From Data to Decisions: Leveraging Large Language Models and Digital Product Passports for Data-Driven Remanufacturing

Remanufacturing plays a key role in advancing the circular economy by extending product lifecycles and reducing resource consumption. However, its effectiveness is often constrained by incomplete lifecycle data and high uncertainty in the returned products. Digital Product Passports (DPPs) enhance transparency and traceability, but their heterogeneous, unstructured, and dynamically evolving data within them limits the effectiveness of obtaining actionable insights for remanufacturing. Generative AI (GenAI), particularly large language models (LLMs) ofer new capabilities for interpreting heterogeneous data, maintaining long-term context across circular production, including collection, disassembly, re-manufacturing, testing, and redistribution. This paper explores the convergence of DPP, GenAI, and remanufacturing, evaluating their synergies and integration challenges. We propose a generative agent-based model where LLMs interact dynamically with DPPs to support real-time quality assessment, operational planning, instruction generation, and interactive diagnostics. Finally, we identify key research directions and technical challenges for building interoperable, GenAI-enhanced platforms that realize data-driven, DPP-enabled remanufacturing.