<p>In this study, a customizable generative artificial intelligence (GenAI) system is constructed to facilitate users to formulate and solve job scheduling problems in a flexible manufacturing system (FMS) consisting of multiple factories. Such problems are usually powered by bio-inspired algorithms, which are difficult to comprehend, trust, and/or apply. In the GenAI system, users first express their scheduling requirements in natural language through the system interface. Since the functional scope of job scheduling is narrow and factory workers have similar knowledge backgrounds and may be asked to enter more professional terms, a pre-trained natural language parser is used instead of a large language model (LLM) to extract the relative position and combination of keywords to establish an extended three-field notation (ETFN) for the job scheduling problem in FMSs. Based on the ETFN, a genetic algorithm (GA) is generated to solve the customized job scheduling problem. The optimal solution can be relayed to the project management system to facilitate subsequent operations. The customizable GenAI system has been applied to an FMS consisting of three factories and eighteen machines. According to the experimental results, the natural language parser successfully identified user’s scheduling requirements with an accuracy of up to 92% for unlearned data. The customizable GenAI system improved ease of application, required minimal execution time, and achieved the highest fitness compared to several current practices.</p>