YantrikDB is a cognitive memory engine designed to go beyond traditional vector databases, which merely store and retrieve. It actively manages AI agent memories by consolidating similar information, detecting factual contradictions, and allowing memories to fade over time based on importance and half-life. This approach ensures higher recall quality and prevents the accumulation of noisy, irrelevant, or conflicting data, enabling AI agents to maintain a more organized, contextual, and efficient knowledge base for improved decision-making and interaction.
Key Features
01Autonomous Memory Forgetting (Temporal Decay)
02Semantic Consolidation of Memories
03Factual Contradiction Detection and Resolution
04Multi-signal Relevance Scoring for Recall
05CRDT-based Multi-Device Memory Synchronization
06106 GitHub stars
