Inspiration
Online beauty shopping still has a confidence problem. Shoppers often hesitate because they do not know how a look, shade, or product bundle will actually work for their face, occasion, and budget. We wanted to turn AI/AR try-on into a complete shopping journey, not just a visual effect.
What it does
GlowCart AI lets a shopper choose or upload a selfie, select an occasion, set a budget, and run a live AI Look Virtual Try-On using Perfect Corp’s YouCam API. It then generates a personalized shoppable beauty bundle, explains why each product fits, and shows retailer signals like conversion intent, shopper segment, bundle value, and fallback status.
How we built it
We built GlowCart AI as a lightweight web app with HTML, CSS, JavaScript, and a Node/Vercel backend proxy. The frontend handles the shopping experience, try-on preview, product bundle, Judge Mode, and fallback demo. The backend protects API credentials and calls Perfect Corp endpoints for template loading, look try-on task creation, and polling.
Challenges we ran into
The biggest challenge was making a live sponsor API integration feel smooth under hackathon conditions. We solved this by adding a secure backend proxy, local and Vercel deployment paths, image compression before upload, and a fallback mode so the shopper journey still works even if an upstream AI/API call fails.
Accomplishments that we're proud of
We are proud that GlowCart AI is more than a mockup: it connects to live Perfect Corp templates and supports a real try-on flow. We also built the product around a clear retail use case, with shoppable recommendations, conversion signals, and resilient fallback behavior that make it feel like a practical B2B tool.
What we learned
We learned that AI/AR demos become much stronger when they are connected to a business workflow. Virtual try-on is compelling visually, but pairing it with shopper intent, product recommendations, and session recovery makes the experience more useful for both shoppers and retailers.
What's next for glowcart-ai
Next, we would add live Nemotron-powered recommendation agents, deeper product catalog integration, Shopify support, more precise shade matching from try-on results, and analytics for conversion lift, abandoned sessions, and post-try-on retargeting.
Log in or sign up for Devpost to join the conversation.