Memory Management and Contextual Consistency for Long-Running Low-Code Agents

The rise of AI-native Low-Code/No-Code (LCNC) platforms enables autonomous agents capable of executing complex, long-duration business processes. However, a fundamental challenge remains: memory management. As agents operate over extended periods, they face memory inflation and contextual degradation, leading to inconsistent behavior, error accumulation, and increased computational cost. This paper proposes a hybrid memory system for LCNC agents that combines episodic and semantic memory with an Intelligent Decay mechanism, which prunes or consolidates memories using a composite score of recency, relevance, and user-specified utility. A user-centric visualization interface enables non-technical users to manage the agent’s memory by tagging which facts should be retained or forgotten. Simulated long-running task experiments show the system outperforms sliding windows and basic RAG in task completion rates, contextual consistency, and long-term token cost efficiency, offering a framework for reliable, transparent agents capable of long-term learning and adaptation.