📄 中文摘要
大多数大型语言模型（LLM）应用在演示中表现良好，但当用户第二天回来时，系统却忘记了所有信息。这并不是模型的问题，而是架构的问题。该指南提供了一种实用的、可生产的方式来为 LLM 应用添加持久内存，使用 Node.js、OpenAI API、Redis（用于结构化内存）以及向量存储进行语义检索。这一模式适用于构建 SaaS 工具、AI 助手或特定领域的 LLM 应用。
如何为 LLM 应用添加持久内存（无需微调）—— 实用架构指南
📄 English Summary
How to Add Persistent Memory to an LLM App (Without Fine-Tuning) — A Practical Architecture Guide
Most Large Language Model (LLM) applications perform well in demonstrations, but when a user returns the next day, the system forgets everything. This issue is not related to the model itself but rather to the architecture. The guide outlines a practical, production-ready approach to adding persistent memory to an LLM application without fine-tuning. It utilizes Node.js, the OpenAI API, Redis for structured memory, and a vector store for semantic retrieval. This pattern is applicable whether developing a SaaS tool, an AI assistant, or a domain-specific LLM application.