Episode 3

Giving persistent memory to a frozen AI (RAG explained simply)

An AI model is frozen at training time: it learns nothing from your conversations. So how do you give it real memory?

26:45

Show notes

An AI model is frozen at training time: it learns nothing from your conversations. So how do you give it real memory?

In this episode

  • Episodic vs semantic memory: an assistant's two brains
  • RAG and vector stores: retrieving the right info at the right time
  • Privacy by design: an external memory that stays under your control

References

  • Retrieval-Augmented Generation (RAG), embeddings, vector databases
  • Memory architecture of AI agents

24h action

Note 3 facts you keep re-explaining to your AI assistant. That's your first "external memory".

#AI#LLM#external memory#RAG#vector store#embeddings#privacy