Using Agentic Artificial Intelligence to Predict Industry Partnerships for Circular Ecosystems

Circularity has been suggested as a means of achieving a sustainable future by closing material and energy loops in regenerative systems. To avoid suboptimization and fully work in a circular manner, companies can form ecosystems where they collaborate on a meso and macro scale to capitalize on joint ventures. However, this poses a transdisciplinary challenge as companies need to find suitable partners where their strategic intents align, taking multiple aspects into account and finding partnerships which may not be obvious. Recent advancements in generative artificial intelligence have combined the reasoning capabilities of large language models with the dynamics of role-playing agents, thereby reducing the need for human guidance in conversations and actions. In this study, agents use annual reports to represent the interests of individual manufacturing companies considering potential partnerships. This agentic framework evaluates the circularity potential of these partnerships, ranking and scoring each for direct comparison. This work highlights the potential of artificial intelligence to effectively model complex behaviors at the scale of enterprise decision-making. Additionally, it raises new research questions about the time-dependent aspects of corporate strategy and macro-scale effects of local policies on incentives to form partnerships with circularity as a fundamental criterion. This information is crucial for companies to optimize potential partnerships and for policymakers and sustainability advocates to encourage joint ventures that align with broader sustainability goals in a transdisciplinary manner.