Complex networks mathematically model the social, biological, and ecological networks. These systems exhibit intricate patterns of connectivity and are dynamic. Dynamic networks are characterized by temporal evolution, where nodes and edges change over time. Such temporal networks mimic real-world phenomena such as social interactions, communication patterns, and gene mutation. Such dynamic interactions result in structural evolution that can be continuous or event-driven. This structural evolution is closely tied to the formation and completion of triads that are the fundamental substructure of network topology. The triads consist of three interconnected nodes. Analysis of triad patterns plays a vital role in link prediction and influence propagation in dynamic networks. This demands identifying the unique triads in the network that would contribute to forming future links in the network. This article aims to predict and enumerate the triad patterns in dynamic networks. Since the underlying network is dynamic, the triadic pattern formation evolves slices. The article proposes a Hybrid GAT-SAGE model, that combines the GAT (Graph Attention Network) and GraphSage (Graph Sample and Aggregation) models to predict the triads. The proposed hybrid model is enabled with a custom attention mechanism, which aims to find the patterns associated with triad formations. The proposed model is empirically evaluated using the SNAP College MSG dataset. The results indicate the model’s effectiveness in predicting and enumerating the triads in future time slices in terms of accuracy.