Many biological decision-making processes can be viewed as performing a classification task over a set of inputs, using various chemical and physical processes as "biological hardware." In this context, it is important to understand the inherent limitations on the computational expressivity of classification functions instantiated in biophysical media. In work done by Carlos Floyd, in collaboration with Aaron Dinner, Arvind Murugan, and Suri Vaikuntanathan, biochemical networks -modeled as Markov jump processes - are trained to perform several classification tasks. Their work reveals several unanticipated limitations on the input-output functions of these systems. They show that these limitations are related to a new thermodynamic constraint governing the response properties of the non-equilibrium steady state. They further show that these limitations can be lifted using biochemical mechanisms like promiscuous binding. Finally their work also provides a framework to understand the classification capacity of these networks. Their findings have implications for understanding how biological systems, constrained by chemical and thermodynamic rules, can read and generate complex codes robustly.