Is AI memory actually a problem, or am I just using agents wrong?
A CS student using AI agents for coding, research and side projects asks how to keep context consistent across sessions and tools. Their manual workaround is to discuss a task with one AI, get a detailed prompt, combine it with a summary of relevant memories, and paste everything into the next session. The thread is a good prompt to discuss lightweight memory habits: one source-of-truth note per project, updated at the end of each session.