Back to projects
Paused before dev

Hero Design System v2

The maturity assessment scored 2.2 out of 5. The restructuring started there.

Company
Conexa Saúde, healthtech
My role
Senior Product Designer (Design Ops & Design System)
Period
Nov 2025 to Aug 2026
Team
8 designers, reporting to the Head of Design. Co-ownership with the front-end Tech Lead.

The assessment

Score of 2.2 out of 5 across 7 dimensions.

I spoke with 10+ people, including PMs, designers and the VP of Product. I asked whether they knew the Design System and what stage they thought it was in. Then I compared Figma and Storybook and evaluated against WCAG criteria.

The problem

The Hero Design System existed, but existing is not the same as working. Teams knew the system was there, but did not know what was standard, what was outdated, or how to contribute without breaking what other teams already used.

The decisions

  • I stabilized v1, which 8 designers and the front-end team used on the patient and healthcare professional platforms. I standardized naming, reorganized the file architecture and published an accessibility guide based on WCAG minimum criteria.
  • I built v2 in parallel, with semantic tokens for light, dark and white label. The system changes colors for each client company.
  • I unified naming between Figma and Storybook, to have a single source of truth.
  • I designed the v2 architecture to be readable by AI (Figma, GitHub and Claude Code): anyone using vibe coding gets the exact component, with its behaviors mapped.
  • I worked in real co-ownership with the front-end Tech Lead: I owned vision, standards and prioritization, and he owned technical stability and releases.

How I used AI in this project

I used generative AI for research, synthesis and drafting models and proposals for v2, before starting work on the Design System.

Current state

Paused before integration with development. The v2 architecture is designed and documented in Figma. The connection with AI tools and the code pipeline have not been implemented. v1 was already being rolled out gradually to the backoffice and management products.

What I would do differently

Include usage instrumentation from the start, to prioritize the migration with more precision than the qualitative assessment alone allows.