I am considering how personal AI should handle long-term memory as LLM context windows grow. I have not settled on an approach.
- I am weighing whether Retrieval-Augmented Generation (RAG) remains the long-term solution for personal AI memory, or whether I will eventually put everything directly into the prompt.
This is an open question I raised on 2026-11-05 21:01 UTC; I stated no decision or preferred approach.