Gigi

Inspiration

Fashion inspiration is everywhere, but turning it into something you can actually wear is harder. A saved outfit might be beyond your budget, miss pieces you already own, or leave you wondering what would make your current look feel complete.

We built gigi around a simple question: “How can I make this outfit better?” Sometimes the answer is a new accessory. Sometimes it is styling something you already own differently.

What it does

gigi helps you improve an outfit from a photo, with your occasion, preferences and budget in mind.

You can review suggested changes, explore selected products from real retailers, choose an alternative you already own, and save your styling plan. Suggestions distinguish between keeping a piece, restyling it, adding something, and swapping it.

The app also includes a wardrobe planner and fashion and beauty inspiration. Its editorial design brings the experience closer to a personal style magazine.

How we built it

We built the interface with JavaScript, HTML and CSS, supported by a Node.js backend and deployed on Vercel.

For outfit understanding, we integrated OpenRouter with Gemini 2.5 Flash Lite. The server resizes uploaded images and requests structured styling advice. Product suggestions are matched to a reviewed catalogue, so prices and shopping links come from known records.

We also built a YouCam virtual try-on integration with explicit consent, garment selection, task tracking and result handling. An earlier local integration completed a garment render using provider demo images. Hosted try-on remains disabled while we work on durable task and session storage.

Challenges we ran into

Our biggest challenge was making the product easy to understand. Early versions asked people to navigate too many categories and planning steps. We refocused the main journey around one photo and one goal: improve the outfit.

Another challenge was connecting creative advice to practical shopping decisions. We needed to preserve budgets, count each product once, and let people use their own alternatives at no additional cost.

Virtual try-on introduced another layer of complexity: keeping the correct person photo, garment and rendering task connected, including when requests fail or return late.

What we learned

We learned that styling advice needs a clear explanation. People should understand what to change, why it helps, and whether buying something is necessary.

We also learned to distinguish between a styling recommendation, a visual preview and a fit guarantee. Each serves a different purpose, and the interface needs to communicate those limits clearly.

Accomplishments we're proud of

We brought together a distinctive editorial interface, a photo styling workflow, budget-aware product selection and a working local YouCam integration.

The app supports changes of mind: users can leave out suggestions, choose owned alternatives, undo changes and save plans without saving their photos.

Built With

• ajv
• css3
• gemini
• html5
• indexeddb
• javascript
• node.js
• openrouter
• sharp
• vercel
• youcam-api