Design direction as code

Your AI coding tools produce whatever you ask for. The problem is knowing what to ask for, and recognising when the output is wrong in ways that don’t throw errors.

AI won’t tell you that your onboarding has three steps too many, or that your pricing page buries the most important comparison. It won’t tell you that your settings screen needs hierarchy, not more toggles. And it won’t tell you which typeface carries the weight of your positioning, or why that particular shade of blue undermines the trust you’re trying to build. These are judgment calls across the full stack of product design, and they’re the only part that hasn’t been automated.

Based in Copenhagen. Working with founders, product teams, and early-stage companies across Europe and globally, wherever the person building the product isn’t a designer and the AI filling that gap isn’t a decision-maker.

Why this exists now

Two years ago, this service wasn’t possible. A founder building a product needed either a full-time designer or an agency engagement. The design work was inseparable from the production work. You couldn’t buy taste without buying hands.

AI coding changed that equation. Production is essentially solved. A founder with Cursor can ship a working product in a week. But production without judgment produces products that all look the same and all make the same UX mistakes: the same rounded corners, the same safe palette, the same three-step onboarding that nobody tested, the same settings page that grows until it’s unusable. Demo day is full of apps that technically work and feel like nothing.

What’s missing isn’t a designer. It’s someone who makes design decisions for a living, embedded in your team on a fractional basis, encoding those decisions so the AI follows them from that point forward.

What makes this different from hiring a designer

Decisions, not deliverables

A traditional designer gives you mockups and expects someone to implement them faithfully. I give you decisions about how your product should work and how it should look, as documentation and code, that your AI implements automatically. No interpretation layer. No design-to-dev handoff. The output is the input.

Continuity, not handoffs

You’re shipping features constantly. Your design judgment should keep pace. An ongoing fractional partnership means the person questioning your feature scope, rethinking your flows, and refining your visual language has full context on your product, your users, and where you’re heading. No re-onboarding, no briefing documents, no lost nuance.

Judgment that compounds

Every flow decision, every design token, every rule in your cursor config, every interaction pattern makes every subsequent feature more consistent and more considered. You’re not buying a one-time intervention. You’re buying a system that gets better the more you build on it.

Embedded understanding, not external opinions

I’ve spent 20 years inside product teams, at agencies like ustwo, with brands like Bang & Olufsen, on platforms serving millions of users. The judgment comes from understanding how products actually get built, not from design theory applied from the outside.

Experience

  • 20 years designing digital products across mobile, web, and multi-platform ecosystems
  • 6 years as independent fractional design leader
  • Clients from early-stage startups to Novo Nordisk, H&M, Adidas, BMW, Bang & Olufsen
  • Deep expertise in brand-to-product translation, design systems, and product strategy
  • Copenhagen-based, serving teams across Europe and globally

Working together

Getting started

We start with your product as it exists: the live app, the repo, the vibe-coded prototype. I review it, use it, ask about your positioning and audience, and then work across the full scope of product design. That means both the structural questions (does this feature make sense, how should this flow work, what can we cut) and the visual ones (palette, type, spacing, layout personality, component design). The initial phase builds the brand direction, establishes interaction patterns, and encodes everything into artefacts your AI tools can use. You get design documentation, rules files, tokens, and a rationale document.

Ongoing fractional partnership

From there, I stay involved. New features need someone to think through the flow before the AI builds it. Product areas expand and need design direction. Components need designing for specific contexts. Having someone with full context who can make those decisions quickly, across both UX and visual design, is the point of the fractional model. I work with a small number of clients on this basis, typically a few days per month, scaled to what you actually need.

€1,080 per day, typically a few days a month.

Let’s talk

Whether you’ve just shipped your first vibe-coded MVP or you’re a funded startup realising your product needs a point of view, let’s explore whether fractional design judgment is what gets you there.

Schedule a conversation

About this site

Built with Svelte. A small example of design-in-code: building products that ship efficiently rather than living in design files.

Set in Special Gothic, Maname & Martian Mono.