The stylist that already knows your wardrobe.

The brief

Most style advice online is generic. It doesn't know what's in your closet, what today's weather looks like, or where you're headed at 7 pm. The VeryChic team wanted the opposite: an assistant that works from the customer's actual wardrobe and puts the taste of a stylist behind every recommendation.

The output had to feel like a designer app first and an AI feature second — sophisticated typography, restrained motion, and copy that reads like a stylist wrote it, not a chatbot.

What I built

• Wardrobe upload. Customers add their own pieces — clothes, shoes, accessories — and the assistant learns what they actually own, not what the catalog assumes.
• Event-driven outfit generation. Enter tonight's occasion ("business dinner", "beach afternoon", "gallery opening") and the assistant returns a full look from the customer's own pieces, laid out visually.
• Concierge notes for every look. Each outfit ships with three short write-ups: a style note explaining the silhouette, a colour note explaining the palette, and an alternative so the customer sees the same base look reinterpreted.
• Tips carousel. Three quick styling tips per outfit — how to wear the coat, which watch pairs, how to style the hair — the small polish that makes a look feel considered.
• Weather widget. Live temperature and a one-line practical read ("Evening 14° · a coat is recommended") integrated into the recommendation logic.

Design language

Dark background, gold accent, generous white space, serif display + sans body — the visual language of a luxury magazine, not a fitness tracker. Every card has a look preview, an AI stylist avatar block, and a soft-shadow surface that feels tactile. The whole flow was designed to feel like opening a private clientelling channel at a boutique.

How the AI is applied

The assistant combines three signals into every recommendation: the customer's own wardrobe (tagged by category, colour, formality), the event context they typed in, and the day's weather. The output is deliberately shaped — three named sections (Style Note, Colour Harmony, Alternative Suggestion) — so the customer can skim and act quickly instead of scrolling through a paragraph.

Behind that, the same system generates a set of daily tips and an optional "look of the day" push, keeping the app useful even on days the customer doesn't open it with a specific outfit in mind.

Analysis

The point of the app isn't to replace the customer's taste — it's to reduce the decision cost of getting dressed. If the assistant returns a coherent, weather-appropriate look built from the customer's own wardrobe, in twenty seconds, wearing something new stops feeling like homework.

For the VeryChic brand the app also opens a natural upsell channel: whenever a recommended look calls for a piece the customer doesn't own, the alternative suggestion can pull from VeryChic's own catalog. The AI is the recommendation surface; the wardrobe is the loyalty layer.