Current strategies for biochemical model generation and interrogation rely on standards-based XML model specifications, such as the Systems Biology Markup Language (SBML), to represent the underlying biology and some aspects of the simulations used to interrogate the model. However, these approaches often lock users into specific software applications, are not well designed from a usability perspective, are often lacking in standards support, and require expert domain specific knowledge typically missing in biologists or clinicians. We have taken a different approach; we’ll present a new framework being developed in our lab at Cornell University for generating course grained effective kinetic models of gene expression and metabolic processes in prokaryotes from natural language. We’ll present a prototype system, written in the Swift programming language, and use this system to generate, and subsequently analyze example models, including a model of carbon catabolite repression in Bacillus subtilis.