Agentic AI-driven closed-loop analytical systems for autonomous pharmaceutical method development and regulatory-aware analysis

Agentic AI is investigated in relation to autonomous closed-loop pharmaceutical analytics within the scope of this review. It will cover frameworks for agentic AI-based pharmaceutical analytics as well as their systems architecture, incorporation into regulations, and practical application areas. Standardized and even AI-based methods used in pharmaceutical analysis still involve certain constraints related to static structure and human involvement. Agentic AI-based systems are goal-directed and self-adaptive, providing decision-making capabilities and optimization processes within closed-loop structures covering designing, executing, evaluating, and improving. Conceptual models of agentic AI-based systems include multi-agent system architecture, systems’ lifecycle validation model, and regulation-compliant AI. Potential benefits relate to reduced method development time up to 30–50%, increased robustness, and real-time performance control. Nevertheless, the following issues need to be solved: validation of an adaptive system, data integrity, lack of explainability, and difficulties with gaining regulatory approval. Prospects include the use of digital twins, federated learning, and explainable AI.