My role: Founder & product designer: research, product strategy, design system, and build direction. Building it with my cousin, Joey Schroeder.
The problem
We wear about 20% of what we own. Existing wardrobe apps just help you catalog your closet, then hand the hard part, the styling decision, right back to you.
Somara makes styling what you already own the entire product. Shopping comes second, and only when there's a real gap.
What I did
18 interviews and usability tests, synthesized into per-participant field guides. Two findings recurred: the "combination problem" (people own the pieces, can’t combine them) and an onboarding "upload wall."
Styling before shopping, a Safe / Bold / Risky variable reward (named by a participant), and privacy as a design requirement, not a footnote.
A two-layer token model with an automated Figma→code pipeline, so design and code never drift.
A programmatic WCAG 2.1 AA audit across 30 token pairs, 20 screens, and 16 components, remediated to 100% pass.
Treated AI as a team of specialists (Gemini, Claude, Cursor, Lovable, Nano Banana / Fal) and shifted from writing prompts to orchestrating agents.
The outcome
A research-validated direction, a working React Native prototype, a real design system underneath, and a 100% accessibility pass. Usability testing caught the biggest mistake (onboarding asked for investment before it delivered any value) which I fixed by resequencing the flow, not adding features.
AI is confidently wrong often enough that the whole job is staying close enough to catch it. No tool caught my biggest mistake. Three users did.
What's next: resequence onboarding so value comes first, build the brand-catalog path users asked for (now wired to the ASOS API so people can browse by brand and tap what they already own), and keep earning the trust the privacy line promises.
Where it stands
This is Part 1. Somara isn't finished, and that's the point: it's the first in a series documenting a product designer's journey from idea to app, and how AI fits into the work.
- Onboarding
- Home
- Avatar
- Closet
- Plan
- Gaps
- New design system + component library (react-native-reusables, shadcn/ui)
- Image tagging
- Conversation flow design
- Onboarding enhancements
- Live API calls (currently the ASOS API)
- More usability testing
Key decisions & trade-offs
Made styling what you already own the whole product and pushed shopping to second, rejecting the "sell you the next thing" model the funded players use. A smaller monetization surface, but it’s what users ranked #1.
After usability testing, resequenced onboarding to deliver an outfit from three items before asking for the full wardrobe upload. Less data up front, but it clears the "upload wall" that kills retention.
Capped the reward at Safe / Bold / Risky. Fewer choices than competitors show, but more would recreate the exact paralysis the app exists to cure (Hick’s Law).
Components reference state tokens, never raw colors, synced Figma→code automatically. Deliberate setup cost up front in exchange for a system that never drifts.
Skills demonstrated
Want the full story?
The research, the tokenization, the prompting, the agents, and the image tools, with all the seams showing.
More case studies