In this solo talk, Daniel Andor, founder of the product design studio Duran, examines whether current AI design tools (Figma Make, V0, Lovable, Plot Design, Claude, etc.) can replace a product designer. He begins by distinguishing two primary uses of design: simple execution of developer‑requested UI versus strategic, problem‑solving design that shapes product direction. For execution tasks, especially when a well‑documented design system and component library exist, AI tools can handle a substantial portion of the work. However, when design must generate solutions, craft high‑quality interfaces, manage complex user flows, or conduct nuanced customer discovery, AI falls short. Andor outlines the specific responsibilities of a product designer—research, ideation, design‑system maintenance, customer discovery, and team alignment—highlighting where AI can assist (e.g., rapid component generation, early‑stage prototyping) and where human judgment is essential (e.g., interpreting feedback, ensuring consistency, making strategic decisions). He then presents a decision framework: consider how central design is to your product’s competitive advantage, the complexity of the flows, and the quality expectations. Depending on those factors, a company might opt for an AI tool, a full‑time designer, or an embedded design studio like Duran. The video concludes with a call to action to contact his studio and subscribe for more UX content.
Chapters
Can an AI tool like Figma Make, Claude, Lovable, or v0 replace a product designer? A founder asked me exactly that recently. They've got a designer role open and they were thinking whether to even fill it. I'm Daniel. I run Duran, a product design studio, and we help growth stage SaaS teams fix the UX that's killing adoption and wasting engineering time. Honestly, the answer isn't a simple yes or no. So let me walk through how I actually think about it. So yeah, tools like Figma Make, V0, Lovable, Plot Design can definitely speed up a lot now. We work with a bunch of companies and some of them use these tools to speed up parts of their process, either on development or on the design side. But it really depends on why you need a designer. And the way I split is this. If you're using design just to implement what your developers ask for, basically a simple execution role, then yeah. AI tools can cover a good chunk of it now, especially if you already have a design system in place and it's well documented. That's actually the biggest lever. The more defined your tokens and components are, the closer the output gets to your real product. The team can throw things around in there and get a decent, sometimes close to usable outcome. The problem usually starts when you want design to be more than simple execution. Because if you're thinking about design as coming up with solutions to problems, not just drawing what's already been decided, then there are a few different functions you want to cover. And honestly, that might be too much to hand to engineers because they got their own role and usually a different skill set. So if you are thinking about that side of design, a more strategic, embedded-in-product decision side, here's roughly what we set up with the customers we work with. First, a feedback database from their customers. So we can collect all the feedback in one place and actually have an overview of what's happening, things are going, what the problems are. Then there's a decision of what the solution should even look like, because once you've defined the problem, the solution can still take very different shapes. It can look different, it can work in different ways, and then you've got to figure out the flow as well. And this is exactly the spot where AI tools like Lovable, Figma Make, Plot Design, and there are a couple of others, usually fall short. So 2 places specifically. On the craft side, they spin up hi-fi looking screens fast now. And honestly, they look decent at a glance. But the output is still generic and recognizably AI. And even with your design system thrown in, which does lift the quality a lot these days, it still drifts. It quietly creates a slightly different version of a component you already have. Break a pattern here and there, and you end up cleaning it up. For standard stuff, it gets you close. For real craft, it's still not quite dev-ready. And on the flow side, they're okay with the standardized flows and standard products. But the moment you've got real complexity or flows that are a bit less standard, that's where they fall apart. They'll make confident mistakes, sometimes get stuck in a fix-break-fix loops, throw in way more options and complexity than you actually need. So you spend a good chunk of time just trimming and correcting. Now, where do I see real value and where we actually use these tools ourselves is in the ideation phase. And this part has genuinely changed. You can now get the whole team building and testing interactive prototypes early, even of fairly complex flows before committing to a direction. That's a real shift and it's super useful. But then you still have to make that decision. You still have to create the actual designs and you still have to prep them for engineering. So it's a starting point, not the finish line. And if you go through the responsibilities one by one, it gets clearer. Research, they can help, but somebody still has to review it and make the call. Concepts and ideation, they are fast and generally useful for exploring options, but the output is usually generic. So you still need someone to push it to something that's actually good and on brand. Design system, this one they can really help with. Generate the components, the states, the variants, and the documentation. The problem is consistency. Left alone, they'll spin up a card that's almost your card. And now you've got 3 versions of it in production. So they helped build it, but you still need someone owning consistency, or you end up with component variations and the redesign you didn't need. Customer discovery, that's definitely not something I'd hand fully to an AI tool. There's a bunch in there that people still handle really, really well. Actually talking to someone, asking follow-up questions, figuring out what to ask and how to ask it. I know there's some testing happening with AI tools now, and from what I read, it's still a mixed feeling about how well it works. You can absolutely use AI to prep for the interview, interpret the data, and pull out the insights. But asking the follow-up questions, the judgment of what to dig into, in my mind, that still sits more with a person than with the AI. And the alignment, obviously. Getting a team on the same page. That's hard to do with a simple AI tool. They tend to generate a lot of data and a lot of versions and maybe hand you something at the end. But kind of, that's it. So in a nutshell, it depends on the product. How important are design, craft, and actually solving your customers' problems? How much competition do you have? And where do they sit in terms of experience they provide? If design is mostly execution for you, a tool might cover a reasonable part of it. If design is how you compete, a tool won't get you there on its own. And look, if you're open to working with a studio instead of a full-time hire, that's an option too. We work with growth-stage tech startups as an embedded team. I run the heavy UX, a designer handles the rest through to-dev handoff. If you've got that hire versus tool decision in front of you, the link's in the description and I'm happy to talk it through. And if you're into UX, products, and figuring out what works, subscribe and I'll see you next time. In the meantime, check out this video. Hopefully it helps you make the call. Thanks.
