A design system isn't a Figma file. It's a set of decisions made once so you don't have to make them again.
Here's how we think about ours at Magnift.
Start with constraints, not possibilities
Most design systems fail because they try to cover everything upfront. Infinite color palettes, dozens of components, elaborate token hierarchies before a single real screen exists.
We did the opposite. We started with three questions:
- What's the one color that signals action?
- What two fonts cover every situation?
- What's the smallest set of components we actually need?
The color answer: Gold
#F5C842. One accent color. It appears once per screen, on the single most important element.
This constraint forces hierarchy. You can't make everything important if only one thing can be gold.
The typography answer: Bricolage Grotesque + DM Sans
Bricolage Grotesque is expressive at large sizes — it gives headlines personality without being loud. DM Sans is neutral and highly legible at small sizes. Together they cover the full range from hero text to input labels without competing.
The rule is simple: headings get Bricolage, everything else gets DM Sans.
Components: only what ships
We didn't design a component library. We built components as products needed them, documented them as they stabilized.
This sounds inefficient but it produces better components. A button designed for a real use case is better than one designed for theoretical completeness.
The result
A system that's actually used, because it was built from real decisions rather than aspirational ones. Every product we ship is consistent because the system earned that consistency by solving real problems.