The Real Reason AI Keeps Forgetting What You Told It and How to Fix It (AI Memory Explained)

Andrew Miles Davis tackles a problem that has been genuinely frustrating him, AI's tendency to forget context that matters even after hours of collaborative work, and breaks it down into three distinct levels of memory. Global memory covers broad, stable facts about a person that reasonably follow them across every conversation, but cannot hold detailed project context without becoming unwieldy and contextually inappropriate. Project memory creates a boundary around a specific piece of work, letting AI hold format, audience, and history for something like a podcast or campaign, but still leaves the AI deciding what within that boundary actually deserves to be remembered. The third and most reliable level involves building an external memory system, real files and notes that the user controls and updates after each session, which AI reads as a record rather than relies on recall for. Andrew shares his own practical example of maintaining a simple status file for major projects that gets updated after every working session, eliminating the need to hope AI remembered the right details. Subscribe to In AI Nutshell for daily ten-minute episodes that dig into how to actually get better results from the AI tools you already use.