In this study, the Jensen inequality has been refined. To achieve this, we employ the fact that an abstract convex function with certain structural properties can be bounded from below in a neighborhood of its global minimum in terms of the norm of its gradient. Some examples have been provided to show the extent of refinement.