Google Cloud 推出 Always-On Memory Agent，用 LLM 持续整合替代 RAG 与向量数据库

Most AI agents forget. They process a request, answer it, then drop the context. Google Cloud’s generative-ai repository now ships a sample that tackles this directly.