请问HN：在群聊中使用RAG、GraphRAG还是其他什么工具来实现人工智能？
嗨，HN，
我正在开发一个聊天应用，用户可以创建包含AI的群聊。这个想法是群组成员可以向AI提问或分享信息，而AI应该能够记住是谁说了什么，以便在后续提供更好的上下文和信息检索。
我正在探索基于RAG的方法，并关注GraphRAG（例如CircleMind或基于Neo4j的解决方案），但这些似乎比较复杂，并且在扩展时可能会很昂贵。
你认为GraphRAG适合这种群体上下文记忆吗？还是说更简单的架构会更合适？任何建议或替代方案都将非常感谢！
非常感谢！
查看原文
Hi HN,
I’m building a chat app where users can create group chats that include an AI. The idea is that group members can ask questions or share information with the AI, and the AI should be able to remember who said what, for better context and retrieval later on.<p>I’m exploring RAG-based approaches and looking into GraphRAG (like CircleMind or Neo4j-based solutions), but these seem complex and potentially expensive to scale.<p>Do you think GraphRAG is the right fit for this kind of group-context memory? Or would a simpler architecture be more appropriate? Any advice or alternatives would be much appreciated!<p>Thanks a lot!