Relax. We're not going to explain this chapter like a textbook — just like a friend talking you through it, in plain, easy-to-follow terms.
Let's think about it this way for a second
Prompt engineering polishes the wording. Context engineering arranges the table. Memory saves notes from the table for a future day. RAG pulls a passage out of a document and drops it into the context. An agent feeds a tool's output back into the context. Vibe coding treats a spec as context. Don't lump these together. On the very first turn you don't need a memory system yet — just pick the 3 sheets of notes you're putting on the table this turn.
Let's connect it to everyday life
"Explain it in 5 sentences" is a prompt. "Read this short note and explain it" is context. "This student already asked about sunlight" is memory. Start out with a short prompt plus a short context. Don't call the whole chat history "memory" — that's just leftover dishes still sitting on the table. It only becomes memory once you deliberately save it to reuse tomorrow.
Let's try it out together
Prompt → say the job
Context → show the notes for this turn
Memory → keep one fact for later
RAG → find a page, then put it in contextTell prompt, context, and memory apart.5-Minute Try
Write 6 tasks. For each one, note whether it's a prompt, context, memory, or RAG — and why.
One Quick Heads-Up
Telling it "remember this" doesn't actually create memory. If you don't add it back in next turn, the model won't see it anymore.