Optimizing Startup Team Composition: A Generative AI and Genetic Algorithm–Based Approach

Abstract

Founding team composition is widely acknowledged as a key determinant of startup ventures success. Entrepreneurship scholars have examined founding teams’ characteristics through various perspectives, including human and social capital, psychological traits, personality, and diversity. Despite the recognized importance of team composition, traditional team formation methods often rely on convenience or heuristic-based combination. These approaches frequently overlook the systematic alignment of critical attributes such as complementary skill sets, personality compatibility, and shared motivations, leading to suboptimal outcomes and ultimately impacting team dynamics and performance. Given the transformative impact of generative Artificial Intelligence (AI) across multiple entrepreneurship domains, we investigate how Large Language Models (LLMs) can enhance the process of founding team formation by making the most of individual profiles. By employing advanced prompt engineering techniques, we properly instructed an LLM to develop a genetic algorithm to iteratively optimize team formation by evaluating and refining groups of three based on multiple criteria. The designed algorithm considers multiple sets of team configurations and efficiently converges on team assignments by applying selection, crossover, and mutation over subsequent generations to maximize demographic variety, hard and soft skills variety, personal value and motivation similarity and personality compatibility (based on the Big Five model). To facilitate reproducibility, we provide detailed prompt engineering strategies for implementation. To validate our method, we conduct an empirical study within a higher education entrepreneurship course where students are required to work in teams. We evaluated each team composition across seven metrics and assigned each team an overall score. Results suggests the effectiveness of our AI-assisted team formation approach. Our study addresses the growing interest in integrating generative AI into entrepreneurship research, offering insights to multiple stakeholders. While we tested our approach in the educational context, its applicability extends also to practitioners, including aspiring entrepreneurs searching for co-founders, startup studios forming founding teams, and organizations enhancing recruitment strategies.