Rogo 联合创始人 Gabe Stengel 指出,持久化智能体记忆是企业尚未解决的核心难题,关键在于"压缩":当用户与同一智能体进行 100 次对话后,模型需记住正确信息并把记忆压缩到每次可用的 token 量内,同时保持足够的连贯性和上下文,才能让人感觉是在与一个真正了解自己的"人"交流。
Persistent agent memory is a major unsolved problem in enterprises.
- Gabe Stengel, co-founder of Rogo(Wall Street AI co)
@patrick_oshag: "What's something that the models currently cannot do that, if they could, would really change the nature of the product?"
@GabeStengel: "Compaction. If you have 100 conversations with a single agent, how does it make sure that it's actually remembering the right things and compacting its memory into an amount of tokens that it can use every time and it has enough coherence and context on who you are and what you care about to make it feel like it's a true person that you're speaking to that learns more and more about you?"
----
(link to the full podcast in the quoted tweet below )
来源:@rohanpaul_ai · x.com