Claude 101

• QuestionWhat is "Project vs RAG corpus"?Tap to revealAnswerA Claude.ai Project pastes every attached file into every chat — it doesn't search them. RAG searches first, then pastes only the relevant bit.
• QuestionIn "Project vs RAG corpus", explain:Tap to revealAnswerProject = in-context file storage. Every file loads every chat.
• QuestionIn "Project vs RAG corpus", explain:Tap to revealAnswerNo embeddings index. No retriever. No vector store.
• QuestionIn "Project vs RAG corpus", explain:Tap to revealAnswerQuality degrades with file count because attention competes against more text.
• QuestionIn "Project vs RAG corpus", explain:Tap to revealAnswerRAG = retrieval before context. Corpus can be unbounded; context cost is constant.
• QuestionAnalogy for "Project vs RAG corpus":Tap to revealAnswerThink of a Project as a desk: every file you've stuck on it is in front of you when you sit down, all the time. RAG is a librarian: you ask a question, and only the relevant pages get handed to you. Stack 40 things on the desk and you can't find anything; ask the librarian for the same 40 and you get one tidy answer.
• QuestionWhat is "Product-tier vs API-level limit"?Tap to revealAnswerClaude.ai stacks subscription-tier limits on top of model capacity. The API exposes model capacity directly with no tier in between.
• QuestionIn "Product-tier vs API-level limit", explain:Tap to revealAnswerClaude.ai = product surface on top of the API. Tier-side caps stack on top of model capacity.
• QuestionIn "Product-tier vs API-level limit", explain:Tap to revealAnswerDaily message caps and dropdown model selection are pure tier features. The API has neither.
• QuestionIn "Product-tier vs API-level limit", explain:Tap to revealAnswerPer-message context on Claude.ai = min(model window, tier cap). The API exposes the model number directly.
• QuestionIn "Product-tier vs API-level limit", explain:Tap to revealAnswerDiagnostic question: does the limit change per subscription? If yes, it's product-tier.
• QuestionAnalogy for "Product-tier vs API-level limit":Tap to revealAnswerThink of the model as the engine and Claude.ai as the rental car. The engine can do 200 mph (model capacity), but the rental contract caps you at 80 (your tier). Renting a different contract changes the cap; switching to your own car (the API) lets the engine run at full.
• QuestionWhat is "Artifact vs inline message"?Tap to revealAnswerArtifacts park long deliverables in a side pane so they don't replay in the chat history every turn. Use them when the output is something the user will iterate on or carry away.
• QuestionIn "Artifact vs inline message", explain:Tap to revealAnswerArtifact = isolated, iterable, re-renderable output. Inline = part of the conversation.
• QuestionIn "Artifact vs inline message", explain:Tap to revealAnswerInlining long output bloats the chat history that replays every turn.
• QuestionIn "Artifact vs inline message", explain:Tap to revealAnswerArtifacts are for deliverables; inline is for conversation.
• QuestionIn "Artifact vs inline message", explain:Tap to revealAnswerToken cost per Artifact is identical to inline — the win is structural, not economic per call.
• QuestionAnalogy for "Artifact vs inline message":Tap to revealAnswerThink of inline replies as talking and Artifacts as handing someone a printout. You wouldn't read a 10-page memo aloud during conversation just to keep talking — you'd hand them the page and say 'see attached.' Artifacts are 'see attached.'
• QuestionWhat is "Context cost of N file uploads"?Tap to revealAnswerN attached files cost roughly N × file_size of context every chat — there is no caching, no retrieval, no separate budget.
• QuestionIn "Context cost of N file uploads", explain:Tap to revealAnswerPer-message cost ≈ N × file_size, paid every turn. No caching at this layer.
• QuestionIn "Context cost of N file uploads", explain:Tap to revealAnswerConversation budget = window − attached files. Heavy Projects shrink the budget visibly.
• QuestionIn "Context cost of N file uploads", explain:Tap to revealAnswerQuality degrades from attention competition before the hard window limit fires.
• QuestionIn "Context cost of N file uploads", explain:Tap to revealAnswerThreshold question: must every chat see all of it? If no, switch to retrieval — don't add files.
• QuestionAnalogy for "Context cost of N file uploads":Tap to revealAnswerImagine printing every attached file at the start of every conversation and stapling it to the front of your question. The pile grows; your speaking room shrinks; eventually the staple bursts. Adding more paper doesn't help — moving the rarely-needed pages off the desk does.

1. 1Diagram the core idea

   Analyse~10 min

   In your own words, diagram the central idea of "Project vs RAG corpus". Aim for one paragraph, no jargon you haven't earned.

   Done when: A non-expert could read your paragraph and explain it back to you.

   Re-read the source concept

2. 2Solve it to a real situation

   Apply~15 min

   Find a real situation from your own work / life where "Context cost of N file uploads" applies. Solve the idea — describe the situation, the move you'd make, and the alternative you considered first.

   Done when: You can name a specific situation (not a hypothetical), state your move, and explain the trade-off in one paragraph.

   Re-read the source concept

3. 3Teach it back

   Evaluate~5 min

   Pick someone (a colleague, a peer, a rubber duck) and explain the section to them in under 5 minutes. Use your own example, not the one from the lesson.

   Done when: Your listener can repeat the core idea + the trade-off back to you without prompting.

Why this matters. Reading + quizzing prove recall; applying it in a real situation is the only evidence of transfer. Treat at least one of these as non-negotiable before you move on.