Build the mental model

Retrieval-augmented generation first finds external evidence and then asks a model to answer from that evidence. It does not teach the model new weights. Quality is a chain: ingestion, chunking, embedding/indexing, query processing, retrieval, reranking, context assembly, generation, citation, and evaluation.

A reliable beginner model has three boxes: input, transformation, and evidence. The input is what enters the system; the transformation is what the technique actually computes or changes; the evidence is how we learn whether the output works beyond one attractive example. For this topic, the transformation is not magic: Documents are parsed, normalized, split into addressable chunks, enriched with metadata, embedded, and indexed. At query time, the system may rewrite the question, apply access filters, run vector or hybrid search, rerank candidates, fit the best evidence into a context budget, and generate an answer whose claims point back to sources.

The boundary matters. Do not confuse a mechanism with an outcome. A mechanism can make a desired outcome more likely while still failing on a particular case. It also does not erase the need for source checks, permissions, or domain judgment. The practical question is therefore not “Does it work?” but “Under which inputs, constraints, and measurements does it work well enough?”

For “Can I carry over vacation?”, hybrid search retrieves policy chunks. A reranker promotes the region-specific 2026 policy over a popular 2024 FAQ. The answer quotes the two-day limit and cites the exact chunk; without supporting evidence it abstains.

The worked number is if 18 of 20 test questions retrieve at least one relevant chunk in top 5, recall@5 = 18/20 = 0.90. State the unit and denominator whenever you report it. A percentage without a denominator can conceal a tiny sample; a latency without a percentile can conceal slow users; a similarity score without a labeled task can conceal irrelevant neighbors. Compare the observed value with a threshold chosen before seeing the final test result.

Now test the tempting shortcut. Suppose the team optimizes only the most visible metric. The result may look better while the system becomes less trustworthy. The reason is concrete: Small chunks retrieve precisely but lose surrounding context; large chunks preserve context but dilute matches. Higher top-k improves recall while adding distractors and tokens. Reranking improves ordering at extra latency. Fresh indexing costs compute but stale evidence damages trust. This is why the decision record must include both the intended gain and the tolerated regression. If the tolerated regression is unknown, the change is not ready for a consequential workflow.

• Prefer a measured baseline over a persuasive demo.
• Keep versions, inputs, and thresholds reproducible.
• Separate syntactic success from semantic correctness and authorization.
• Escalate or abstain when evidence falls outside the contract.
• Re-evaluate when data, traffic, models, providers, or user goals change.

These rules turn the topic into an engineering decision rather than a slogan. They also make disagreement productive: another person can challenge the assumptions, rerun the evaluation, and reach a documented conclusion.

Create one near-example and one counterexample. The near-example should differ from the scenario in only one important way; the counterexample should look similar while requiring a different technique. For each, label the input, the transformation that actually occurs, and the evidence you would accept. This exercise prevents the topic name from becoming an all-purpose explanation. If you cannot say what would falsify your mental model, it is still a story rather than a model. End with one sentence beginning “This technique does not guarantee…” and make that limitation observable.