AI · Consumer Shopping
Glance
An intelligent shopping agent that learns your style, predicts trends, and turns any screen, phone, TV, lockscreen, into a personalised shopping surface.
Role
Senior Developer
Year
2024
Timeline
9 months · team of 12
Stack
React Native · Python · Vector DB
4.9 ★
App rating
Phone · TV · Lockscreen
Surfaces
Tech stack
React NativePythonVector DBGenAI
The problem
Shoppers were tired of generic feeds and endless scroll. Retailers wanted a way to reach intent-rich users without paying the ad-network tax. Both sides needed a smarter surface between browsing and buying.
Approach
- Built a style-embedding model that captures each user's taste from swipes, saves, and conversational feedback.
- Fused a generative try-on pipeline with a live catalog of 1M+ SKUs from official brand stores.
- Shipped the experience across lockscreen, phone, and TV, same personalised feed, different form factor.
- Instrumented an evals harness so every model update ships only after beating the previous week on CTR and NPS.
Results
- 4.9★ average app rating across Galaxy Store, Play Store, and App Store.
- Personalised feed drives materially higher click-through than generic retail placements.
- Deep partnerships with global brands to sell directly through the agent, not ads.
See it in production
Explore the live product to see how it comes together.
