UX Perspectives: Navigating AI-generated prototypes in design

Clients, developers and product managers can now build high-fidelity, working applications themselves, using tools like Claude or Lovable, without writing a line of code. A rough idea can become something clickable in an afternoon instead of a two-week wait to get a static mockup.

We asked designers working across agencies, in-house teams and freelance practices what that shift means for UX. The consensus was clearer than you might expect. Almost everyone agreed that AI is a genuinely useful addition to the design process — faster iteration, quicker buy-in, fewer weeks lost to guessing what a client meant. But that same consensus came with a condition attached: AI can accelerate the work, but it can't replace the judgement that decides whether the work is any good.

Olatunde Ayilara RGD, Senior UX Specialist and Content Designer at the Ontario Public Service

I think this is a good for our industry. Being able to go from idea to working prototype quickly is great since it means we  can test ideas with users faster. The negative change will be the security implications; people are sending a lot of half-baked solutions in today's market. Also, as a creative, it can limit creativity and brainstorming—you can literally have AI do your brainstorming for you. I always make sure to find time to think on my own, writing my thoughts down with pen and paper. You can't code common sense. AI hasn't gotten to a point where it can think strategically the way humans will, especially for a custom-made solution.

Travis Gobeil RGD, Branding & UX Specialist at Kern Unit

With AI, we can iterate faster than ever, and coding skills are no longer a barrier between an idea and something you can put in front of a user. The problem is that bad UX doesn't always look bad. Someone without UX knowledge can generate a flawed experience and have no way of understanding that it's flawed. The output looks finished, so it doesn't invite scrutiny. AI doesn't know it's bad, it only knows what it sees. Bad UX ships, gets fed back in as training data and the snake starts eating itself. What I tell my students and clients is that AI amplifies judgement, it doesn't supply it. If you can't tell a good result from a plausible one, you'll ship the plausible one every time. The value isn't in producing; it's in knowing what's worth building.

Samiksha Makhijani RGD, Lead Product Designer at Zero Hash

It really depends on how these tools are being used. If someone is using an AI tool to quickly bring an idea to life, share a concept with stakeholders or get early buy-in, that's a great use of AI. Where designers have an important role is creating governance around how AI is used in the design process. Just because we can build something faster doesn't necessarily mean we're building the right thing. At the end of the day, product designers still own the quality of the experience. We need to stay involved in what gets built, make sure it actually works for users and stay close to engineering through implementation. AI can help us move much faster, but the judgement behind what we build matters more.

Aaron Neilson-Belman RGD, Manager, UX & Brand Strategy at Brafton Inc.

AI is making it much easier to produce polished, functional interfaces. I don't see that as inherently positive or negative. The value depends on how these tools are used. When foundational UX work such as research, flow planning, information architecture and design systems is in place, AI can accelerate prototyping and iteration in a very positive way. The risk comes when speed is mistaken for process. Jumping straight to a high-fidelity application can produce something polished without establishing whether it has solved the right problem or supports real user needs. For me, ensuring UX delivers its best value means getting involved as early as possible, ideally before a project even begins. It means asking the right questions early and clearly communicating the business risks of rushing or skipping foundational UX work.

John Ryan, Co-founder at Significant Other

I know it's a cliché for designers to grin and bear the clunky PowerPoint "mockup" clients share as inspiration, but we've always taken the stance that it's a better starting point than a wish list that's untethered from reality. A client who has built a rough working prototype with AI knows what they want, and as long as they're open to us asking lots of hard questions and taking them beyond their draft, it's for the better.

But an AI-generated prototype can look "finished" in a way a sketch never did, so it's easy for it to just become the de facto solution. Our approach is to treat it as an indication of intent, not a design. We go back to the questions the prototype likely skipped: who is this actually for, what are they trying to do and what does the organization need it to do over time. Prototype fidelity used to be a signal that someone had done all of the thinking and refining. It isn't anymore, and making sure UX gets its due mostly means refusing to be impressed by how "real" the AI-assisted draft looks.

As for our own use of AI, we've learned that AI is good at producing that first draft of something, but unreliable at  anything meaningful beyond that first draft. Our approach is very much human-centred: Designers and developers define the desired outcome, build a clear implementation plan, then use AI to execute discrete flows or pieces of code. But we're never using AI front to back. Our work has to be maintainable and secure with or without AI in the mix, and a black box of generated code isn't either.

Josh Skinner RGD, Senior Staff Product Designer at Apartment List

In June of 2026, it was reported that Ford had rehired 350 veteran engineers after its big bet on AI to automate its design and quality systems failed. Stories like this interest me because they illustrate a reality of using AI tools for creation: To get it to create anything meaningful, you still need the expertise with a fundamental understanding of what you're trying to create. AI doesn't know whether it's right or wrong, or whether it's doing something correctly; it merely outputs what it was told to output. Clients, developers and project managers may skip that critical step of understanding and go straight to creating, which is a negative. At best, it will require someone with knowledge and expertise to fix the output. At worst, something unusable or even harmful gets put out into the world. In an age where you can prompt a webpage, designers are essential because they know how AI can best augment their workflow, and they can recognize when AI gets something wrong, understand why it's wrong and fix it.

Sean Archie Urriza RGD, Senior Production Designer at DIRTT

Positive, and I say that as someone who builds this way himself. When a PM or developer hands me a working thing instead of a paragraph describing it, we skip two weeks of guessing at what they meant. Stakeholder presentations got easier too. People can click around and use it themselves instead of watching me walk through static screens. The flip side is that a polished prototype can hide the problem. The client sees something beautiful, they nod and nobody asks whether it solves th fundamental problem. When I present, the research and the reasoning come out alongside the prototype, not after it. My approach hasn't really changed. I still start with people. Building got cheap. Knowing what to build didn't, and that's where I put the UX weight.