Beauty & Cosmetics
USA
D2C
Managing a beauty collection can be overwhelming, so we
leveraged AI to streamline the process.
Dew is an AI platform that helps you keep
your beauty products organized, learns about your
preferences, and provides you with suggestions regarding
future products according to your personal taste and skin
type.
Lexogrine has been part of the journey from
the simple idea to the whole application development. We
took full ownership of the technical execution, delivering a
seamless mobile and web experience powered by custom AI
algorithms and a robust cloud infrastructure.
Spanning the entire process: from fine-tuning complex AI models to designing a delightful UI/UX.
Identifying the relevant email
Parsing emails and digital receipts from various retailers to automatically populate the product database.
Designing the model
Engineering a specialized AI model for precise email parsing and product categorization.
Creating appealing UI
Designing a beautiful user interface and UX to match the beauty market and user expectations.
Delivering Smart Recommendations
Leveraging the user’s personal product history to generate beauty recommendations tailored to their unique needs.
Our approach combined the precision of AI with a deeply human-centered interface.
Training on diverse datasets
We trained the model using a vast range of email formats and order confirmations. This diversity enabled the AI to establish robust patterns, allowing it to identify and extract the correct messages with high precision.
Robust architecture & infrastructure
We built the core application using TypeScript, utilizing React Native for mobile and Node.js for the backend. The entire infrastructure is powered by AWS to ensure high availability and computing power.
Designing for user delight
Understanding the nuances of the beauty market, we aimed to create an environment that feels familiar to enthusiasts. We centered the design around a soft color palette and custom iconography, giving the application a distinct yet delicate visual identity.
AI-driven personalization
We developed a dedicated AI engine to analyze each user’s inventory. The system not only predicts which beauty products would best complement their collection but also actively scouts for relevant discounts.
eCommerce best practices
We integrated established eCommerce design patterns to minimize the learning curve, ensuring a familiar and frictionless shopping journey that is elevated by intelligent AI personalization.
We built a unified, cross-platform ecosystem using TypeScript, React, and React Native to ensure seamless performance on all devices. The robust backend relies on Node.js and MySQL hosted on AWS, while PyTorch drives the core pattern recognition intelligence.
TypeScript
React
React Native
Node.js
MySQL
AWS
PyTorch
One of our primary challenges was balancing the constraints of an MVP with the desire to create a delightful user experience. By carefully prioritizing the feature scope and making strategic design choices, we moved beyond a "viable" product to create one that was genuinely useful and lovable from day one.
We adopted a unified TypeScript architecture across the entire stack, allowing our developers to switch seamlessly between frontend and backend tasks. We utilized NestJS to structure the code efficiently and avoid "reinventing the wheel", ensuring rapid development. For the core intelligence, PyTorch powered our robust pattern recognition module, enabling precise detection and cataloging of user products.
