Agentic memory is persistent storage, not RAM; however, the computer industry has been mixing up these two terms forever (see
storage vs. memory). The two types of storage used with AI agents are the vector database and graph RAG (retrieval-augmented generation).
A vector database, which converts text into math, is good for questions and answers like regular database queries. To obtain answers, a straightforward search is used.
A graph RAG uses a knowledge graph to store entities and relationships. Good for reasoning, graph RAG results are made by traversing the graph. See
AI agent and
RAG.