Enable any language model with permanent, searchable memory using a lightweight middleware for on-demand retrieval and continuous learning.