External Memory: Vector Stores & Databases

~15 min read

External memory persists information outside the context window — in vector stores or databases — so an agent can recall relevant facts from far earlier, or from entirely separate past sessions, without needing everything to fit in one prompt.

External Memory: Vector Stores & Databases is a Pro topic

Sign in, then upgrade to Pro or Power to unlock this topic and the full AI Engineering curriculum.

Key points

  • External memory persists outside the context window — in vector stores or databases — surviving past a single session
  • Vector stores suit memory retrieved by MEANING (semantic similarity), reusing the same mechanism that powers RAG
  • Traditional databases suit memory retrieved by EXACT KEY, like structured user preferences or account details
  • Without external memory, every new session starts as a blank slate no matter how rich the in-context memory was during a prior session
  • External memory is what turns a stateless, single-session assistant into one that accumulates knowledge about a user or task over time