, LLMs and generative AI is a new technology that, yeah, it's helpful to have experts that are knowledgeable in the field of machine learning so they can quickly help understand this new technology.

But that new technology itself changes quite a bit. And I think it's taken everybody a bit by surprise. Like there was this time from when ChatGPT was released and the next couple of weeks on a daily basis, you would see new improvements in the space and new things that people were building with it and new ways that people were getting value from it and new technologies being released. So it was at this time where if you just log on to Twitter, you would see almost every day somebody posting, hey, this new record basically has been broken.

It was new to everyone, including me, including booking, and probably including all those machine learning scientists that were working with different types of ML technology in the years prior to that. So how did we do it? And it's a very much cross-functional effort. I can stress that enough.

Obviously, designers are always working with tech and product and other crafts. But in this thing in particular, you really have to learn about this technology together. And you really have to figure out ways to collaborate and communicate about it together. So designers have to learn vocabulary that machine learning scientists have already been using.

Same with product managers. They were also in this position suddenly where they had to learn new vocabulary. And then you have to decide on okay what parts of this new technology are actually part of which craft So that just one one example Prompt engineering that sounds that would be an engineer job right That sounds like something tiny. Yeah, it makes sense for a machine learning engineer to engineer the prompts.

And it does actually happen in the tools that machine learning engineers use because it is part of the code, but it's the tool, the biggest lever you have to really impact the experience because the way how you engineer those prompts will vary, will have the direct impact on what the generative AI, what the LLM essentially creates as an output. So it is something where it is essential, it is crucial for UX roles and not just designers, also researchers and also writers to be involved and have a say in how those prompts are engineered.

So it's a bit of this new space where, let's say new responsibilities, new skill sets are required. And it's not clearly defined who does what and how it works. So prompt engineering is one thing where basically designers need to work alongside machine learning engineers or maybe in other companies generally, which is tech, with developers to basically craft those prompts and then find ways to understand how they would impact the experience. And then a new thing is there is a lot less that is actually under your control when you work with this kind of technology.

So typically as a designer, you design something and you know that what the interaction is going to be like, what the experience is going to be like, because we designed it. But with Gen AI, you don't quite exactly know how it turns out. And even if you go to chat GPT and you put in the same prompt twice, you will get a different answer the second time than you got the first time. Even the companies building this type of technology don't know 100% all the details of how it generates the answer and how it works.

So you have to be able to deal a little bit with that ambiguity, but you can design how you deal with it. So you can design failovers. You can design how to steer it in the direction that you want. You can design how to avoid situations that would lead to a less optimal experience.

So that's new that you have to look into these things. And then another thing that is new for designers is you really have to design conversations more than you have to design interfaces or more than you have to design even. So you use tools that you would use or techniques and approaches that you would use to design user journeys, user flows, but you apply it to designing conversations. So, for example, you want to say, hey, we want this.

In this case, for Booking.com, obviously it is an AI trip planner. So it's essentially a virtual travel agent. You want this AI to behave like a travel agent and to respond like a travel agent.

How do you design how it responds to certain conversations? And it's not as simple as just connecting this question will lead to this answer. Because it's completely open, like users could in any type of question. So yeah, those are exciting new things to consider when designing with this type of technology.

And the key to it is working very closely with the engineers that are then actually implementing the code that will build this tool that you're working on with this technology. I think what's also really interesting to me is that kind of the expectation on the other side has suddenly changed for two things. And one thing when you're already saying for a product, hey, this is like AI powered or something like that, people already have a different expectation and at the same time don't know what the output is, but you're working with a whole backlog of things that one expects when one is in booking.

com. So, you know, there's part of the essence of what you're providing as a result that can really drift too much. But at the same time, as you said, the unpredictability of it and the kind of the new type of interaction that one has really changes things. So I think that's a really interesting thing.

And even to see, will these processes that we've been using for all the other interaction design, will they still apply? Will they have to be updated to actually be more permissive for certain processes? And my next question comes exactly around this. Like you think that all these new technical possibilities that we have with all these models and with all these possibilities start making a requirement for designers to learn new skills on how to work in these types of projects.

Because what I believe is that if I learned how to implement user journey, it doesn't necessarily mean that I can translate that from like a Miro template into this new thing So what your opinion on that Yeah no absolutely So there are a bunch of skills that will help you adapt to the new skills that are required when working with this kind of technology. That's more what I was saying. Not so much that we already have those skills that we already have the expertise, but more that we have a lot of very relevant skills and expertise as designers and actually in different UX roles that will help us position ourselves as key strategic roles in this new era, really, where different AI technology is becoming more core to creating experiences.

And it will become more core to creating experiences because it also becomes a lot more accessible and a lot more affordable. And startups of pretty much any size can afford to incorporate external ML capabilities into their products now. Something that previously was something that the big tech companies could afford to build an organization around themselves. But right now you can use software as a service for Gen AI capabilities and just build your product around it at a relatively low cost and relatively low risk.

So more and more companies are going to actually use that. And as a designer, you won't really be able to completely ignore the field. So, yes, I do believe that it's becoming more and more important for designers to at least learn about the technology and learn about the different ways how it can be used to improve the experience. And that's really where then also the key value of our role lies.

obviously companies are looking into using this technology, but they're using this technology or they want to use this technology to ultimately improve the experience because that drives value for the user, which in turn drives value for their business. So how do you get value for the user out of it? That's where UX is crucial. And that's why I think design roles need to position themselves as the key roles to involve in projects working with AI.

And yeah, they need to upskill essentially or expand a little bit their skill set and expertise into this field. And has there been any moments building these new features powered by AI where you were like, okay, this really changes everything. Some kind of breakthrough moments. I mentioned costs already, obviously, which is an important part of it.

But from the design side, was there anything specific that you found specifically surprising or interesting? One thing that I mentioned already is the unpredictability of it. That is a big, like you cannot really overestimate this change, but also it can do so many things out of the box, understand user intent, personalized experiences that it offers a wealth of ways to improve an experience just by tapping into an existing. And so I think it's the vast amount of things that you can do with it.

That is going to revolutionize a lot of products, but then there's also the unpredictability of working with it. That's going to require us to adapt how we work with it. And we won't get to a point where we can fully steer it. Like you won't be just able to say, always work and respond as a travel agent.

And then everything just works out. you can still trick a travel agent into giving you inappropriate travel advice. And there's no 100% fail-safe way to avoid that. So these things, working with this kind of ambiguity, working with this kind of unpredictability, it's just something that we need to get better at and manage around it.

Okay, I won't spoil it much more about the topic. For everything else, people can watch your talk, either in person or online. Are you excited about visiting in Berlin again? And what are you most looking forward to do here besides catch?

Yes, obviously super excited. The thing I'm looking forward to most, obviously, is to connect with people. So looking forward to all these talks between the actual talks at the conference. And besides that, I won't really have so much time.

So I'm mostly planning to see family. Okay, nice. Thank you so much for this. This is probably one of the last podcasts that's airing.

So for everyone who's listening, we'll see you this week, probably. Thank you so much. and yeah see you in a few days in berlin thank you and see you over there bye bye that's it this was the last episode of this year we'll see each other in person or online next week and remember that you're still on time to get a 15 discount to attend online using the code hatch design talks at checkout on www.hatchconference.