Understanding the structural dynamics associated with enzymatic catalysis has been a long-standing goal of biochemistry. With the advent of modern cryo-electron microscopy (cryo-EM), it has become conceivable to redefine a protein's structure as the continuum of all conformations and their distributions. However, capturing and interpreting this information remains challenging. Here, we use classification and deep-learning-based analyses to characterize the conformational heterogeneity of a class I ribonucleotide reductase (RNR) during turnover. By converting the resulting information into physically interpretable 2D conformational landscapes, we demonstrate that RNR continuously samples a wide range of motions while maintaining surprising asymmetry to regulate the two halves of its turnover cycle. Remarkably, we directly observe the appearance of highly transient conformations needed for catalysis, as well as the interaction of RNR with its endogenous reductant thioredoxin also contributing to the asymmetry and dynamics of the enzyme complex. Overall, this work highlights the role of conformational dynamics in regulating key steps in enzyme mechanisms.