I help SaaS and fintech teams make their design systems AI-ready, so tools like Claude Code and Cursor generate UI that's actually on-brand. From Figma tokens to production code.
space.stack-xl
Daniel Vinersar · Product Design Manager · 10 years across frontend, UX and design systems. Currently leading Design Systems & AI across 10+ products.
color.action.primaryradius.controlfont-family.displayblue-500 tells AI what color to use, not when. So it guesses.
Button and PrimaryButton both exist. Nothing says which one is current, so AI picks one, and not always the same one.
Spacing logic, states, do's and don'ts: undocumented. Every generated screen drifts a little further.
The result: faster output, slower product. More review, more rework, more drift.
I asked an AI model to build the same three screens twice: once with a typical design system, once with one built for AI. Same model, same prompt. Only the system changed.
Pick a prompt · read my review under the panels · switch runs 1–3 to compare consistency
// prompt: Create a pricing card for our Pro plan: €29 per month, billed annually, with four features and a primary call to action to start a free trial. Mark it as the most popular plan.
Real components, but values instead of intent and no usage rules. Like most libraries.
For growing teams that need more power and flexibility.
€29/ month
Billed annually (€348 per year)
Daniel Vinersarblue-500, teal-600) and nothing says which one means action. All three unstructured runs made the CTA and the tag blue. All three AI-ready runs used Button with variant=primary, which is the brand color by definition.The AI-ready side used the brand color and the same core components in every run. The unstructured side made every main action blue instead of the brand color, and varied more from run to run.
Both outputs generated with the same model and the same prompt. The only difference is the design system it was given.
Read how this was madecolor.action.primary) instead of raw values (#3B82F6).For teams that want a clear picture, fast. Design-side only.
For teams ready to fix it properly. Design and code.
For teams that want senior design systems help every month, without a full-time hire.
The Review is fully credited toward the Audit if you book it within 30 days.
The Audit roadmap is a natural starting point for a retainer.
I reply within one business day.
You get a first read on where you stand.
I go through your system, design and code.
You get a prioritized plan and a session with your team.
A fitness SaaS with no Figma, where Claude co-authored 76% of the commits. It started on Tailwind, moved to a custom SCSS token system, and I audited the result the way I'd audit a client's. Here's what held up, what drifted, and what I'd do differently.
As of October 6, 2026
A design system built from scratch for one main site, four microsites and an internal app, with four brand themes on one core. The build took two months; adoption took much longer. Here's how it worked, how it got into code, and what I'd do differently.
July 2022 – March 2026, through AscentCore
I started in sales, moved into frontend development and spent six years writing production React. Then UX, then design leadership.
Today I'm a Product Design Manager leading Design Systems & AI work across 10+ products, and I take on a small number of consulting engagements at a time.
That path is the point: I understand design systems from the Figma file to the pull request, and I know what AI needs from both.
Most of my design systems work is under NDA, so what you see here shows the approach, not the client.
LinkedInYes. Your team will, and an AI-ready system is also a cleaner system for humans.
Yes. I come from frontend development, so the conversation goes both ways.
A small number at a time, so every client gets my full attention, not a junior team.
Find out in 3 minutes, or talk to me directly.