Built the product,raised the team.

Head of Product & Design

Player-coach, building product teams that work AI-native. Over 13 years I've shipped products across fintech, public sector, B2B and consumer, some with 100K+ users across the EU and US, and built the teams that ship them.

Trusted by

PracticeSince 2013

I work as a player-coach, close to the product and the actual work while setting direction for the team around it. What I care about most is that the product and the team keep getting better.

Selected work

Mostly leadership engagements and 0 to 1 products. A lot of my enterprise work is under NDA, so I'm happy to walk you through it on a screenshare. For older visual work, have a look at my Dribbble.

01 / 11

Highlights

A few things I'm proud of from the last couple of years.

Mentored 50+ career-switchers into UX through Designlab, several of whom now work in UX designDesignlab

How I work with AI, and how I bring a team along

AI runs through a big part of my research, synthesis and decision making. More importantly, I've helped a whole design team find a shared way of working with it.

Raising the team.

For me leadership is something you practise, not a title. I try to stay close enough to the work to set the bar by example, and far enough back to give people room to grow. It's about giving people the support and guidance they need to hit the team's goals, and in the end the real check of quality is the product that ships.

Growing people through critique

Honestly, I think designers grow most in the review of their work, a lot more than in a one-on-one. Group critiques also help a team set its own standards and learn from each other.

Writing the standard down

Quality only becomes real once it's written down. Design systems, principles and review criteria help turn taste into something a large team can refer back to.

Staying close to the work

I still design myself. Doing hands-on work on the hardest problems keeps me credible in my knowledge, and it keeps strategy honest about what it really costs to ship something.

Understanding the problem first

I don't open Figma, or an AI tool, before I really understand the problem. AI is super valuable for decisions and ideation, but it doesn't replace judgement on key issues.

15+designers and front-end engineers led at VDAB

2design teams built at startups

50+students mentored through Designlab

100,000+users served in the various products I've designed

13years across fintech, public sector, B2B and consumer

Clients

Recognition and awardsRed DotiF Design AwardHenry van de Velde LabelBootstrap ExpoUnizo

Working together

Fractional head of product & design

Ongoing leadership for teams that need senior product and design direction.

Interim leadership

I step in, get the team and the work stable, set the standard and hand over to a permanent lead when you're ready.

Advisory

For founders and product leaders shaping a 0 to 1 product, or building out a design function.

Hands-on lead designer

When that's what the product needs, I work hands-on, from discovery and research to the design system.

The easiest way to start is a short call - I promise 15 mins will be worth it :)

sven@umber.me

Back to overview

How I work with AI

AI made generating options cheap. So for me the interesting work moved to what surrounds it - the context a model works from, and the judgement we apply to what comes out.

I think we're not only designing screens anymore. We're also rebuilding the conditions design work happens in, and making them readable for people and for machines.

Research, principles, design systems and past decisions are now inputs a model works from. So the quality of that foundation pretty much sets the ceiling for everything AI produces for a team.

The maturity ladder

I see a lot of teams that are still at prompting and try to jump straight to agents. That usually doesn't work, because an agent is only as good as the context and skills underneath it.

So I build it up step by step, and most of the value sits in the middle. It starts with context - research, principles and constraints written down once, so a model can reuse them instead of starting from scratch every time. On top of that come skills: a proven way of doing something well, which the model applies the same way every time. Skills then chain into workflows the whole team shares, and only once those hold up does it make sense to give a model more autonomy as an agent. Context and skills are usually where I start with a team.

The context stack

Six layers a model needs to design well. Every organisation already has them - they just live in wikis, Figma files and people's heads, and very little of it is machine-readable.

So the work is mostly about translating and structuring what's already there, not inventing something new. Deciding what stays persistent, what gets added per task and how it all reaches the model is basically information architecture, but for machines.

The loop

The immature way of working is prompt, generate, accept. The loop I run starts earlier and ends later.

Bringing a team along

Adopting AI as one designer is a workflow question. Doing it with a whole design team quickly becomes a leadership question.

A safe space first

Freedom to explore, make mistakes and use their own tooling, without being too prescriptive about which technology to use.

A learning environment

People share what they've learned and how they want to apply it in their day-to-day work.

Then working on it together

This is the most important part. Otherwise everyone builds their own process, components and AI workflows, and the team loses cohesion.

How it lands

The freedom and the sharing stay. On top of that there's a consolidated backbone - same repo, source material and context, shared MD/JSON files, tokens and design system. People can still deviate with a good reason, as long as they share it back with the team.

No AI design department

AI belongs to content, product and platform together, so I don't believe separate teams owning it is the right setup.

Judgement I don't delegate

Where I think the value moved now that the mechanical work got cheaper.

Taste, more than execution

Execution is getting commoditised quite fast. Knowing what good looks like and holding the output to that is what I think stays valuable. The ceiling of quality is also how well you provide context. From shipping the right thing to embedding taste and craft, that's where the effort sits.

The human problem

Solving for the person and where they want to get to in an experience. AI is becoming a capability inside every product rather than the headline, and that's what's left to design.

Want to talk through any of this?

sven@umber.me Back to overview