- Project:
- AI-Powered Retail Assistant Robot
- Context:
- Software Development Internship at Bumerania
- Tech Stack:
- React Native (Expo), TypeScript (Node.js & Fastify), MongoDB, Google Gemini AI
Overview
The AI-Powered Retail Assistant Robot is an innovative solution designed to transform the in-store shopping experience for clothing retailers. This project combines artificial intelligence and computer vision to deliver a more seamless, personalized, and efficient shopping journey.
This robotic assistant enables customers to scan items via QR codes to check availability and explore options like in-store pickup, warehouse requests, or online purchases. It also features voice search capabilities and advanced recommendation systems based on both scanned items and the customer's current outfit.
- React Native
- Expo
- TypeScript
- Node.js
- Fastify
- MongoDB
- Google Gemini
- TensorFlow.js
Key Features
Product Scanning: Allows customers to scan QR codes on clothing items to retrieve detailed information about availability, sizes, colors, and purchasing options. The robot can also walk the customer straight to the product inside the store.
Voice Search: Integrates a voice recognition system that enables customers to search for specific products using spoken commands, facilitating navigation.
Intelligent Recommendations: Utilizes AI algorithms to offer personalized suggestions based on both specific items and the customer's complete outfit.
Omnichannel Experience: Bridges the physical and digital experience, allowing options like in-store pickup, warehouse requests, or online purchases from the same device.
Technical Development
The frontend was developed with React Native (Expo) to ensure a fluid and responsive interface on the robotic device. The backend was built with TypeScript using Node.js and Fastify to create a fast and efficient API. MongoDB was implemented as the database to manage inventory, user preferences, and recommendation data.
The integration of computer vision technologies allows the robot to recognize clothing items and analyze the customer's current outfit to improve the accuracy of recommendations and offer a truly personalized experience. We leveraged Google's Gemini AI model to power the visual analysis, and TensorFlow.js for body detection and outfit capture.
Impact on Retail Sector
This project represents a significant innovation in the retail sector, combining the convenience of e-commerce with the tangible experience of physical stores. Initial results show increased customer satisfaction, greater shopping efficiency, and an increase in cross-selling thanks to personalized recommendations.
Team & Acknowledgements
Built during my software development internship at Bumerania (Alicante, March - June 2025) together with Jesús Martínez Manchado, Alberto Rodríguez and Vicente Hernández Maciá. Thanks to Bumerania for providing the hardware that made this independent project possible.
The conceptual UI was inspired by Zara and created for educational and portfolio purposes only. This project is not affiliated with, endorsed or approved by Zara. Zara is a registered trademark of its respective owner.