# Nuance

- **Event:** [Gemini 4 NYC Hackathon](https://cerebralvalley.ai/e/gemini-4-nyc)
- **When:** Sat, Nov 21 at 9:00 AM – 10:00 PM (EST)
- **Where:** New York, NY
- **Team:** [Kelly Zheng](https://cerebralvalley.ai/u/kzzz), [Kira Cui](https://cerebralvalley.ai/u/kirakiracui)
- **GitHub:** https://github.com/KkZzzzzzzz/gemini
- **Demo video:** https://youtu.be/3krjPfk5eC8
- **Gallery:** https://cerebralvalley.ai/e/gemini-4-nyc/hackathon/gallery
- **Page:** https://cerebralvalley.ai/e/gemini-4-nyc/hackathon/gallery/5

Nuance — Find what you love. Change what you don't.

Have you ever found a product you almost loved, except for a few small details?

Maybe you love a jacket's texture and style, but wish it were longer and came in black. Traditional shopping search forces you to start over with generic keywords, often losing the specific qualities that made you like the original product.

Nuance lets you start with something you already love and find a version that better matches your preferences.

Users can discover products through four entry points: Search, Live Camera, Link, and Photo. Using Gemini 4's multimodal capabilities, Nuance analyzes the original product and breaks it down into editable attributes such as color, material, shape, and length.

Instead of writing another search query, users can directly choose which features to keep and which to change.

For example, our live demo starts by capturing a fur jacket through the camera. We keep its fluffy texture and overall style, but change the length to longer and the color to black.

Nuance then searches Shopify's cross store product catalog for real alternatives. It evaluates how closely each product matches the user's preferences and highlights the differences, helping shoppers understand the trade offs between options.

Unlike AI generated product concepts, our recommendations correspond to real products that shoppers can actually purchase.

Under the hood, we use Gemini 4 Argon for visual product understanding, Gemini 4 Neon for evaluating alternatives, and the Decisions API to rank recommendations based on the user's constraints.

Our goal is to move shopping beyond keyword search and chatbots, toward a more visual, intuitive experience where people can express exactly what they like and what they'd change.

Find what you love. Change what you don't.

## More from Gemini 4 NYC Hackathon

- [Goal Goose](https://cerebralvalley.ai/e/gemini-4-nyc/hackathon/gallery/2)
- [ClearLedger](https://cerebralvalley.ai/e/gemini-4-nyc/hackathon/gallery/3)
- [AgenticSwarms](https://cerebralvalley.ai/e/gemini-4-nyc/hackathon/gallery/4)
- [Showunce](https://cerebralvalley.ai/e/gemini-4-nyc/hackathon/gallery/6)
- [Murder Board](https://cerebralvalley.ai/e/gemini-4-nyc/hackathon/gallery/7)
- [Patient Engine](https://cerebralvalley.ai/e/gemini-4-nyc/hackathon/gallery/8)

---

Markdown version of https://cerebralvalley.ai/e/gemini-4-nyc/hackathon/gallery/5. Site index for agents: https://cerebralvalley.ai/llms.txt · full text: https://cerebralvalley.ai/llms-full.txt
