Every team builds its own buttons.
Your product looks a little different on every page. Designers and developers align every detail one by one, and the same component exists in five variants because nobody knows which one is right.
Does this sound familiar?
The same elements look different in different parts of the product.
Developers implement designs by eye because Figma and code don't match.
Every new page is designed and built from scratch.
How I approach it
Inventory
I capture which components, colors and patterns already exist in your product and show where the biggest inconsistencies are. This becomes the foundation of the system.
System in Figma and code
I define design tokens and build the most important components in Figma and as a component library in code at the same time. Under my guidance, AI agents write much of the variants, tests and documentation, while I make the design decisions myself. Both sides stay in sync for good.
Rollout in your team
Documentation, examples and a step-by-step rollout into existing pages make sure your teams use the system and keep developing it themselves. This includes rules for AI assistants and automated checks that make sure people and agents use the right components instead of inventing new ones.
What waiting costs you every month
Without a shared system, the same component gets built several times, often by different teams at once. So you pay for the same development more than once. Afterwards every variant has to be tested, maintained and developed further on its own: a small design change is not one adjustment but four, and each turns out slightly different. AI tools make this worse, because without guidelines they create yet another slightly different variant with every request. With every new feature, this extra effort keeps growing.
Why AI rules alone are not enough
Many teams give their AI assistants rules and skills: use the components from the design system, no hard-coded colors, no custom spacing. That helps, but it is no guarantee. Rules and skills are instructions in natural language, not linters and not parsers. An agent reads them, interprets them and follows them most of the time, but not always. The deviation only becomes visible once it has spread through the product.
A design system only becomes reliable with tools that actually check the code: lint rules that reject hard-coded colors and custom spacing, typed components that only allow approved variants, and visual tests that reveal every deviation. They run on every change, whether it comes from a person or an agent, and they fail instead of politely reminding.
I build both together: rules and skills so AI agents use the system from the start, and linters, types and tests that enforce it. That is the only way the code stays consistent and readable in the long run, for your team as well as for the AI agents that keep working with it.
Frequently asked questions
Is a design system worth it for us?
Yes, as soon as several people design and develop one product or you have several products under one brand. The effort pays off because every component is built and maintained only once.
Do we have to redesign our interface for it?
No. The system builds on your existing design and makes it consistent. Existing pages are migrated step by step without your users noticing a break.
Which tools and technologies do you use?
For design I work with Figma, in code with design tokens and a component library for your framework, such as React, Angular or Vue. Documentation lives where your team already works, for example in Storybook.
Who maintains the system afterwards?
Your team. I set up a process for how new components are proposed, reviewed and added, and hand over the system so it keeps growing without me.
How long does it take to build?
A usable first version with the most important components usually takes a few weeks. After that, the system grows with your product instead of covering every edge case up front.
Do we even need a design system now that we have AI?
More than ever. AI assistants build interfaces in seconds, but without guidelines a little differently every time. A well-documented design system is the foundation that lets AI deliver consistent results. I design the system and prepare it so that people and AI tools can work with it alike.
Let's talk about your situation
In a free initial call we take a look at your problem together. You get an honest assessment of whether and how I can help.