The evolution of design

The tension between traditional design methodologies and emerging AI-powered approaches highlights a fundamental transformation in how we create digital products. As designers increasingly embrace rapid prototyping and AI-assisted tools, we're witnessing not the death of design but rather its evolution into something more responsive, iterative, and immediately testable.

The shifting design landscape

The traditional web design workflow has typically followed a structured path: manage expectations, define objectives and audience, plan site structure, and move through distinct phases of research, design, and development (Teamwork). This methodical approach has served as the foundation of user-centered design for decades.

But today, we're seeing a shift. Designers are bypassing some of these established practices—not because they no longer matter, but because modern tools allow for new, faster ways to explore ideas and test them live.

The traditional process under pressure

Iterative design—research, prototyping, testing, refinement—remains powerful because it enables learning from failure before release (Interaction Design Foundation). As M Cobanli said: “Great design is the iteration of good design.”

However, the process is under strain. Some design leaders argue that AI can replace parts of this workflow out of convenience (UX Design). This raises questions: are we solving the right problems, or just shipping faster?

The rise of AI in design: evolution or regression?

AI-powered design tools are reshaping how we prototype and build. Tools like ChatGPT, Replit, and v0.dev let designers build functional, testable products in minutes—dramatically reducing the time to validate ideas (Lenny’s Newsletter).

We’re seeing a major acceleration. As one expert observed: “We just created a working prototype of a CRM in less than five minutes.” This kind of speed was unthinkable even a year ago.

Designing in the browser: a new paradigm

Browser-based design—long a controversial idea—is becoming normalised. It doesn’t mean ditching design tools. It means making the browser your canvas for rapid iteration (Cloud Four, Vanseo Design).

Instead of mockups handed to developers, we co-create live experiences. This speeds up decision-making and shortens feedback loops.

AI in research: helper, not replacement

AI excels at certain research tasks—helping write interview guides, summarising usability feedback, and accelerating planning and documentation (NN/g). These tools can boost efficiency, especially during early stages like desk research or scripting usability tests.

But AI cannot replace the deep, human work of synthesis.

As Jon Kolko argues in Synthesis and ChatGPT, the output of research is not the point of research. The true value lies in the sensemaking: the mental and collaborative process of forming meaning, drawing connections, and building shared understanding within a team. That process is not just a step in the workflow—it is the work.

ChatGPT can make you that report. It can't form new connections in your head for you.

Generative research is about immersing yourself in people’s contexts and learning to see the world through their eyes. The hours spent clustering quotes, debating interpretations, rewatching recordings, and externalising your thoughts through frameworks and diagrams—that’s where understanding emerges. AI might summarise data, but it can’t reflect, empathise, or reframe your worldview in the same way a messy, human synthesis session can.

Even if we imagine an ideal AI that can form rich meaning from data, outsourcing this thinking to it robs us of the learning. As Kolko notes, “In some perverse way, you've become the dumb utility, and the AI has become richer.” You gather the data; it makes the meaning.

In that scenario, you don’t grow. You don’t deepen your understanding of your users, your product, or your own assumptions. You just get the scraps—pre-digested summaries devoid of context or insight.

AI is a tool. But synthesis is a craft. And that craft is core to the value of research.

As Jakob Nielsen puts it, AI-led research without human context is “much worse than nothing” (Jakob Nielsen). We should embrace AI where it enhances our work—but never let it replace the hard, generous, and deeply human job of making sense of what people truly need.

Testing in production: the new normal

Another shift is testing in production (TiP)—rolling out features live to a small audience to monitor real usage (BrowserStack, LaunchDarkly).

For example, an e-commerce platform might test a new recommendation engine on just 5% of traffic, measuring real-world clickthrough and conversion rates. This level of rapid validation is now essential for staying competitive.

Finding balance: speed with intent

The goal isn’t to replace traditional design—it’s to evolve it. We need to strike a balance between speed and rigour.

Browser-based prototyping, continuous testing, and AI tools are powerful when used alongside—not instead of—research, collaboration, and insight. These tools help us move fast, but they don’t replace the fundamentals.

Personalisation and real-time design

AI enables more dynamic, personal experiences. We can now design with live user data, co-create with real people, and adapt in real time.

As long as our work is grounded in research and done collaboratively with product peers, we’re not abandoning design—we’re modernising it.

Conclusion

Design isn't dead. But it's changing. Fast.

We’re shipping earlier, learning faster, and collaborating more fluidly across design and code. The playbook may look different—but the goals haven’t changed: solve real problems, understand users, and build things that matter.

The tools may be new, but the heart of great design remains the same.

Design isn’t dead. It’s just shipping faster.

Further reading: