Oracle Agent Memory as an Enterprise Memory Substrate for Long-Horizon AI Agents

Short summary

This report presents Oracle Agent Memory, a database-native memory substrate for long-horizon AI agents built on Oracle Database. It addresses memory as a lifecycle—ingestion, extraction, consolidation, retrieval, summarization, and revision—with layered architecture separating active memory from passive storage. LongMemEval results show 93.8% accuracy using 10.7x fewer tokens than flat-history baselines.

• •Oracle Agent Memory provides database-native memory lifecycle management for long-horizon AI agents
• •Layered architecture separates active memory core from passive store with explicit scope control across users, agents, and threads
• •Achieves 93.8% LongMemEval accuracy using 10.7x fewer tokens than flat-history baselines

Generated with AI, which can make mistakes.