Developing new catalysts to tackle environmental issues and meet energy demands is a complex task. Traditional design and optimization methods often struggle with the vast and intricate catalyst parameter space. Machine Learning (ML) has opened new avenues in catalyst optimization, addressing some limitations of conventional methods. However, current approaches do not fully utilize the extensive information available in scientific literature on catalyst synthesis.

To overcome this challenge, this study introduces a groundbreaking Artificial Intelligence (AI) workflow that combines large-language models (LLMs), Bayesian optimization, and an active learning loop to accelerate and improve catalyst optimization. This innovative approach merges advanced language processing with robust optimization techniques, effectively translating knowledge from diverse literature into actionable parameters for practical experimentation and optimization.

The article demonstrates the application of this AI workflow in optimizing catalyst synthesis for ammonia production. The results highlight the workflow's ability to streamline the catalyst development process, providing a fast, resource-efficient, and precise alternative to traditional methods.