From conventional monitoring to agentic AI: integration of data streams and computational tools for real-time microalgal surveillance

Microalgal cultivation generates continuous and diverse data streams, including physicochemical sensor measurements, optical and spectroscopic signals, and episodic molecular tests. Together, these data characterize cultivation conditions at different biological scales. Existing AI/ML frameworks can interpret these streams independently; however, human intervention is required to translate predictions into cultivation decisions, adding delays and limiting adaptability across strains, reactors, and environmental conditions. This mini-review describes the transition from conventional monitoring to agentic AI as the progressive integration of sensing, interpretation, and action. The real-time sensing modalities that constitute an agent’s perceptual field include AI/ML models and genome-scale metabolic models, which serve as callable tools in the proposed agentic architecture for coordinated data-model integration in algal systems biology. Emerging evidence from agentic systems in microbiome research and precision oncology supports the conceptual and technical plausibility of a multi-agent, tool-enabled framework for autonomous microalgal cultivation and management. However, the proposed integrated architecture holds significant potential for microalgal cultivation systems, it has not yet been implemented and requires experimental validation.