An AI-Powered Prototyping Platform for Designers
- Date
- Apr 2026
- Client
- Waitwhile
- Service
- Product design0 to 1 toolingConversational UXAI integrationSystem designPrototypingFrontend build
In short
I designed and built a self-service prototyping platform for Waitwhile's designers. Idea to a live, shareable prototype in about two minutes, with zero engineers in the loop. The whole thing came together in a week.
The old way
Designers prototyped in Figma, wiring up every interaction by hand. But the interactions that actually matter are subtle, and Figma can't really fake them. You only feel them in real code. Vibe coding finally makes that reachable, but there's still a setup hill to climb. And if you want to prototype on top of an existing product, you end up rebuilding the same components from scratch every time, which is a waste.
The real problem was access, not infrastructure
None of the building blocks were missing: a base template to start from, a repo, and CI that publishes a live link on every push. The catch is that branches, CI and public URLs are foreign territory for most designers. So I packaged it all: a boilerplate so nobody rebuilds the same components twice, a shared gallery as the home for every prototype, and a one-step way to spin up a new one. The bar: idea to live URL in under two minutes, no engineer.
The gallery
The gallery is a retro Windows 9x desktop, and the metaphor does real work: each prototype is an icon you open in a window, and archiving is dragging it onto the Recycle Bin. It also signals "made with care, but not production," which makes people comfortable experimenting. They started talking about it, which did more for adoption than any doc.
The creation flow
The "New Prototype" modal keeps it to three fields: name, squad, and a starting point (full app, one page, or blank). That last choice changes what actually gets deployed. Each prototype window also has a mobile, tablet and desktop selector, since people share mobile flows but review them on desktop.
Meeting designers where they already were
A web form is fine for deliberate creation, but designers were already living inside Claude. So I built a Claude skill: say "create a prototype called Appointment Redesign, starting from the Calendar page," and it spins up a branch, prunes to that route, and hands back a live URL, without ever leaving the chat.
Decisions I'm proud of
- Branch-per-prototype, not repo-per-prototype, so the gallery lists everything from one call and nobody loses work deleting a stray repo.
- Prune at create time, not first session, so the very first build shows exactly what was asked for instead of the whole app.
- The real UX wins here were infrastructure choices, not button placements.
Role and scope
I designed and built all of it solo in about a week: the gallery, the creation flow, the Claude skill, and the wiring underneath. A new designer can self-onboard from the built-in guide in a few minutes.
