StockSight | Devpost

The Inspiration

Our journey began with a simple observation at a local grocery store. We saw a manager walking up and down aisles with a clipboard, manually counting cans of soup. Simultaneously, a customer left the store frustrated because the item they saw "In Stock" on the website was actually missing from the shelf.

Despite living in the age of AI, the physical shelf—the most important point of sale in the world—is still "offline." We realized that stores are currently "blind" to their own inventory until it's too late. We asked ourselves: “What if the shelf could speak for itself?” The Development Journey

We didn't want to build just another security camera. We wanted to build a Retail Agent.

Using Kiro’s AI-powered coding environment, we rapidly prototyped a bridge between high-speed Computer Vision and the IoT world. We chose the TuyaOpen Platform because of its unparalleled ability to connect diverse hardware. By integrating our custom YOLOv8 model with the Tuya Link SDK, we successfully turned a standard Raspberry Pi into a native Tuya device.

The "Aha!" moment came when we connected our Tuya-enabled shelf to AWS Bedrock. Suddenly, we weren't just looking at data points on a dashboard; we were having a conversation with the store. We could ask, "Which aisle needs attention?" and the AI Agent, fed by real-time Tuya sensor data, would answer instantly. The Impact

RetailTwin AI transforms the retail experience:

For the Manager: No more clipboards. The store "calls" them when it needs help.

For the Business: Reduced "Out-of-Stock" losses and optimized staff labor.

For the Ecosystem: It proves that with Tuya and AWS, building sophisticated, commercial-grade AI hardware is as easy as stacking blocks.

The Future

We see a future where every shelf in every store is a "Retail Twin." By leveraging the global Tuya AI+IoT ecosystem, we aim to scale this solution from a single local shop to global retail chains, making "Out-of-Stock" a thing of the past.