What Will Save Design?

UI execution is being outsourced, teams are shrinking. The industry is more of a mess than it looks. In uncertain times, shrinking the team and increasing efficiency through AI is precisely what will get applause across board & management teams.

With output commoditized, judgement about what actually is good becomes the scarce thing. But here's the catch: non-designers also have a notion of what's good. The real battle is less about producing (better?) work, and more about proving that your judgement is the better one. The market will shrink, but the demand for talent up the chain will grow.

When mediocre (yet solid) work is cheap and everywhere, beating precisely that is what will separate products. That is what will require strategic judgement, not more screens. Hence the current shift you're seeing from Figma to more code is, in my humble opinion, not what will save the profession. It's a different flavor of output.

So how do you become the trusted signal for what's superb versus merely good? The AI will suggest UI confidently. The right designer will cross-check whether actual well-written UX copy was used, or whether it's consistent with another flow, or whether this might be the area where you do want to break the consistency paradigm of the design system. Or whether this feature was supposed to be built in the first place, looking at the research.

That's my Friday-in-a-heatwave take: output, even if it's code, will not save design.

Thoughts on AI and Design

Recently, I was in New York to spend a day with design leaders across companies to understand the impact of AI on our field.

As technology is evolving rapidly, we focused our discussion on many themes.

The first thing I want to reflect on is the general sentiment that I believe is applicable to anyone working in technology: there can be a lot of doom thinking regarding how rapidly advancing AI tooling will take over (aspects of) our jobs as UI, UX, and product design become democratized. Anyone can design now.

However, there’s a lot to be optimistic about. Because ultimately, what are the talents of good product and design folks?

  • High levels of craft and taste can help products stand out from mediocrity.

  • Style is something that AI can replicate, but taste will always remain a human perspective.

  • The ability to craft a vision, but also show individual steps toward that vision to your team.

  • The ability to display empathy, be user-centric, and focus on the desirability of products and features.

  • Collecting and presenting evidence for fact-driven design.

So ultimately, I look at it as a force multiplier for great designers.

Don't derail the train

Overloading users with complexity in product design can derail their journey. The goal is to avoid cognitive overload by using progressive disclosure. Start simple, then reveal advanced features as needed

Key guidelines:
- Strip away non-essentials: keep primary screens clean.
- Gradually unveil complexity: introduce advanced use cases step by step.
- Guide the journey: build user confidence through above principles.

Remember, exceptions exist: some products, like trading software, require more complexity. But as a rule, we tend to try to do too much within our interfaces. If a flow takes longer, but the overall understandability of the screens is improved, it is better UX. There's a reason a guest checkout flow isn't completed in a single step.

The Empathy Gap

You can never truly imagine all the ways a user interacts with your product. This is due to what we call the empathy gap: it’s a bias where we aren’t fully able to predict someone’s feelings, thoughts and behaviours. So how do you fix this?

First, understand why this exists: we stand too close to the product, knowing its assumptions. Users have different contexts, levels of expertise or expectations that are not necessarily aligned with your team.

Next, look at what stage you’re in the product development cycle and attempt to close the empathy gap:

  • Are you in the early stages? Use circumstantial research such as interviews, diary studies, …

  • Are you in the late stages? Shift to validation research: think of usability testing, tree testing, heuristic evaluation, …

  • Are you post-launch? Continuously observe real-world usage with tools like Smartlook to surprise yourself in the behavioural patterns.

You can never close the gap 100%, but you can avoid mistakes that in hindsight are obvlious.