A plain-language guide to Retrieval-Augmented Generation (RAG) — its key components and how it enhances large language models with external knowledge.https://retrieval-augmentedgeneration.com/https://retrieval-augmentedgeneration.com/blogs/corpus-you-dont-own/https://retrieval-augmentedgeneration.com/blogs/corpus-you-dont-own/Swap a vector index for a live search retriever and you lose curation. Here's what curation was doing, and how to get the important parts back.Sun, 16 Aug 2026 00:00:00 GMTretrievalweb-searchtrusthttps://retrieval-augmentedgeneration.com/blogs/answer-boxes-as-a-source/https://retrieval-augmentedgeneration.com/blogs/answer-boxes-as-a-source/A featured snippet is already an answer. Treating it as a retrieved source means grounding your generation on an extraction you didn't perform.Sat, 15 Aug 2026 00:00:00 GMTretrievalweb-searchgroundinghttps://retrieval-augmentedgeneration.com/blogs/search-api-as-the-retriever/https://retrieval-augmentedgeneration.com/blogs/search-api-as-the-retriever/A RAG system doesn't have to retrieve from a vector index. Here's what the R stage looks like when it's an HTTP query to a search engine.Fri, 14 Aug 2026 00:00:00 GMTretrievalarchitectureweb-searchhttps://retrieval-augmentedgeneration.com/blogs/answering-in-the-readers-language/https://retrieval-augmentedgeneration.com/blogs/answering-in-the-readers-language/The question is in French and the corpus is in English. What that does to quotes, citations, defined terms and your faithfulness checks.Fri, 31 Jul 2026 00:00:00 GMTgenerationanswer-designattributionhttps://retrieval-augmentedgeneration.com/blogs/questions-with-false-premises/https://retrieval-augmentedgeneration.com/blogs/questions-with-false-premises/Asked why a policy allows fifteen days when it allows ten, a grounded model will explain the fifteen. How to make it check the premise first.Fri, 31 Jul 2026 00:00:00 GMTgenerationgroundinganswer-designhttps://retrieval-augmentedgeneration.com/blogs/the-model-does-not-know-what-day-it-is/https://retrieval-augmentedgeneration.com/blogs/the-model-does-not-know-what-day-it-is/Documents talk about now and next quarter. The model has no clock. How to make time resolvable in a grounded answer instead of guessed.Thu, 30 Jul 2026 00:00:00 GMTgenerationgroundingcontexthttps://retrieval-augmentedgeneration.com/blogs/when-the-answer-is-a-procedure/https://retrieval-augmentedgeneration.com/blogs/when-the-answer-is-a-procedure/A how-to answer someone will execute fails differently from a factual one. Dropped steps, merged variants, and lost conditionals.Thu, 30 Jul 2026 00:00:00 GMTgenerationanswer-designextractionhttps://retrieval-augmentedgeneration.com/blogs/arithmetic-the-sources-didnt-do/https://retrieval-augmentedgeneration.com/blogs/arithmetic-the-sources-didnt-do/A total the model computed appears in no document. Derived figures are new claims, they can't be cited, and they are wrong more often than they look.Wed, 29 Jul 2026 00:00:00 GMTgenerationgroundingreliabilityhttps://retrieval-augmentedgeneration.com/blogs/what-to-log-about-a-rag-answer/https://retrieval-augmentedgeneration.com/blogs/what-to-log-about-a-rag-answer/When an answer is wrong, you need to reconstruct how it was produced. The trace that makes that possible, and the privacy cost of keeping it.Tue, 28 Jul 2026 00:00:00 GMTgenerationobservabilitydebugginghttps://retrieval-augmentedgeneration.com/blogs/structured-output-from-retrieved-context/https://retrieval-augmentedgeneration.com/blogs/structured-output-from-retrieved-context/Extracting fields rather than prose changes the problem: schemas can be validated, and every field needs a way to say it was not found.Mon, 27 Jul 2026 00:00:00 GMTgenerationstructured-outputextractionhttps://retrieval-augmentedgeneration.com/blogs/the-shape-of-a-good-answer/https://retrieval-augmentedgeneration.com/blogs/the-shape-of-a-good-answer/Length, structure and ordering are design decisions with real effects — starting with putting the answer in the first sentence.Sun, 26 Jul 2026 00:00:00 GMTgenerationanswer-designuxhttps://retrieval-augmentedgeneration.com/blogs/decoding-settings-for-grounded-answers/https://retrieval-augmentedgeneration.com/blogs/decoding-settings-for-grounded-answers/Temperature and its neighbours do something different in RAG than in creative writing. Sensible values, and what they can't fix.Sat, 25 Jul 2026 00:00:00 GMTgenerationtuningreliabilityhttps://retrieval-augmentedgeneration.com/blogs/streaming-a-grounded-answer/https://retrieval-augmentedgeneration.com/blogs/streaming-a-grounded-answer/Streaming makes a system feel fast and makes validation impossible, because you have published the answer before you could check it.Fri, 24 Jul 2026 00:00:00 GMTgenerationstreamingvalidationhttps://retrieval-augmentedgeneration.com/blogs/summarising-across-many-documents/https://retrieval-augmentedgeneration.com/blogs/summarising-across-many-documents/Some questions need fifty documents, not five. The staged approach that handles them, and where accuracy leaks out of it.Thu, 23 Jul 2026 00:00:00 GMTgenerationsummarisationarchitecturehttps://retrieval-augmentedgeneration.com/blogs/more-than-one-retrieval-pass/https://retrieval-augmentedgeneration.com/blogs/more-than-one-retrieval-pass/Some questions can't be answered from a single search. How iterative retrieval works, and the loop control it needs to stay sane.Wed, 22 Jul 2026 00:00:00 GMTgenerationiterativearchitecturehttps://retrieval-augmentedgeneration.com/blogs/multi-turn-rag/https://retrieval-augmentedgeneration.com/blogs/multi-turn-rag/Follow-up questions break retrieval, because 'what about contractors?' contains almost nothing to search for. How to handle conversation.Tue, 21 Jul 2026 00:00:00 GMTgenerationconversationretrievalhttps://retrieval-augmentedgeneration.com/blogs/permissions-at-generation-time/https://retrieval-augmentedgeneration.com/blogs/permissions-at-generation-time/Retrieval can surface a document the asker is not allowed to read, and the answer will summarise it. Where access control has to live.Mon, 20 Jul 2026 00:00:00 GMTsecuritypermissionsgenerationhttps://retrieval-augmentedgeneration.com/blogs/prompt-injection-in-retrieved-documents/https://retrieval-augmentedgeneration.com/blogs/prompt-injection-in-retrieved-documents/Retrieved content goes into your prompt as instructions the model might follow. Why RAG makes this worse, and what actually reduces the risk.Sun, 19 Jul 2026 00:00:00 GMTsecuritygenerationinjectionhttps://retrieval-augmentedgeneration.com/blogs/quoting-versus-paraphrasing/https://retrieval-augmentedgeneration.com/blogs/quoting-versus-paraphrasing/Verbatim quotes are checkable and stiff; paraphrase reads well and can drift. When each is right, and how to combine them.Sat, 18 Jul 2026 00:00:00 GMTgenerationcitationsanswer-designhttps://retrieval-augmentedgeneration.com/blogs/when-sources-disagree/https://retrieval-augmentedgeneration.com/blogs/when-sources-disagree/Two retrieved documents give different answers. What a good system does about it, and why silence is the worst option.Fri, 17 Jul 2026 00:00:00 GMTgenerationconflictreliabilityhttps://retrieval-augmentedgeneration.com/blogs/checking-citations-after-generation/https://retrieval-augmentedgeneration.com/blogs/checking-citations-after-generation/A validation pass that confirms every claim is supported before the answer ships. What it can catch cheaply, and what it can't.Thu, 16 Jul 2026 00:00:00 GMTgenerationvalidationattributionhttps://retrieval-augmentedgeneration.com/blogs/citations-that-point-somewhere-real/https://retrieval-augmentedgeneration.com/blogs/citations-that-point-somewhere-real/Asking a model to cite its sources is easy. Getting citations that actually support the claims attached to them is a separate problem.Wed, 15 Jul 2026 00:00:00 GMTgenerationcitationsattributionhttps://retrieval-augmentedgeneration.com/blogs/teaching-a-model-to-say-i-dont-know/https://retrieval-augmentedgeneration.com/blogs/teaching-a-model-to-say-i-dont-know/Abstention is a feature you build, not a default you get. Where the decision belongs and how to keep it from swallowing good questions.Tue, 14 Jul 2026 00:00:00 GMTgenerationabstentionreliabilityhttps://retrieval-augmentedgeneration.com/blogs/grounding-instructions-that-work/https://retrieval-augmentedgeneration.com/blogs/grounding-instructions-that-work/The few sentences telling the model what to do with retrieved material have an outsized effect. What belongs in them, in what order.Mon, 13 Jul 2026 00:00:00 GMTgenerationpromptinggroundinghttps://retrieval-augmentedgeneration.com/blogs/why-the-model-ignores-your-context/https://retrieval-augmentedgeneration.com/blogs/why-the-model-ignores-your-context/You retrieved the right documents and the answer came from the model's training instead. The causes, and what actually fixes each one.Sun, 12 Jul 2026 00:00:00 GMTgenerationgroundingtroubleshootinghttps://retrieval-augmentedgeneration.com/blogs/how-many-chunks-to-include/https://retrieval-augmentedgeneration.com/blogs/how-many-chunks-to-include/More retrieved context is not reliably better. Why adding chunks stops helping, and how to find where that happens for your system.Sat, 11 Jul 2026 00:00:00 GMTgenerationcontexttuninghttps://retrieval-augmentedgeneration.com/blogs/assembling-the-prompt/https://retrieval-augmentedgeneration.com/blogs/assembling-the-prompt/Retrieved chunks, the question, the instructions and the history all go in one prompt. How you arrange them changes the answer.Fri, 10 Jul 2026 00:00:00 GMTgenerationpromptingpipelinehttps://retrieval-augmentedgeneration.com/blogs/what-happens-after-retrieval/https://retrieval-augmentedgeneration.com/blogs/what-happens-after-retrieval/Most RAG writing stops once the right documents are found. That is the halfway point — here is the rest of the pipeline, stage by stage.Thu, 09 Jul 2026 00:00:00 GMTgenerationpipelinebasics