Work / Design Systems

Design Systems / Systems / Accessibility / AI readiness

Making design-system knowledge usable by people and AI

I transformed an existing design system into an accessible, AI-compatible foundation, rebuilding its components, connecting Figma with GitHub and Storybook, and creating the guidance needed for teams and AI-assisted workflows to apply and evolve it consistently.

Project meta

Role: Lead Product Designer
My contribution: I designed the system’s accessibility, canonical knowledge and AI governance architecture, enabling AI to work from trusted sources, recognise when change was needed and support governed evolution while keeping authority with the team.
Timeline: 9 months
Team: 1 PM, 1 Designer, 2 Engineers

Andrei Stanescu

4 September 2026

Design systems · AI-enabled design operations · Accessibility · Component architecture · Design governance · Storybook · AI-assisted workflows

OUTCOMES

A design system designed to outlive its creator

300+Interconnected documentation pages

6Codex skills for accurate system use

2Canonicalization workflows for preview and update

4Connected system surfaces across design, documentation and code

1Operating model designed to preserve quality after handover

Write-up in progress

The results are in. The story behind them is still being written.

This case study is being written up in full—the challenge, the shift in approach, and what was built. The outcomes above are real and already delivered; the full narrative is still being prepared.

RELATED WORK

More case studies

Designing the evidence AI needs to improve

I designed a human-in-the-loop audit platform that helped subject matter experts resolve disagreements between researchers and an AI agent—while capturing the reasoning behind every decision.

View case study

Turning a design system constraint into a shared way of working

ReVolve started with a practical delivery problem: teams were solving the same interface challenges independently, without an owner, roadmap or dedicated engineering capacity. I turned that constraint into a problem-solving strategy that connected real product work to reusable patterns, clearer decisions and adoption across teams.

View case study

Making a complex candidate journey explain itself

A ground-up rebuild of the candidate screening experience, separating the constraints we had to respect from the legacy behaviours we no longer needed to inherit.

View case study

All case studies →

Work / Design Systems

Design Systems / Systems / Accessibility / AI readiness

Making design-system knowledge usable by people and AI

Project meta

Role: Lead Product Designer
My contribution: I designed the system’s accessibility, canonical knowledge and AI governance architecture, enabling AI to work from trusted sources, recognise when change was needed and support governed evolution while keeping authority with the team.
Timeline: 9 months
Team: 1 PM, 1 Designer, 2 Engineers

Andrei Stanescu

4 September 2026

Design systems · AI-enabled design operations · Accessibility · Component architecture · Design governance · Storybook · AI-assisted workflows

OUTCOMES

A design system designed to outlive its creator

300+Interconnected documentation pages

6Codex skills for accurate system use

2Canonicalization workflows for preview and update

4Connected system surfaces across design, documentation and code

1Operating model designed to preserve quality after handover

Write-up in progress

The results are in. The story behind them is still being written.

This case study is being written up in full—the challenge, the shift in approach, and what was built. The outcomes above are real and already delivered; the full narrative is still being prepared.

RELATED WORK

More case studies

Designing the evidence AI needs to improve

I designed a human-in-the-loop audit platform that helped subject matter experts resolve disagreements between researchers and an AI agent—while capturing the reasoning behind every decision.

View case study

Turning a design system constraint into a shared way of working

View case study

Making a complex candidate journey explain itself

A ground-up rebuild of the candidate screening experience, separating the constraints we had to respect from the legacy behaviours we no longer needed to inherit.

View case study

All case studies →