Chemical industry to bet on agentic AI

Agentic artificial intelligence, a form of AI capable of making autonomous decisions, is gaining traction in the laboratory and on the factory floor. The chemical industry will increasingly adopt it in 2026.“AI agents are autonomous units, something that can think and do stuff on its own. But, of course, it needs to be instructed,” says Marc Feldmann, senior principal at Alexander Thamm, an AI consulting firm that has advised several chemical companies. Feldmann says that, for now, he is seeing adoption mainly among big firms such as BASF and Bayer. In a blog post published on the Chemical Industry Journal website, Feldmann cites the example of the chemical giant Celanese, which created JO.AI, a platform of specialized AI agents, for its plant operations.“One agent optimises shift checklists. Another flags anomalies—‘Corrosion might be an issue in heat exchanger 3.’ A third drafts work orders,” he writes about the Celanese agent.Traditional AI applications are often preprogrammed and operate with limited autonomy. In contrast, agentic AI is more autonomous and can independently review the literature, formulate hypotheses, design experiments, learn from feedback, and collaborate with humans through conversations.The advent of agentic AI is driven by the growing popularity of large language models, which can understand and generate human language. “The successful integration of large language models (LLMs) into laboratory workflows has demonstrated robust capabilities in natural language processing, autonomous task execution, and collaborative problem-solving,” write Yi Luo and colleagues in a paper about their creation of what they call a multiagent-driven robotic AI