OutbuildDesign systems2025–2026

Design System 2.0: a token-based foundation for dense construction software

Role

Senior UX/UI Designer

Team

one additional designer, frontend engineering, PMs, mobile engineering

Timeline

Sep 2025 – ongoing

Platform

Web, iPad, mobile (Figma, Supernova, Storybook)

Scope

Self-initiated in my first week · New system, rolled out across a live product

Aa

TL;DR

The stake

The Figma files were complex and inconsistent, which slowed both design and engineering.

What I did

I defined the foundations, built subsystems for dense data surfaces, and set up documentation that syncs with Storybook.

How I led

I flagged the problem in my first week, drafted a phased rollout plan with our engineering lead, and set the standard engineering now builds against.

Results

From 249 components to 800+.

Design System 2.0 components

249 in four weeks
800+ today

249 components in my first four weeks, grown to 800+.

Feature shipping speed

About 3× faster, together with AI-coded prototypes. Outbuild.

Design-to-development time

40% faster with the TeamGantt Design System, saving an estimated 90+ hours a week.

Outcomes

Adopted by every product team.

  • 249 components and 612 design tokens in my first four weeks.

  • Every product team adopted it; engineering builds against it as its shared standard.

  • Later releases built on it: Roadblocks in March 2026, Company and Project Settings in October 2026.

  • A mobile system for the iPad and mobile apps.

01 · Context & problem

Inconsistent files show up on every dense screen.

In my first week I flagged the complexity in our Figma files to an engineering lead. For a product made of tables, cells, and Gantt charts, inconsistency in the source files shows up on every screen.

I treated the design system as the first piece of the design practice. Every other improvement, including the Lookahead redesign, would depend on it.

02 · Constraints

A live product on three platforms.

  • A live product. Engineering couldn’t stop to migrate everything at once. I drafted a plan with our engineering lead to implement the system efficiently and over time.

  • Three platforms. The system had to serve web, iPad, and mobile.

  • Density. Generic UI kits don’t cover the hard parts: tables, columns, and Gantt bars.

03 · My role & influence

I started the system and built most of it.

I started the system and was its main builder. A second designer later expanded it. Frontend engineering implemented it in Storybook. A PM introduced the new system to engineering in October 2025 as the home for all modules, buttons, and typography.

04 · Strategy

Foundations, then subsystems, then the kit.

  1. Foundations: typography, colors, grids, breakpoints, border, elevation, spacing.

  2. Subsystems: a fully responsive platform layout, plus menu, modal, cell, table, column, form input, Gantt chart, and Gantt bar systems.

  3. UI kit: button, icon button, checkbox, radio, toggle, tab, scrollbar, date picker, pagination, avatars, tooltip, banner, divider, and color selection.

05 · Key decisions & trade-offs

Four decisions behind the system.

1. Build subsystems, not just atoms.

I could have shipped a basic kit first. Instead, I built systems for the dense surfaces users live in: tables, cells, columns, and the Gantt chart. That took more upfront work, so the hardest screens would share the same rules.

2. Documentation that syncs with code.

I started the documentation site in Figma Sites in September 2025. In October I moved it to Supernova, because Supernova syncs with Storybook. Design documentation and engineering components now come from one source.

3. Platform-wide token changes, announced where engineers work.

In April 2026, I changed the default button text weight from bold to medium. The token moved from body.bold.m to body.medium.m. That one change affected every normal text button in the platform. I posted it in the Storybook channel so engineering could see the full scope.

4. Test color in a sandbox, not in production.

For the new executive dashboard, I compared the 500 and 600 color steps in a duplicate testbed file. The real design system file stayed stable while we decided.

I also built a mobile design system file with buttons, icon buttons, sliders, sheets, and form components. I later proposed adopting Figma Slots for the menu and table systems.

06 · Earlier

I had built one before, at TeamGantt.

Before Outbuild, I built TeamGantt’s design system from scratch as the company’s sole designer. It grew to more than 1,800 responsive components.

The approach followed four steps: audit, tokenization, component standardization, and cross-team collaboration. Tokens used a three-tier structure: Foundation, Type, and Modifier. The color system had 10 shades per color, from 100 to 1000, and inverted automatically for dark mode. Themes were built to accessibility standards.

Results:

  • Design-to-development became 40% faster, saving an estimated 90+ hours a week.

  • Documented for engineering and used by 17 designers and engineers, on a three-tier token architecture with light and dark themes.

The main lesson carried over to Outbuild. A design system is a product for the engineering team.

07 · Reflection

The system now lives in more than Figma.

The system now changes in more places than Figma. Designers refine components inside coded prototypes, and those changes need a reliable path back into Storybook. Solving that path is the next piece of governance.

Next case study