NxtPeople

I lead product design for NxtPeople, a multi-product HR SaaS platform for employers and candidates. I built NxtPeople-DS, a design system with a two-tier token architecture that keeps every module and sub-brand consistent as the platform grows.

View Website
  • SaaS
  • Mobile
  • Enterprise
Product Type
Multi-product SaaS platform
Industry
HR
Timeframe
2018–now
Role
Product Design Lead & UX Architect
Team
Me, one designer I lead, and the NxtPeople product and engineering team.
Scope
Design system, Product UX strategy, Component architecture, AI-enabled design-to-code workflow
NxtPeople product interfaces, including login, AI assistant, mobile planning and scheduling.

The challenge.

NxtPeople grew from one product into a platform of modules: NxtForm, NxtMatch, NxtPlan, NxtPay and more. Each module risked developing its own patterns, and a sub-brand, GoTemp, needed its own identity on the same foundation. Without a shared system, every new feature meant redesigning basics and every release added inconsistency.

The key decision.

I separated brand from meaning in a two-tier token architecture: brand tokens hold the raw values, semantic tokens describe their purpose. Components only use semantic tokens. A new module or brand changes values in one place, without touching components.

The rejected option & why.

A separate component library per product or brand. It looked faster at the start, but it would have multiplied maintenance and let the products drift apart. A second token set on the shared NxtPeople-DS gave GoTemp its own brand without a second system.

The outcome.

One design system now supports multiple products and a separate brand. Product teams build new screens from shared components instead of designing from scratch.

The final verdict.

I own the NxtPeople product experience end to end, aligning design, product and engineering. The design system runs from Figma tokens through Tokens Studio and Style Dictionary into production CSS, feeding a shadcn-based component library in Laravel and Vue. I use Figma MCP and Claude Code to speed up the design-to-code handoff, but every generated component is checked against the system’s rules before it ships. That’s the point: AI makes the components fast, the system decides whether they belong.

Facing the same challenges?

Every case here came down to one decision that was hard to undo. Book a free sparring session: I’ll look at your product and tell you which decision matters most right now.

Book a free sparring session

Ask Superflow

AI assistant · answers from this site

Superflow AI

Questions about working with Bart? Ask about the approach, services or your next step.

AI can make mistakes. Messages are sent to OpenAI to answer your question. Don’t share sensitive information. Privacy