Currently, Raglight retrieves documents based on the input query (RAG) but does not maintain any session-based memory. Adding memory would allow the agent to remember previous interactions, summarize them, and use them in subsequent queries for better context-aware responses.