The memory singularity
Автор: AI Memory Forum
Загружено: 2025-11-25
Просмотров: 6
Cofounder Gabriel Kreiman describes what does it mean for an AI agent to have memory. Most frameworks treat it as a retrieval problem: vector databases, embeddings, and top-k search. But true memory requires more than finding relevant information: it requires persistent state, structured knowledge, and the ability to learn and evolve over time. This talk provides a framework for understanding memory as a first-class architectural component, explores the limitations of retrofitting memory onto stateless systems, and demonstrates design patterns for building agents with memory at their core. During this talk, I'll provide an overview of how memory means much more than just RAG, and demonstrate live what it looks like to build a stateful agent. Memory isn't just a feature you add, but the foundation on which intelligence is built.
Доступные форматы для скачивания:
Скачать видео mp4
-
Информация по загрузке:
