💻
💻 Technology
Claude memorises session transcripts instead of learning from them — a growing agentic problem
Developers using Anthropic's Claude in agentic mode have flagged an unwanted behaviour: instead of learning from past sessions, the model attempts to memorise entire session transcripts, wasting context window space and degrading performance. The issue is described in a blog post on 12gramsofcarbon.com, which calls on Anthropic to address it. The post sparked debate on Hacker News about how AI agents should handle memory.
Comments
No comments yet
Comments
No comments yet — be the first to weigh in 👇
No comments yet. Be the first!