A Collaboration for Sustainable Innovation: Neworn GmbH, a secondhand marketplace for baby and children’s items, partnered with the nonprofit research center SWISDATA to develop a novel recommendation system. Supported by the FFG Innovation Voucher, this collaboration aimed to address a critical challenge: traditional recommendation systems are ill-suited for children’s clothing due to the rapid and predictable growth of children, which renders sizing irrelevant over time. By integrating novel neural network technologies and Large Language Models (LLMs), SWISDATA helped Neworn create a solution that not only enhances user experience but also supports sustainability. The result is a sophisticated system that helps parents in Austria and Germany discover, buy, and resell high-quality secondhand items with ease, contributing to a more conscious and responsible future.

Screenshots from the Neworn app (available for iOS and Android)

SWISDATA’s work focused on developing a proof of concept (PoC) that combines cutting-edge AI models to deliver personalized recommendations. The system uses a customized Two-Tower model to learn user preferences and calculate product similarity scores. To address the unique challenge of children’s growth, a growth prediction model trained on statistical data predicts body height, chest and hip circumference, and shoe size. This ensures that recommended items remain size-appropriate over time.

Additionally, a locally hosted open-source LLM enriches product listings by categorizing them into seasonal classifications, enabling recommendations that align with both growth patterns and seasonal needs. For example, a winter jacket might be recommended as one size too large in November–December, perfectly fitting until February, and snug until March. Each recommendation includes a timeline showing how well a product will fit over months, allowing parents to shop in advance. The system also incorporates a fit prediction model that evaluates brand-specific cuts, sizing, and fit to score clothing alignment with a child’s anatomy and parent preferences.

The PoC, now deployed on SWISDATA’s infrastructure, synchronizes data with the Neworn marketplace to generate recommendation models and use statistical data for predictions. It is fully functional and undergoing evaluation for integration into Neworn’s live system. Key innovations include modular design, enabling future additions like quality checks, and a circular system for secondhand clothing. By predicting not only current needs but also resale opportunities, the system supports a sustainable cycle of reuse.

SWISDATA’s expertise in AI-driven R&D projects has empowered Neworn to reduce user browsing time, increase sales conversion rates, and offer a seamless shopping experience. This collaboration highlights how nonprofit research centers like SWISDATA can drive innovation for European companies, transforming challenges into opportunities through advanced technologies.

Neworn’s partnership with SWISDATA, supported by an FFG Innovation Voucher, has resulted in a novel recommendation system that redefines secondhand shopping for parents. By leveraging AI, growth predictions, and seasonal classifications, the system ensures that recommendations are not only personalized but also practical and sustainable. This success story underscores SWISDATA’s role in advancing European companies through R&D projects in Artificial Intelligence. For businesses seeking to innovate with AI, the message is clear: collaborate with experts like SWISDATA to turn challenges into competitive advantages.

Take action today! Explore how SWISDATA can support your company’s R&D goals in AI and help you deliver smarter, more sustainable solutions.