BetaMay 1, 2026

Vessura

A digital wardrobe that thinks like a stylist. Tag once, get daily outfits adapted to your weather, calendar, and how recently you've worn each piece — built on Next.js + OpenAI.

Next.jsTypeScriptSupabaseTailwind

Source

Yashwanth Kamireddi · Designer · Full-Stack Developer

Overview

Vessura is a digital wardrobe designed to feel like a private atelier. It catalogs the pieces you already own, learns how you wear them, and surfaces complete outfits — calibrated to the weather, the calendar, and your taste. The product thesis is simple: most people own enough clothing to dress beautifully every day; what's missing is the editor.

The problem

The average closet is wildly under-utilized — research repeatedly puts active rotation around 20% of what's hanging there. The reason isn't shopping habits; it's friction. People can't recall what they own at the moment of decision. Existing wardrobe apps treat the closet as a spreadsheet. Vessura treats it as a collection worth showcasing.

What it does

Catalog with intent

Photograph each piece once. Vessura auto-tags by category, color, season, and silhouette, and stores a clean editorial-style render so the wardrobe looks as good as the clothes do.

Outfits, not items

The recommender composes complete looks — top, bottom, layering, shoes, accessory — instead of leaving you to assemble them. Suggestions adapt to today's forecast, your week's calendar, and how recently you've worn each piece.

A preference signal that learns

Every accept, reject, and wear event updates a lightweight preference signal — so suggestions sharpen over time. Fewer suggestions, better hits.

Inspiration without noise

A curated discovery feed surfaces ways to re-style what you already own, instead of pushing you to buy more.

Design philosophy

Vessura was built around a "Digital Atelier" design system: heavy negative space, editorial typography, soft motion, and no decorative chrome. The interface gets out of the way so the clothes can speak. Every screen — from onboarding to the daily look — was prototyped against the same question: would this fit in a luxury magazine?

Stack

Frontend — Next.js (App Router) + TypeScript + Tailwind, motion via Framer
AI — OpenAI for tagging, outfit composition, and natural-language wardrobe queries
Data — Supabase + Drizzle ORM with multi-tenant isolation
Imagery — automatic background-removal pipeline for editorial product shots

What's next

Try-on previews using diffusion models, capsule planning for travel, and a Capsule Score metric that rewards owning fewer, better pieces.

Gallery

Back to all projects