Qdrant developer relations engineer Dylan Couzon argues that locally held model weightsA local model runs on computing infrastructure controlled directly by the user rather than only through a remote model service. and persistent memory give users more control over their AI systems. He describes writing, retrieving and forgetting memoriesAI agent memory is stored information that an agent can retrieve and use across steps, sessions, or changing contexts. as distinct design concerns.
In a company demonstration, object detections from drone footage are embedded in a local vector store and retrieved by semantic searchRetrieval is the process of selecting relevant stored information and returning it to an AI system for the current task.. Couzon discusses optional synchronization and consent, and presents longer-term views about personal AI memory as predictions.
Watch on YouTube
