Memora integrates vector and graph databases to efficiently manage and search memory data. The vector database, such as Qdrant (with more options coming soon), utilizes both dense and sparse embeddings (such as SPLADE) to perform hybrid search. Memories are first searched in the vector database, where the vector IDs correspond to the memory_ids of memory nodes in the graph database. This design allows for seamless retrieval and resolution of memories from Neo4j, ensuring that the most up-to-date version of a memory is utilized.


Token Account