Teach, don’t just recommend
Every suggestion includes a plain-language reason. If the model cannot explain a credible pairing, it should not suggest it.
Your wardrobe.
Rediscovered.
An AI wardrobe stylist that helps people wear what they already own, instead of selling them something new.
Product strategy / AI experience / iOS design02 / Product strategy + iOS design
excessorizePersonal style.
Not another shopping app.
Not another shopping app.
More you.START WITH WHAT YOU OWN.
The problem
Most people own more accessories and clothes than they actually wear. Pieces sit unused because pairing them takes taste and confidence. Existing styling apps often push products to buy and rarely explain why a pairing works.
Excessorize surfaces what you already own, removes the pairing guesswork, and teaches the reasoning so the user gets better over time.
“You already own great outfits. We’ll show you how to wear them.”
My role
I wrote the PRD, built the brand, designed the full system, wrote the UX copy, and made the technical and privacy architecture calls. Product thinking came first, AI multiplied execution, and I remained the quality gate on every pass.
Process
I started with a real PRD before touching a screen: problem, goals, non-goals, user flows, MVP scope, and the two risks that decide whether the product feels magic or mediocre, recommendation quality and AI vision reliability.
- Each iteration began with layout, hierarchy, and state-level direction, not a request to “make it nicer.”
- Every pass ended with an audit. The first found a cold-start problem; a source-code scan later caught missing safe-area handling and repeated header treatments that made the work feel generic.
- A competitive scan across Whering, Indyx, Acloset, Stylebook, and Cladwell confirmed the teaching layer was open territory and that time-to-value needed to be faster.
- A dedicated UX-writing pass set the voice: an encouraging best friend with taste. The AI stays humble, benefits lead, and a forgotten item becomes a “forgotten favorite,” never “underused.”
I also killed an entire aesthetic direction. The magazine energy of v5 looked promising on paper, but felt thin and dated in use. Cutting it made room for a more native, confident system.
Key decisions
Make the AI visible and editable
Style DNA lets people see, correct, and reset the taste profile. Training becomes a trust feature instead of a black box.
Context over fine-tuning
An editable profile and feedback memory are passed as generation context. The architecture choice is what makes the trust feature possible.
Switch to native when the experience demanded it
A guided live viewfinder and richer iOS motion required SwiftUI. The product experience mattered more than staying inside a familiar stack.
Favorites-first onboarding
Five favorite pieces lead to a real outfit in under three minutes. Value first, cataloging later.
Outcome
Excessorize is design-complete: a full Figma system with 17 color roles across light and dark modes, a type scale, component library, working HTML prototypes from v6 through v12, brand guidelines, the PRD, architecture documentation, and a written build handoff.
The cover shown here is a typographic study for this portfolio. It is not presented as a screenshot or generated product render.
What I’d do next
- Build the capture, tag, and outfit loop from the handoff brief.
- Test recommendation quality against real wardrobes before polishing anything else.
- Measure time-to-first-outfit with a small user group.
Next projectExplorite
↗