Predictive Process Monitoring aims to anticipate the future execution of business process instances, with suffix prediction focusing on forecasting the remaining sequence of events in an ongoing case. Most existing approaches assume that cases are executed in isolation, relying exclusively on intra-case information while neglecting inter-case dynamics. However, in real-world settings, cases often influence each other or share common system-wide constraints, such as resource availability. Current methods for incorporating inter-case information primarily focus on remaining time prediction by using manually engineered inter-case features. In this work, we propose I3SP (Integrated Inter-case Informed Suffix Prediction), a novel encoder-decoder architecture that integrates both trace and log prefixes for suffix prediction. Our approach directly learns inter-case dependencies from the log prefix, eliminating the need for selectively encoding inter-case features. Experimental results on real-life event logs demonstrate that I3SP effectively draws additional predictive power from the log prefix. Furthermore, an analysis of example predictions provides explanatory insights into the learned patterns that contribute to improved predictive performance.