Company

About Onefold

One person, in Cologne, building the layer that sits between a design system and the agents that now write against it.

Why this exists

Design systems spent a decade solving a coordination problem between people. That work was not wasted, but the audience changed. A meaningful share of interface code is now written by agents, and an agent cannot ask a colleague, cannot read a Figma file, and cannot infer a rule from three examples it has not seen.

What it can do is retrieve. So the systems that will hold up are the ones whose knowledge is retrievable, and the ones that will quietly erode are the ones where the important parts only ever lived in people's heads. That gap is what Onefold measures and then closes.

What Onefold refuses to become

  • Another component library. The world does not need one more, and yours is not the problem.
  • A documentation site. Documentation that lives somewhere new is documentation that goes stale somewhere new.
  • A Figma replacement. Design belongs in Figma.
  • A chat window bolted to a design system. The useful version of this is a structured query answered from a graph, not a conversation with a documentation dump.
  • A tool that asserts things it cannot back up. Every claim in the product carries its evidence, or it does not ship.

How it is being built

In phases, in public, and in the order that produces something real soonest: GitHub analysis and a readiness report first, then Figma reconciliation, then MCP access for coding agents. Anything that does not move that sequence forward gets questioned rather than scheduled.

The architectural decisions are written down as they are made, including the ones that turned out to constrain later choices. The product page describes what a run actually does, and security covers what happens to your code.

Who

Oliver Pitsch. Cologne, Germany. Reachable at onefold@pitsch.me, and the person who answers is the person building it.

If you own a design system and something on this site is wrong about your situation, that is worth an email. Being corrected early is cheaper than being wrong in public for a year.

Come and break it

Early access is small on purpose. A real repository and honest feedback are the whole ask.

Request early access