Architecture Reference

Appearance

Search, Retrieval, and RAG ​

Search, Retrieval, and RAG Graphics Coverage ​

Primary chapter graphic: RAG Request Grounding Flow, RAG and Agentic RAG Comparison, Open Source RAG Stack, Basic RAG Application on Cloud, RAG Application Stack. Accepted graphics: 5. Reviewed non-signal pages: 9. Open graphics in review: 0. QA status lives in graphics audit and visual review ledger.

Corpus pages: p. 25, p. 102-103, p. 130, p. 148, p. 164-165, p. 196-197, p. 208-209, p. 223, p. 250, p. 274-275, p. 291-292, p. 365, p. 388-389, p. 417-418 Coverage: 22 pages; low-confidence extraction ranges: p. 25, p. 365, p. 388-389

This chapter is part of Marius's owned architecture build corpus. The text routes decisions; durable implementation signal is carried by accepted graphics, reviewed non-signal decisions, and the linked QA audit.

Chapter Visuals ​

Accepted graphics carry the canonical design signal for this chapter. Each selected source page is either accepted as a graphic or explicitly marked non-signal in the source-faithful ledger. Review and QA state live in visual inventory, visual review ledger, and graphics audit.

RAG Request Grounding Flow ​

RAG Request Grounding Flow

RAG and Agentic RAG Comparison ​

RAG and Agentic RAG Comparison

Open Source RAG Stack ​

Open Source RAG Stack

Basic RAG Application on Cloud ​

Basic RAG Application on Cloud

RAG Application Stack ​

RAG Application Stack

Open Review Queue ​

  • none

Reviewed Non-Signal Pages ​

  • Search, Retrieval, And RAG: Database + Index Map: source p. 223; batch 04; status non-signal/reviewed; ledger reason in visual-review-ledger.json
  • Search, Retrieval, And RAG: LLM + RAG Map: source p. 209; batch 06; status non-signal/reviewed; ledger reason in visual-review-ledger.json
  • Search, Retrieval, And RAG: LLM + Embedding Map: source p. 164; batch 09; status non-signal/reviewed; ledger reason in visual-review-ledger.json
  • Search, Retrieval, And RAG: API Gateway + Database Map: source p. 389; batch 11; status non-signal/reviewed; ledger reason in visual-review-ledger.json
  • Search, Retrieval, And RAG: Database + Index Map: source p. 275; batch 12; status non-signal/reviewed; ledger reason in visual-review-ledger.json
  • Search, Retrieval, And RAG: Database + LLM Map: source p. 197; batch 16; status non-signal/reviewed; ledger reason in visual-review-ledger.json
  • Search, Retrieval, And RAG: Embedding + Pattern Map: source p. 165; batch 25; status non-signal/reviewed; ledger reason in visual-review-ledger.json
  • Search, Retrieval, And RAG: LLM + Embedding Map: source p. 365; batch 28; status non-signal/reviewed; ledger reason in visual-review-ledger.json
  • Search, Retrieval, And RAG: Session + Index Map: source p. 418; batch 31; status non-signal/reviewed; ledger reason in visual-review-ledger.json

Use When ​

  • Users or AI systems need source-grounded lookup over documents, records, or knowledge.

Avoid When ​

  • Permissions cannot be enforced before retrieval.

Core Model ​

  • Retrieval is a derived read model optimized for intent, relevance, freshness, and access control.
  • Prefer explicit ownership over accidental coupling. Every boundary should say who owns correctness, cost, data, recovery, and change.
  • Use corpus page pointers for inspection, and keep the chapter notes focused on reusable design decisions.

Implementation Guidance ​

  • Choose chunking, metadata, indexing, ranking, permission filters, and reindex flow before trusting answers.
  • Write the smallest useful design note: purpose, inputs, outputs, state, failure behavior, observability, and rollback.
  • Choose the first implementation that can be tested against the real workflow without hiding a known production risk.

Tradeoffs ​

  • Vector search broadens recall, while keyword and filters improve precision and control.
  • Centralization reduces duplicated work but can become a bottleneck when every team needs exceptions.
  • Specialized infrastructure helps at scale, but it must earn its operational cost.

Failure Modes ​

  • The answer cites stale or unauthorized source content.
  • The diagram shows boxes but not ownership, retry behavior, data freshness, or user-visible failure.
  • The system has no proof path for the highest-risk assumption.

Decision Checklist ​

  • Evaluate recall, precision, citation quality, freshness, and permission filtering with real examples.
  • Name the owner, source of truth, timeout, retry policy, and evidence that the path works.
  • Add one regression check for the failure mode most likely to recur.

Neutral Automation Examples ​

  • A policy assistant retrieves only documents visible to the current role, then returns citations with the answer.
  • A neutral internal automation starts with fixtures, then adds credentials, permissions, and production scheduling only after the boundary is tested.
  • A customer-facing workflow keeps irreversible actions behind explicit approval until metrics show it is safe to automate further.