Agent memory layers don't need an LLM deciding what to remember

Most agent memory setups run an extra LLM inference on every interaction to judge, rewrite, and tag what to keep. The author argues this second inference pass is wasted, suggesting memory selection should be handled without a model call.