אל תשכירו את הזיכרון של ה-AI שלכם — דילן קוזון, Qdrant

תקציר מקורי באנגליתSuperintelligence is coming. The real question is who it belongs to. Dylan Couzon, developer relations engineer at Qdrant, argues that frontier-class models now run on hardware you can own, but the most valuable part of the stack, the memory that makes an assistant yours, still lives in someone else's cloud. He breaks memory down into write, retrieve and forget, explains why retrieval beats a big folder of markdown, and runs a live, fully offline demo: a drone that builds its own searchable memory with sub-millisecond queries in a 15 MB footprint, using Qdrant Edge. In this talk: • Why owning inference gives you autonomy, and owning memory gives you continuity • The CPU / RAM / disk model for LLMs, context and memory • Why retrieval beats dumping everything into the prompt • A live offline