#137 Generative UI: Lessons from Google Labs Disco

Generative AI is quietly changing how digital products work. And Google Labs Experimental Browser Disco is a perfect example for that. Hello and welcome back to the future of UX. My name is Patricia and today I want to talk about Generative UI. Using Google Disco and its GEN-TAPS feature as a concrete example, but really zooming out to what this means for UX thinking. Not what buttons look like, but how interfaces are created in the first place. So in this episode of the Future of UX, I explore what happens when interfaces are no longer fixed screens, but generated on the fly based on user intent, context and goals. Using this go and the GEN-TAPS feature as a lens, we talk about how browsing is shifting from search first to goal first, how UI becomes a temporary situational response rather than the static artifact. And of course, my trust, transparency and responsibility become the core UX challenges in generative systems. This is not a tool review or something like that or a super hype episode, but I would rather describe it as a UX first perspective on what generative UI signals for designers, for product teams, and basically anyone shaping digital experiences. So if you work on UX and product and design strategy, this episode will help you understand what's actually changing and why it matters. I've been noticing something lately. A lot of conversations around AI and design still sound like "oh, a tool should I learn next?" or "what feature will replace this workflow?" But when I look at some of the newer experiments, especially around generative UI, it feels like we are asking the wrong questions. Or maybe not the wrong questions, but not the right questions, maybe. to frame it like that. Because AI is not only about the tools, but I think what's even more exciting for me is the interfaces that are starting to behave differently. So definitely like the UX perspective. And one experiment that really made that obvious to me is the Google Labs Disco. So what is this goal? Disco is an experimental invite only browser experience from Google Labs. It's built on ChromeMilm, but it's not meant to replace Chrome. At least this is what they say. It's rather a playground for rethinking browsing. And I think it's just pretty interesting because there were a lot of different browser experiences. I also talked very in depth about Atlas, the browser from OpenAI with TGPT. And now Google released their own generative AI browser experience. It's still experimental. This one is called Disco. And this is even more interesting, I think than the other browsers. Because it does not just show you web pages. It's not only about the web pages, but Disco takes it much further, I would say. Of course you can open tabs, you have chat input, and it's all about your current goal. And with all that in mind, it generates a custom interface, an interactive interface on the fly for you. Which is not a template or a saved layout. No, it's a new I that exists because of what you're trying to do right now. And that's where things get really, really interesting. So we will dive into three main topics that I find super interesting. The first will be that interfaces are shifting from the structure first to intent first. Then we will talk about the behavior, and then in the end we will talk about the responsibility. So let's dive into the structure first to intent first. This is actually a phrase that I got from Jacob Nielsen. I can highly recommend his newsletter. He always shares super insightful stuff. I will link this specific newsletter where he talks about generative AI in the show notes in case you want to check it out. So one thing that this code makes very clear is that the starting point has changed. Traditionally interfaces are structure first. So that we as designers, we design the navigation, we design the layout, we design the hierarchies, and then users adapt basically their own behavior. So they learn how to use the web page to what we designed. And with generative UI, it all lips. You or the user doesn't adapt to the interface, but the interface adapts to your intent. So for example, you say, "I want to play a plan, plan a trip," or you say, "I want to learn this specific topic." "Maybe I want to learn about AI for designers," or maybe I want to compare different options, maybe different shoes that I can use to block my docs. And instead of navigating the system really responds by assembling an interface that fits that intent. And I think what's important here is that the UI is no longer the primary object, but the goal is, and this really reminded me so much about the whole concept of jobs to be done. And I know that this is generally a concept that is also used in a lot of like your acts practices. The idea of the concept jobs to be done, there's also a wonderful book that I will link in the show notes, if we free to check it out. It's basically that you think about what is the goal that the user wants to achieve. And then what does the user needs to do, or get done to get to the goal. For example, hang a painting on the wall. Okay, this is the goal that I want to do. Okay, what do I need to do actually to get this done? I need a hammer, I need a nail, I need this painting, you know, and do all the specific steps. And then the past, we were thinking about, okay, where should we place the hammer, that the user can grab it, where can we place a nail, what kind of other instruments does the user maybe needs, maybe something to measure, whatever, right, maybe a letter, something like that to climb up. And now with this new approach, the intent first, you think about what does the goal, so how could maybe everything basically happen, or how could we maybe create something where we don't even have a hammer, where we maybe don't have a nail, but where we have a different structure, just thinking about something spontaneously, this could be having maybe a certain glue, that you can pin on your painting and place it basically on the wall, or let it hangs itself, something like that. That it's all about like what does the goal, what does the goal, the user explains, okay, my goal is I want to hang this painting, and then things happen in the background maybe autonomously. I hope I explain it well, I'm not sure if this example was so, what was so, doesn't make so much sense, but I think you get the idea. So from your ex perspective, this means that we're designing less for the path and much more for the outcomes. And the interface becomes this temporary manifestation of the intent, and not this permanent structures, you just have to learn. And I think this is a fundamental mental arm shift, super, super exciting. Okay, that also brings us to the second important topic, because you I become a behavior of the system, and not this fixed artifact. Another thing that really stands out in this code is how the UI feels less like this product and more like a behavior. So the interface isn't just there until you need it. And it don't need to stay once the task is done, you know, it can be completely vanished because you don't need it anymore, because you complete this task and no one needs the structure in the background anymore. And this aligns with a broader generative UI idea. UI is no longer something that we ship, but it's something the system does on itself autonomously. So the system looks at the context, it needs a behavior, and they're available later, and then decides on how to present itself. For designers, or for us, I think that this means that the work moves away from designing static artifacts, designing static interfaces, designing static UI as well. And instead, we are moving into a much more conceptual space. We're thinking about what a certain boundaries, what our guard, what our constraints, what our priorities, what is logic behind it. [BLANK_AUDIO] are acceptable variations. So we talk about, okay, when there's a generative, I, what are some constraints? If thinking about, okay, what shouldn't we place there? Are there any topics that we don't want users to create apps on the go on? Are there any priorities? Maybe thinking about accessibility? How big generally should the buttons be? Do we have any guardrails, any guidelines, basically? Not thinking about a static design library, but more something that's a bit like floating, moving around. And we are especially defining the decision space in which you I can emerge. This sounds very abstract, but I think that this is also something that we need to document at some point. I'm not sure how this will look like. This is like a written document. If this is something visual, for example, if this is something coded, where we have basically all the rules explained, or if this is a second moving document that also recreates itself, I'm, I don't, I don't know yet. But what I know is that instead of asking, what does this screen look like? We will ask what should the system consider before it shows anything at all? There's a much more abstract, but also more powerful role for you. And I'm super excited about it. I can't wait the day when we don't need to design interfaces anymore, but more think about the strategy and the structure behind it. Because this is where design, I think also really shines. And designers are so to do people who really, I think have the skills to understand these nuances. And yeah, so I think pretty exciting. I'm also very curious to hear your thoughts about the topic of the narrative UI of disco. And of course, you can always share comments underneath the podcast episode. And this is possible. And Spotify, I would love to hear your thoughts as well. Or another option is I'm going to link a LinkedIn post where I share it a bit about my thoughts as well. So basically this podcast episode, but a little bit condensed. And if you have any thoughts on it, please share it. I am super curious to hear your thoughts about it. And I think it's a great place also where we should like connect and discuss and learn from each other as well. Okay, let's move to the third insight or the third idea that I that I want to share with you. And this is that generative UI makes their responsibility that design has visible again. Because now, yeah, yes, where generative UI gets a little bit uncomfortable. But also in a, I think in a good way. Because when you I, oh, yeah, when interfaces when you I are generated, someone still has to be accountable for the decisions behind it. We can say like, oh, the AI all did that. I think that would be a bit easy. So even interface is confusing misleading or biased. You can just say, oh, the AI did it. And this got tries to build trust by linking every generated element back to its source. And I think this is a very good start, because that highlights a bigger UX challenge users are no longer interacting with raw information. They are interacting with AI interpretation of information, which means that your X has to actively design for transparency and traceability and recovery. And those are not new rules. And if you design an AI feature, traceability, transparency, recovery are super important rules or guidelines. So how does the user understand why the AI looks the way it is? Or how can they maybe also challenge it, maybe give feedback? Can they correct it? And generative UI doesn't reduce the responsibility of your X. I would even say that it amplifies it. So when I look at this goal, I don't really see a broader experiment with a lot of other people seeing I see a signal that interfaces are becoming adaptive, contextual and deeply system driven. And that pushes your X designers into a different role less about like crafting the perfect screens. And I mean, we are moving away from that, I think for the last years anyway. And moving more into like shaping how the systems behave towards humans. So we become a lot more focused on intent and not so much the interaction itself. And this also feels like your X is growing up quite a bit. For me, personally, the whole topic of generative UI is an absolutely fascinating topic. I'm also working with different clients or asked here the year before I work with different clients on this topic. How you can really use AI to create custom interfaces for users. So I am pretty deep into the topic. I did a lot of research on it. I tried a lot of things out myself. So I'm I know that there is quite a lot happening. And I think a lot of people don't really see that or think like, oh, this is totally sci-fi, but it's not. We are seeing those trends, we are seeing those only single signals. And this doesn't mean that everything's changing tomorrow. And we have only generative UI interfaces, 100% not. But we are seeing these tendencies and for me, as someone who's so excited and interested in the future and how things change, this is something. Yeah, that I want to emphasize for you as a designer, keep an eye on it. Because this will change a lot also how we work. So learn about it, try it out. If you read about it, get a little bit excited about it as well, but also curious and cautious. Curious and cautious, I think is the best mindset ever. Because hype is definitely not helping anyone. Yeah. Okay, so to summarize a little bit, I think that maybe the real question is not, yeah, will generative UI replace all the designers because the screens are now designing themselves. But I think it's more about, have you ready to design systems that decide how they show up? Because that's what the future that this goes quietly pointing towards. And whether this exact product chips or not, that shift is already happening. What an exciting world we're living in, right? And yeah, super curious to hear your thoughts as well. I mentioned that I'm linked in post. You'll be to share your thoughts there. So before I wrap up, I just want to say a huge thank you. I really hope that you enjoyed not only this podcast episode, but also all of the podcasts episode that I shared this year. I hope that it really gave you a few, a few new thoughts that you could take, yeah, into your new week, into your design life, and really got some insights. Now we're going to take a short mini winter break. I think it's only one week. So wherever you are, I hope that you get to slow down a little bit and sometime with the people you love, or also maybe some time alone to really connect, think about your goals and the next year and where you would like to focus yourself. I think it's a super helpful once in a while. I wish you all very Merry Christmas and all the best worth a days ahead. Thank you so much for an entire year of support, of curiosity, of all the thoughtful messages and what also learning together. For me, this has been a very, very exciting year. You know, my wonderful listeners, you probably know that I became a mom this year. So a lot of things has also changed for me. I might also do a little reflection session about that at some point. So I'm very, very thankful for your support. And for being part of this wonderful community, where we all support each other, where we all learn from each other and right to make the future battle with design, I think this is our our mission to learn and be the design and the room who really speaks up or knows what to ask who is curious and cautious about the future. So thank you so much for your support. It generally means so much. And now I wish you a very Merry Christmas, hopefully some calm and relaxing days ahead and see you very, very soon. My wonderful friends, thank you.