Basis Lab logo

Basis Lab

Chrome extension design system for Bubble—1,500+ users, 40% paid conversion. Unstyled-first architecture solved compatibility for existing apps.

Basis Lab Component Library

Overview

Chrome extension design system for Bubble (no-code platform). Drop-in component library that works with existing apps. Built 0→1: designed and developed end-to-end.

Role: Founder, Designer, Developer

Timeline: 2023-2025

Impact:

  • 1,500+ users
  • 40% paid conversion rate
  • Unstyled-first architecture unlocked adoption for existing apps
  • Layered IA matched user mental models

Component Library

Drop-in component library with pre-built layouts, data patterns, and UI elements.

Component Library Overview Component Detail View

Unstyled-First Architecture

The problem: V1 pre-styled templates looked broken on existing apps with their own styles. Templates forced rebuilds—blocked users who needed it most.

The solution: Style as a choice, not a requirement. Unstyled (default) inherits host styles. Styled (optional) design token set. Clear upgrade path.

Styled and Unstyled Switcher

Layered Information Architecture

The problem: V1 organized by popularity (Buttons, Inputs, Lists). Design-savvy users moved fast. Everyone else: "When do I use what?"

The solution: Organized by how screens are built—Shells → Page Templates → Containers → Data Lists → Elements. Matches user mental models. "Removing is easier than building."

Component Sheet

Component Sheet

Interface Details

Interface Detail Interface Detail Interface Detail Interface Detail Interface Detail

Figma Design System

Sibling Figma system for builders to design in Figma first—components are 1:1 with the Basis extension component library.

Figma Template Design System

Key Outcomes

Adoption: Unstyled-first removed barriers—users could drop components in without breaking existing designs

Validation: Support questions shifted from "make it pretty" to "how do I compose screens"—right mental model

Growth: 1,500+ users with 40% conversion validated compatibility-first approach

Learning: Users needed structure, not opinions. Mental models matter more than technical accuracy. Built production-quality tool within platform constraints.

More work