Transdisciplinarity drives innovation, particularly in fields requiring novel approaches, such as energy technologies. This study introduces a method for assessing the transdisciplinarity of startups by analyzing employees’ research backgrounds. We leverage Natural Language Processing (NLP) techniques, using predictions from a document classification model, to classify articles from scientific literature. Using Open Corporates and Semantic Scholar data, we identify scientific publications linked to these individuals and classify their articles into research fields. This model, used to classify articles on Semantic Scholar, and available on HugingFace, refines classification and enhances understanding of interdisciplinary connections. By detecting publications that bridge multiple research categories, we develop a transdisciplinarity index to evaluate startups’ innovative potential and capacity to integrate diverse knowledge domains. A taxonomic approach structures research fields hierarchically for a more systematic assessment. Focusing on renewable energy startups, we apply our methodology to uncover hidden patterns in knowledge integration, providing a robust framework for evaluating the transformative potential of emerging companies in this sector.