Based on previous experience in prediction systems with Fuzzy Deformable Prototypes, improvement in their functioning and deformation capacity is necessary, a new idea/concept for this is proposed. We propose a system with artificial intelligence that is capable of characterizing new situations, within the context of a PhD. thesis, capable of recognizing samples in a cognitive environment, in addition to testing its viability and performance in a non-cognitive environment a preliminary experiment is carried out, classifying handwritten digits from a reference database (MNIST) with a good success rate.