Rebuilding
From a scattered MVP to an integrated
shopping & styling ecosystem, powered by AI.
Vêtir launched with a big vision: shopping, professional styling, personalization, and AI. The ingredients were there, but the experience had grown feature by feature, without a clear product language connecting everything together.
Shopping wasn't as effortless as it could be. Styling was difficult to recognize as a core part of the experience. And the early AI Stylist felt more like search than an intelligent styling partner.
Across the product, inconsistent visual patterns, weak hierarchy, fragmented navigation, and disconnected experiences made Vêtir feel more like a collection of features than one product.
As the sole designer, I set out to change that.
The goal wasn't simply to make Vêtir look more luxurious. It was to make the product easier to understand, easier to use, and much clearer about the value of styling and personalization at its core.
I rebuilt it module by module, inside the existing roadmap. As the product evolved, so did the experience—from a rudimentary AI layer and fragmented shopping flows toward richer styling experiences, including AI-assisted outfit creation and avatar try-ons.
Vêtir had strong individual features, but no shared language connecting them.
The product had grown quickly, and every new feature added another pattern. The result was an experience that felt inconsistent visually, structurally, and functionally.
ios / closeteverything in
ios / aiAI help lived inNo shared language. The app used Helvetica, the web mixed Libre with assorted fonts, and there was no type hierarchy, which also hurt SEO. For a luxury fashion platform, the result felt more mass-market than it should have.
Engineering had developed its own component set independently of design. Every new feature introduced another pattern, making the experience harder to keep coherent.
Strong features in silos. Shopping and styling both existed, but neither felt as seamless as it could be. Navigation hid parts of the product. Important functionality was difficult to discover. And the styling proposition wasn't immediately obvious.
Each feature worked on its own.
Together, they didn't add up to one product.
Navigation hid the product's range, so existing features went unused.
Google Analytics · UX audit · Product feedback · Competitor analysis
A stylist serves you a look. Search hands you a list. Every AI feature was tested against this, and wherever something returned a static list when it could collaborate with the user instead, that became the problem to solve.
Minimal and editorial: imagery does the emotional work, restraint does the credibility work. Every new feature is built inside the system, not bolted on as an exception.
Traction and revenue were low, and people were confused about what Vêtir actually offered: a store, a styling app or a stylist marketplace?
Adding more features felt like the obvious fix. I needed to show that the real problem was the experience holding those features together, and make that case on business terms, not aesthetics.
I mapped where the experience broke down, where functionality was difficult to discover, and where the visual language felt inconsistent with the luxury brands Vêtir sold.
Rather than proposing a disruptive redesign of the entire product, I used the Outfit Module as the trial. It was a high-value part of the product where the new approach could be demonstrated clearly.
The ops team, who onboard clients and stylists every day, started relaying positive feedback. The redesign continued module by module. The redesign became the way the roadmap was executed.
A big-bang redesign wasn't realistic for a five-person engineering team at our funding stage.
Instead, every incoming feature request became the entry point to redesign its whole module in the new system. The redesign absorbed the roadmap rather than competing with it. Many smaller flows shipped between these milestones.
For months, two design systems ran side by side. Keeping the product feeling coherent across converted and unconverted pages took constant judgment calls. High-value surfaces went first, internal tools last, and the AI Stylist's agentic flows were scoped into phase two. The old system is now fully retired.
Black and white carry the product. Gold is reserved for AI, so users recognize "the stylist is here" at a glance. Nine warm neutrals and muted status colours keep the editorial mood.
Type App styles at actual size, headings in the order they appear on a page. On web the same roles scale up: Display 48, Banner title 40, Page title 80, Section heading 24.
16
20
24
32
Three audiences, three homepages. Shoppers and clients open on weekly try-ons across diverse avatars, followed by stylist picks, new arrivals, edits and a direct way into the AI Stylist. Guests get a lighter introduction to the product.
Deep dive · coming soonEvery screen scrolls on its own. Hover to pause, click for the full page.
Auto-arranged compositions, auto-filled outfit details, image-based search, looks built from a reference photo, and a calendar to plan ahead. The canvas is a flat lay, so that's how it's shown here.
Deep dive · coming soon
Drag to compare the real screens. Plus an 8–10s recording of the AI Stylist inside the canvas.
Built around a stylist's actual job: performance and client roster first, turning data into next steps. This is the answer to "Won't I rather just use Excel?"
Rolling the system out module by module meant components arrived gradually, instead of a disruptive handoff. Engineering adopted the new components, and once we added an AI-assisted Figma-to-code workflow, variables and components mapped almost one-to-one into code.
Stylists were Vêtir's growth engine: each one brought their own clients. So the product grew a toolkit around a stylist's working day: commissions on every client purchase, digitized client closets, AI-assisted styling, a look calendar, invoicing and a performance dashboard.
Together, they made Vêtir a place where a stylist could run their whole business.
Stylists on the platform (~14 → 190), as it became something stylists could run their business on.
More search traffic and orders after the type and hierarchy overhaul.
Lower drop-off, higher engagement. More active users, longer sessions.
7.4M interactions since launch.
"Personalized, chic, and finally in line with luxury fashion."
Users & stylists paraphrased, real quote?