--- title: Types of RAG #1 | daily.dev description: A structured overview of seven RAG (Retrieval-Augmented Generation) architecture types, covering Simple RAG, Simple RAG with Memory, Agentic RAG, Graph RAG,... canonical: https://daily.dev/posts/types-of-rag-1-5jyhs12df twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Types of RAG #1 | daily.dev og:description: A structured overview of seven RAG (Retrieval-Augmented Generation) architecture types, covering Simple RAG, Simple RAG with Memory, Agentic RAG, Graph RAG,... og:url: https://daily.dev/posts/types-of-rag-1-5jyhs12df og:image: https://api.daily.dev/og/posts/5jYhs12Df.png og:image:alt: Types of RAG #1 og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # Types of RAG #1 **[Medium](https://daily.dev/sources/medium_js)** · 19 min read · 2 upvotes · 0 comments ## Summary A structured overview of seven RAG (Retrieval-Augmented Generation) architecture types, covering Simple RAG, Simple RAG with Memory, Agentic RAG, Graph RAG, Self RAG, Branched RAG, and Multimodal RAG. Each type is explained with how it works, primary use cases, pros and cons, and a concrete worked example — mostly using financial analysis scenarios. The progression moves from basic single-step retrieval to increasingly sophisticated pipelines involving memory, autonomous agents, knowledge graphs, self-critique loops, parallel branching, and cross-modal retrieval. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### What is the difference between Simple RAG and Agentic RAG? Simple RAG performs one retrieval step then generates an answer, using embedding generation, a vector similarity search, and context injection into the LLM prompt, with no feedback loop. Agentic RAG instead uses an LLM as a reasoning engine to decompose complex queries into sub-tasks, dynamically select tools or data sources, and run iterative evaluation loops that refine searches until sufficient evidence is gathered before answering. _daily.dev surfaces practical breakdowns like this for engineers choosing a RAG architecture to build._ ### How does Self-RAG decide whether to retrieve information again? Self-RAG scores retrieved fragments on relevance and faithfulness using internal critique tokens or evaluation modules, checking whether the retrieved context matches the query and whether the generated answer is grounded in the source text without hallucination. If the evaluation flags the result as insufficient or the answer incomplete, an internal LLM rewrites the query and performs a refined secondary retrieval before generating again. _Engineers debugging retrieval quality issues track architecture patterns like self-critique loops on daily.dev._ ### What embedding models are used for Multimodal RAG to search across text, images, and audio together? Multimodal RAG relies on unified cross-modal embedding models such as CLIP or ImageBind, which map different media types—text chunks, image pixels, and audio frequencies—into a shared vector space so semantic concepts align regardless of original format. This lets a single similarity search retrieve a mix of relevant text, images, or video timestamps for a query that itself can be text, image, or a combination. _Teams evaluating multimodal search stacks follow architecture comparisons like this on daily.dev._ ## Similar posts on daily.dev - [Agentic RAG Explained in 3 Levels of Difficulty](https://daily.dev/posts/agentic-rag-explained-in-3-levels-of-difficulty-9v2stnekw) · Machine Learning Mastery · 0 upvotes · 0 comments - [RAG Architecture Explained: How Retrieval-Augmented Generation Actually Works](https://daily.dev/posts/rag-architecture-explained-how-retrieval-augmented-generation-actually-works-spmhltlnd) · BigData Boutique blog · 2 upvotes · 0 comments - [RAG, Agentic RAG, and AI Memory](https://daily.dev/posts/rag-agentic-rag-and-ai-memory-t92tejzkq) · Daily Dose of Data Science \| Avi Chawla \| Substack · 3 upvotes · 0 comments - [Why RAG Complexity Should Be Earned](https://daily.dev/posts/why-rag-complexity-should-be-earned-114pe1rxo) · Towards Data Science · 0 upvotes · 0 comments - [Agentic RAG: When Retrieval Needs to Think Before It Answers](https://daily.dev/posts/agentic-rag-when-retrieval-needs-to-think-before-it-answers-y10lymkah) · Medium · 2 upvotes · 0 comments --- Tags: [#llm](https://daily.dev/tags/llm), [#ai-agents](https://daily.dev/tags/ai-agents), [#rag](https://daily.dev/tags/rag), [#vector-search](https://daily.dev/tags/vector-search) [View this post on daily.dev](https://daily.dev/posts/types-of-rag-1-5jyhs12df) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"Types of RAG #1","url":"https://daily.dev/posts/types-of-rag-1-5jyhs12df","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/types-of-rag-1-5jyhs12df"},"datePublished":"2026-07-23T13:10:30.550Z","dateModified":"2026-09-14T08:32:53.984Z","description":"A structured overview of seven RAG (Retrieval-Augmented Generation) architecture types, covering Simple RAG, Simple RAG with Memory, Agentic RAG, Graph RAG,...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/14613dff525b431a3b29d4dba0076ec1?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/14613dff525b431a3b29d4dba0076ec1?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"Medium","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"Medium","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/medium","url":"https://daily.dev/sources/medium_js"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/types-of-rag-1-5jyhs12df","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":2},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"llm,ai-agents,rag,vector-search","timeRequired":"PT19M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Medium","item":"https://daily.dev/sources/medium_js"},{"@type":"ListItem","position":3,"name":"Types of RAG #1"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/types-of-rag-1-5jyhs12df#faq","mainEntity":[{"@type":"Question","name":"What is the difference between Simple RAG and Agentic RAG?","acceptedAnswer":{"@type":"Answer","text":"Simple RAG performs one retrieval step then generates an answer, using embedding generation, a vector similarity search, and context injection into the LLM prompt, with no feedback loop. Agentic RAG instead uses an LLM as a reasoning engine to decompose complex queries into sub-tasks, dynamically select tools or data sources, and run iterative evaluation loops that refine searches until sufficient evidence is gathered before answering. daily.dev surfaces practical breakdowns like this for engineers choosing a RAG architecture to build."}},{"@type":"Question","name":"How does Self-RAG decide whether to retrieve information again?","acceptedAnswer":{"@type":"Answer","text":"Self-RAG scores retrieved fragments on relevance and faithfulness using internal critique tokens or evaluation modules, checking whether the retrieved context matches the query and whether the generated answer is grounded in the source text without hallucination. If the evaluation flags the result as insufficient or the answer incomplete, an internal LLM rewrites the query and performs a refined secondary retrieval before generating again. Engineers debugging retrieval quality issues track architecture patterns like self-critique loops on daily.dev."}},{"@type":"Question","name":"What embedding models are used for Multimodal RAG to search across text, images, and audio together?","acceptedAnswer":{"@type":"Answer","text":"Multimodal RAG relies on unified cross-modal embedding models such as CLIP or ImageBind, which map different media types—text chunks, image pixels, and audio frequencies—into a shared vector space so semantic concepts align regardless of original format. This lets a single similarity search retrieve a mix of relevant text, images, or video timestamps for a query that itself can be text, image, or a combination. Teams evaluating multimodal search stacks follow architecture comparisons like this on daily.dev."}}]} ```