<p>Multiprofessional collaboration is essential in palliative care (PC), where diverse healthcare professionals work together to support terminally ill patients. However, current hospital information systems often fail to support this collaboration effectively, especially when it comes to leveraging the vast amount of routinely collected, unstructured data such as narrative clinical documentation. In this study, we explore how state-of-the-art natural language processing (NLP), particularly large language models and multi-agent artificial intelligence (AI), can utilize PC routine data to improve multiprofessional case reviews in PC. Following a design science research approach, we developed and evaluated a multi-agent AI-based prototype that enables contextual information retrieval, symptom-centered visualizations, and synthesized patient reports. The resulting nascent design theory comprises three artifacts: a working prototype, a reference architecture with orchestrated agent workflows, and 14 technological rules. Our results help address challenges in multiprofessional collaboration in PC, including asynchronous communication, fragmented data, and role-specific information needs. By making routine data more accessible, interpretable, and actionable for multiprofessional teams, this study offers theoretical and practical guidance for designing NLP-enabled information systems that support multiprofessional sensemaking, with potential applicability beyond PC to other collaboration-intensive healthcare settings. At the same time, our findings are constrained by small sample sizes, limited field use, prompt-related variability, and reliance on a single AI provider (i.e., OpenAI). A promising next step is a field deployment of NLP-enabled information systems in routine multiprofessional case reviews, accompanied by pre- and post-measures to assess their impact on collaboration quality and clinical decision-making.</p>