Prompt engineering for relational contracts

This academic article investigates how artificial intelligence (AI) technologies can facilitate the success of open innovation projects, particularly when participants exhibit overconfidence. Open innovation involves collaboration among diverse firms, necessitating coordination mechanisms such as relational contracts to address project uncertainty and varying boundaries. However, relational contracts can suffer from misunderstandings, particularly in the early stages. This design science research study proposes an AI-based solution to mitigate these challenges and support the formation and execution of relational contracts. Drawing on insights from organizational economics, psychology, and software engineering, the authors develop a multi-agent system that simulates negotiations. Our system, leveraging prompt engineering techniques, breaks tasks into sub-tasks to manage overconfidence and enhance clarity. We illustrate how to use our system by replicating the first stage of the empirical paper on relational contracts of Gibbons et al. (2023). Our research highlights the potential of AI to reduce misunderstandings and improve coordination in open innovation initiatives. Moreover, it suggests that overconfidence might not always be detrimental, potentially solving credibility problems in relational contracts once clarity is established. While the study has limitations, it offers valuable insights for both managers and academics studying open innovation and relational contracts.