Article · November 2026

How AI-assisted design is changing my workflow as a product designer

AI is changing how we research, explore, prototype and build digital products. For me, the biggest shift isn’t simply that we can work faster. It’s how AI lets us experiment earlier, explore more ideas and spend more time on the decisions that make an experience better.

I’ve spent much of my career in technology-driven product environments, where experimenting was often the best way to find out what a new possibility could mean for people. That curiosity hasn’t changed. What has changed is how quickly an idea can become something we can try.

This article isn’t about designing AI features. It’s about designing with AI: how it is changing the way I work, where it helps and where it doesn’t.

What is AI-assisted product design?

Put simply, it’s using AI to support the work we already do: synthesising research, framing problems, generating ideas, exploring interfaces, building prototypes, supporting tests and iterating on what we learn.

What it isn’t is handing over the decisions. The designer is still accountable for what gets built, why and for whom. AI can suggest, summarise and produce. It can’t know which trade-off is right for these users, in this context, with these constraints.

AI extends the work. The responsibility stays with us.

From a linear process to a fluid workflow

Many of us learned product design as a sequence of stages: research, define, wireframe, design, prototype, hand over, build and test. Each stage produced a deliverable, and each handover was a moment where context could get lost.

In practice the process was rarely that tidy. But because building something real was expensive, testing with working software often came late, sometimes after most of the important decisions had already been made.

AI-assisted workflows change that cost. An idea can become a functional prototype much sooner, which means we can move between understanding, exploring, building and testing many more times before we commit.

From handoffs to continuous experimentation

Comparison of a staged traditional product design process with an iterative AI-assisted workflow that moves between understanding, exploring, functional prototyping, user testing and refinement.

Traditional workflow

Stages and handoffs

  1. Research
  2. Define
  3. Wireframes
  4. UI design
  5. Clickable prototype
  6. Developer handoff
  7. Implementation
  8. Testing

AI-assisted workflow

A continuous learning loop

UnderstandResearch and context, all the way through
  1. Explore
  2. Build a working prototype
  3. Test with people
  4. Refine

AI supports exploration and production; people provide context, judgement and validation.

An illustrative comparison. Real projects are rarely this linear, or this tidy.

That doesn’t make research or engineering optional. If anything, understanding the problem becomes more important, because it has become so easy to build the wrong thing quickly.

How AI supports each stage

Research

AI can help synthesise interview notes, cluster observations and spot patterns across sources. I treat its output as a first draft: every insight goes back to the source material before it shapes a decision, because a confident summary can still be wrong.

Define

I use AI as a sparring partner to challenge assumptions and sharpen the problem statement. Asking it to argue against my framing, or to describe the problem from different users’ perspectives, often reveals what I’ve taken for granted.

Explore

It’s useful for comparing alternatives quickly and for questioning established patterns. Why does this flow need five steps? What would it look like if it needed none? Exploring more directions early makes it easier to see which ones are worth pursuing.

Design

AI can generate starting points for layouts and copy. The design work starts after that: hierarchy, clarity, accessibility, tone and consistency with the design system. These are judgement calls, and they’re where much of the quality comes from.

Prototype

This is where I see the biggest shift. With AI-assisted coding, a prototype can have real behaviour: data that changes, states that update and edge cases that actually happen. That makes it possible to test how something feels to use, not just how it looks.

Test and iterate

Nothing replaces watching real people use what we’ve made. AI can help analyse feedback and suggest changes, but the learning comes from the users. And because changes are quicker to make, the loop from insight to improvement gets shorter.

Tools in the workflow

The tools change quickly, so I think of them by the job they do rather than by name. These are the ones I use today:

  • Thinking and synthesis: ChatGPT and Claude.
  • Interface exploration and design systems: Figma.
  • Functional prototyping: Claude Code.

None of them is the point. What matters is how they fit together, and that the work stays connected to the people we’re designing for.

Why design systems matter more, not less

When AI generates interfaces and code, it produces whatever it’s guided towards. Without shared rules the result is a patchwork: similar components that behave slightly differently, inconsistent spacing and colour combinations that aren’t accessible.

Reusable components, design tokens, interaction rules, accessibility requirements and clear documentation give both people and AI the same foundation. A well-documented design system becomes something AI can work within, which makes it easier to stay consistent as the pace increases.

Closer collaboration between design and development

The gap between design intent and working implementation is shrinking. A designer can share something that behaves the way it should, not just a picture of how it should look, and that makes conversations with engineers more concrete.

But a working prototype isn’t production software. Engineers bring expertise in architecture, security, performance, reliability and scale that a prototype doesn’t address. The opportunity is to work together earlier, not to blur responsibilities.

The skills that matter more

When producing becomes easier, other skills become more valuable:

  • Problem framing: knowing which problem is worth solving.
  • Design judgement: deciding what is good enough, and what isn’t.
  • Technical curiosity: understanding enough about how things are built to explore what’s possible.
  • Experimentation: being comfortable trying, failing and learning in small steps.
  • Systems thinking: seeing how a change affects the whole service, not just one screen.
  • Human-centred research: staying close to the people we design for.

A small experiment to try

If you want to explore an AI-assisted workflow, start small:

  1. Pick a real problem, ideally one you already understand well.
  2. Explore a few different approaches before choosing one.
  3. Build a small functional prototype with an AI-assisted coding tool.
  4. Test it with a handful of people.
  5. Iterate on what you learn.
  6. Reflect on where AI helped, and where it got in the way.

The last step matters most. It’s how you find the balance that works for you and your team.

Knowing what is worth creating

AI is making the design workflow more fluid, connecting research, exploration, interface design, prototyping and testing in shorter loops. The real benefit isn’t only speed. It’s more time for the parts of the work that need people: understanding, judgement and learning from real users.

Because when creating becomes easier, knowing what is worth creating matters even more.

Thinking about this too?

Shall we make something good together?

If you’re rethinking how your team designs and builds, or want to explore what an AI-assisted workflow could mean for your product, I’d love to talk.

  • AI-assisted design
  • Functional prototypes
  • Product and design leadership
+46 73 388 84 94

Get in touch

Tell me what you are working on and I will get back to you.

All fields are required.

I only use your details to reply to you.