{"id":5106,"date":"2026-08-18T11:45:10","date_gmt":"2026-08-18T02:45:10","guid":{"rendered":"https:\/\/cinnamon.ai\/?post_type=ideas&#038;p=5106"},"modified":"2026-08-18T11:45:11","modified_gmt":"2026-08-18T02:45:11","slug":"super-rag-tech-blog-07","status":"publish","type":"ideas","link":"https:\/\/cinnamon.ai\/en\/ideas\/super-rag-tech-blog-07\/","title":{"rendered":"[Episode 7] Super RAG API Use Case: &quot;Document vs. Document&quot; Search \u2014 Searching past accidents and near misses from work plans to predict hazards."},"content":{"rendered":"<ul><li><a class=\"aioseo-toc-item\" href=\"#aioseo-\u306f\u3058\u3081\u306b-1\">Introduction<\/a><\/li><li><a class=\"aioseo-toc-item\" href=\"#aioseo-\u672c\u9023\u8f09\u306b\u3064\u3044\u3066-5\">About this series<\/a><\/li><li><a class=\"aioseo-toc-item\" href=\"#aioseo-\u696d\u52d9\u30b7\u30fc\u30f3-\u88fd\u9020\u696d\u306e\u5b89\u5168\u7ba1\u7406\u3068-ky\u6d3b\u52d5-8\">Workplace scenarios \u2014 Safety management and hazard prediction activities in manufacturing<\/a><\/li><li><a class=\"aioseo-toc-item\" href=\"#aioseo-\u306a\u305c\u901a\u5e38\u306erag\u3067\u306f\u8db3\u308a\u306a\u3044\u304b-\u8cea\u554f\u306b\u7b54\u3048\u308b\u3068\u6587\u66f8\u304b\u3089\u6587\u66f8\u3092\u63a2\u3059\u306e\u9055\u3044-23\">Why a standard RAG isn&#039;t enough \u2014 the difference between &quot;answering questions&quot; and &quot;searching for documents within documents.&quot;<\/a><\/li><li><a class=\"aioseo-toc-item\" href=\"#aioseo-\u691c\u7d22\u30a2\u30fc\u30ad\u30c6\u30af\u30c1\u30e3-\u6587\u66f8-vs-\u6587\u66f8\u306e\u30ab\u30b9\u30bf\u30e0\u30d1\u30a4\u30d7\u30e9\u30a4\u30f3-43\">Search Architecture \u2014 Custom &quot;Document vs. Document&quot; Pipelines<\/a><\/li><li><a class=\"aioseo-toc-item\" href=\"#aioseo-\u306a\u305c\u6587\u66f8-vs-\u6587\u66f8\u691c\u7d22\u306f-super-rag\u306e\u524d\u51e6\u7406\u3092\u5fc5\u8981\u3068\u3059\u308b\u304b-70\">Why does &quot;document vs. document&quot; search require Super RAG preprocessing?<\/a><\/li><li><a class=\"aioseo-toc-item\" href=\"#aioseo-\u3053\u306e\u30a2\u30fc\u30ad\u30c6\u30af\u30c1\u30e3\u306f\u4f55\u3092\u610f\u5473\u3059\u308b\u304b-\u7b2c2\u56de\u7b2c6\u56de\u306e\u8a2d\u8a08\u5224\u65ad\u3068\u306e\u6574\u5408-107\">What does this architecture mean? \u2014 Consistency with the design decisions in Parts 2 and 6.<\/a><\/li><li><a class=\"aioseo-toc-item\" href=\"#aioseo-\u6a2a\u5c55\u958b\u8ef8a-\u88fd\u9020\u696d\u5185\u3067\u306e\u4ed6\u306e\u6587\u66f8-vs-\u6587\u66f8\u30b7\u30fc\u30f3-116\">Horizontal Expansion\/Axis A \u2014 Other &quot;Document vs. Document&quot; Scenarios within Manufacturing<\/a><\/li><li><a class=\"aioseo-toc-item\" href=\"#aioseo-\u6a2a\u5c55\u958b\u8ef8b-\u696d\u7a2e\u3092\u8d85\u3048\u308b\u30c8\u30e9\u30d6\u30eb\u691c\u7d22\u30d1\u30bf\u30fc\u30f3-131\">Horizontal Expansion\/Axis B \u2014 Trouble Search Patterns Across Industries<\/a><\/li><li><a class=\"aioseo-toc-item\" href=\"#aioseo-\u30b9\u30c8\u30fc\u30ea\u30fc\u306e\u7d9a\u304d-super-rag\u3092\u5165\u308c\u305f\u5f8c\u306e-m\u3055\u3093\u306e\u696d\u52d9-143\">Continuing the story \u2014 M&#039;s work after installing Super RAG<\/a><\/li><li><a class=\"aioseo-toc-item\" href=\"#aioseo-\u6301\u3061\u5e30\u308a\u691c\u8a0e\u6750\u6599\u81ea\u696d\u52d9\u306e\u6587\u66f8-vs-\u6587\u66f8\u30b7\u30fc\u30f3\u68da\u5378\u3057\u30d5\u30ed\u30fc-149\">[Things to take home and consider] Flowchart for inventorying &quot;document vs. document&quot; scenarios in your own work<\/a><\/li><li><a class=\"aioseo-toc-item\" href=\"#aioseo-\u6b21\u56de\u4e88\u544a-175\">Next episode preview<\/a><\/li><li><a class=\"aioseo-toc-item\" href=\"#aioseo-\u307e\u3068\u3081-178\">summary<\/a><\/li><\/ul>\n\n\n<div style=\"height:85px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<h2 id=\"aioseo-\u306f\u3058\u3081\u306b-1\" class=\"wp-block-heading\"><a>Introduction<\/a><\/h2>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>The message of this article<\/strong>: The normal RAG is &quot;<strong>Enter your question \u2192 Search and refer to related documents to answer each question.<\/strong>It operates in the pattern of &quot;. On the other hand, the work on site is &quot;<strong>I want to use the document itself as input and retrieve a related set of other documents.<\/strong>&quot;\u2014For example, there is a need to input tomorrow&#039;s work plan and retrieve past accident and near-miss incidents that occurred in similar tasks. This is what makes it possible.&quot;<strong>Document vs. Document<\/strong>In the custom search pipeline of &quot;, the foundation is <strong>Super RAG preprocessing engine and search infrastructure (structure preservation by DocReader, hybrid search, dictionary preprocessing)<\/strong> This can be achieved by combining these elements using APIs. In this article, we will delve deeper into this pattern, using safety management in the manufacturing industry as an example.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">For example, at a manufacturing company<strong>safety manager<\/strong>Please think of Ms. M, who serves in this role. Based on the Industrial Safety and Health Act, workplaces in designated manufacturing industries with 50 or more employees are required to appoint a safety manager, and Ms. M is in that role. One of her monthly duties is:<strong>Pre-operation risk assessment for special tasks in the following month<\/strong>The process involved reviewing accident reports and near-miss incidents from the past five years, identifying similar work conditions, and sharing them at KY (Kiken Yochi - Hazard Prediction) meetings\u2014a time-consuming process each time. It relied on the memories of veterans, partial searches by file name, and patrolling folders on the company&#039;s file server and full-text search systems. &quot;I think that incident was similar, but where did I save it?&quot; and &quot;I thought this incident was similar, but actually the procedure is different.&quot; These were common occurrences.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Up to the 6th round, Super RAG <strong>Extraction (DocReader) \/ Hybrid Search \/ Answer Strategy<\/strong> I&#039;ve been talking about how we build up accuracy using these three pillars. In this article, I&#039;ll explain how these three pillars are effective for on-site work like Ms. M&#039;s\u2014especially for situations that can&#039;t be handled in a normal chat format.<strong>Document vs. Document<\/strong>This article will focus on search scenarios related to &quot;[...]&quot;.<\/p>\n\n\n\n<div style=\"height:50px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<h2 id=\"aioseo-\u672c\u9023\u8f09\u306b\u3064\u3044\u3066-5\" class=\"wp-block-heading\"><a>About this series<\/a><\/h2>\n\n\n\n<p class=\"wp-block-paragraph\">This is the seventh article in the series. The series is structured into five chapters: (I) Why Super RAG, (II) How to implement it, (III) Visualizing the contents, (IV) What is happening in the field, and (V) Decision on implementation. Chapter III (Parts 4-6)<strong>The three pillars are extraction, search, and answer strategies.<\/strong>Now that all the maps are ready, we will begin Chapter IV &quot;<strong>What is happening on the ground?<\/strong>Next, we will introduce how this map functions in business scenarios.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">The 7th installment is<a href=\"https:\/\/cinnamon.ai\/en\/ideas\/super-rag-tech-blog-02\/#aioseo-3\u3064\u306e\u30d1\u30bf\u30fc\u30f3\u3092api\u547c\u3073\u51fa\u3057\u306e\u89b3\u70b9\u3067\u898b\u76f4\u3059-6\" target=\"_blank\" rel=\"noopener\" title=\"[Part 2] A Thorough Explanation of Three Embedded Patterns \u2014 Analyzing API Call Flows and Dify Expense Review\">The three embedded patterns introduced in Part 2<\/a>among<strong>Pattern \u2461 (Search-focused)<\/strong>This is an example of a real-world case. Super RAG is &quot;<strong>Preprocessing engine and search platform components<\/strong>It is used as &quot; and on top of that<strong>We develop our own logic for decomposing business-specific queries and integrating results.<\/strong>This is an architecture. The example in this article uses accident summary reports published on the internet by third parties as experimental data for a manufacturing site.<strong>Extracting accident and near-miss incidents from work plans.<\/strong>This explanation is based on a test implementation of the search pipeline.<\/p>\n\n\n\n<div style=\"height:50px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<h2 id=\"aioseo-\u696d\u52d9\u30b7\u30fc\u30f3-\u88fd\u9020\u696d\u306e\u5b89\u5168\u7ba1\u7406\u3068-ky\u6d3b\u52d5-8\" class=\"wp-block-heading\"><a>Workplace scenarios \u2014 Safety management and hazard prediction activities in manufacturing<\/a><\/h2>\n\n\n\n<p class=\"wp-block-paragraph\">In manufacturing establishments,<strong>KY activities (hazard prediction activities)<\/strong>or<strong>TBM (Toolbox Meeting)<\/strong>Risk assessments like the one described above are conducted on a daily basis before any work begins. The cycle involves identifying potential hazards in tomorrow&#039;s or today&#039;s work in advance, sharing countermeasures, and then starting the work.<\/p>\n\n\n\n<div style=\"height:30px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<figure class=\"wp-block-image size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"398\" src=\"https:\/\/cinnamon.ai\/wp-content\/uploads\/2026\/07\/01_ky_workflow_before-1024x398.png\" alt=\"\" class=\"wp-image-5113\" srcset=\"https:\/\/cinnamon.ai\/wp-content\/uploads\/2026\/07\/01_ky_workflow_before-1024x398.png 1024w, https:\/\/cinnamon.ai\/wp-content\/uploads\/2026\/07\/01_ky_workflow_before-300x117.png 300w, https:\/\/cinnamon.ai\/wp-content\/uploads\/2026\/07\/01_ky_workflow_before-768x298.png 768w, https:\/\/cinnamon.ai\/wp-content\/uploads\/2026\/07\/01_ky_workflow_before-18x7.png 18w, https:\/\/cinnamon.ai\/wp-content\/uploads\/2026\/07\/01_ky_workflow_before.png 1194w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><\/figure>\n\n\n\n<p class=\"wp-block-paragraph\"><em>Workflow for KY (Kiken Yochi - Hazard Prediction) activities (Before)<\/em><\/p>\n\n\n\n<div style=\"height:30px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p class=\"wp-block-paragraph\">At the core of this cycle is,<strong>Reference to past accident reports and near-miss incidents.<\/strong>is.<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Without even needing to invoke Heinrich&#039;s Law, what happens on the ground is<strong>Close call<\/strong>Such incidents have been accumulated in large numbers as precursors to serious accidents.<\/li>\n\n\n\n<li>In past similar tasks, what accidents or near misses occurred, what were the causes, and what countermeasures should be taken\u2014these are all things that are important within the organization.<strong>experience<\/strong>It is recorded as &quot;<\/li>\n\n\n\n<li>In the KY meeting, we will draw out this experience<strong>Let&#039;s all review the key points to keep in mind for today&#039;s work.<\/strong>That is the purpose.<\/li>\n<\/ul>\n\n\n\n<div style=\"height:30px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p class=\"wp-block-paragraph\">But what is the reality on the ground?<strong>Past accident reports and near-miss reports are indeed kept as records.<\/strong>In many cases, this data is stored on the company&#039;s internal file server in an unstructured state.<strong>The problem is that it&#039;s difficult to access when you need it.<\/strong>That&#039;s the point.<\/p>\n\n\n\n<div style=\"height:30px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Even if you search by file name,<strong>The file name is a business-like name such as &quot;20230815_Accident Report_Manufacturing Department 2.docx&quot;.<\/strong>And it doesn&#039;t show the contents.<\/li>\n\n\n\n<li>Even when categorized into folders, they are organized along departmental or formatal axes, such as &quot;Manufacturing Department 2 Folder&quot; or &quot;Maintenance Case Studies Folder.&quot;<strong>&quot;Examples under the same working conditions&quot; cannot be used as a reference.<\/strong><\/li>\n\n\n\n<li>in the end,<strong>A veteran&#039;s memory: &quot;That case from back then was similar, wasn&#039;t it?&quot;<\/strong>This leads to relying on others, and the knowledge is not internalized within the organization.<\/li>\n<\/ul>\n\n\n\n<div style=\"height:30px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p class=\"wp-block-paragraph\">In places where skilled labor is required, such as chemical plants,<strong>These sounds and smells are signs of something wrong.<\/strong>\u201d, \u201c<strong>Always check for insulated gloves in this situation.<\/strong>&quot;such as<strong>Tacit knowledge<\/strong>However, a structural problem has been pointed out: this knowledge is only accumulated in the minds of veterans. The reason why Ms. M spends time identifying cases is ultimately<strong>The process of transforming this tacit knowledge into organizational knowledge, all by one person.<\/strong>That&#039;s how it was.<\/p>\n\n\n\n<div style=\"height:50px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<h2 id=\"aioseo-\u306a\u305c\u901a\u5e38\u306erag\u3067\u306f\u8db3\u308a\u306a\u3044\u304b-\u8cea\u554f\u306b\u7b54\u3048\u308b\u3068\u6587\u66f8\u304b\u3089\u6587\u66f8\u3092\u63a2\u3059\u306e\u9055\u3044-23\" class=\"wp-block-heading\"><a>Why a standard RAG isn&#039;t enough \u2014 the difference between &quot;answering questions&quot; and &quot;searching for documents within documents.&quot;<\/a><\/h2>\n\n\n\n<p class=\"wp-block-paragraph\">Here, we will introduce the following in the first to sixth installments.<strong>Regular RAG<\/strong>Please recall how to use &quot;&quot;.<strong>Enter a question in natural language.<\/strong>The AI then extracts the relevant section from the related documents.<strong>Answer in a question-and-answer format.<\/strong>Returning this\u2014this is the basic pattern for a chat-style RAG.<\/p>\n\n\n\n<div style=\"height:30px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<figure class=\"wp-block-image size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"575\" src=\"https:\/\/cinnamon.ai\/wp-content\/uploads\/2026\/07\/02_qa_vs_doc_vs_doc-1024x575.png\" alt=\"\" class=\"wp-image-5114\" srcset=\"https:\/\/cinnamon.ai\/wp-content\/uploads\/2026\/07\/02_qa_vs_doc_vs_doc-1024x575.png 1024w, https:\/\/cinnamon.ai\/wp-content\/uploads\/2026\/07\/02_qa_vs_doc_vs_doc-300x169.png 300w, https:\/\/cinnamon.ai\/wp-content\/uploads\/2026\/07\/02_qa_vs_doc_vs_doc-768x432.png 768w, https:\/\/cinnamon.ai\/wp-content\/uploads\/2026\/07\/02_qa_vs_doc_vs_doc-18x10.png 18w, https:\/\/cinnamon.ai\/wp-content\/uploads\/2026\/07\/02_qa_vs_doc_vs_doc.png 1180w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><\/figure>\n\n\n\n<p class=\"wp-block-paragraph\"><em>&quot;Answering Questions&quot; RAG vs. &quot;Document vs. Document&quot; Pipeline<\/em><\/p>\n\n\n\n<div style=\"height:30px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p class=\"wp-block-paragraph\">Applying this to Ms. M&#039;s work, it would look like this.<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>If you ask, &quot;What accidents have occurred in the past during the container replacement process for product A?&quot;, you can retrieve relevant accident examples.<\/li>\n\n\n\n<li>If you ask, &quot;Tell me about some near-miss incidents involving working at heights during the nighttime hours,&quot; you&#039;ll likely find some plausible examples.<\/li>\n<\/ul>\n\n\n\n<div style=\"height:30px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p class=\"wp-block-paragraph\">This is certainly convenient.<strong>However, the real need for the work is to &quot;identify relevant examples without overlooking any perspectives for the entire task of tomorrow.&quot;<\/strong>That&#039;s the thing. Tomorrow&#039;s work plan is packed with multiple processes, multiple target equipment, multiple working conditions, and multiple hazard perspectives.<strong>1 question 1 answer<\/strong>So, if you ask them in order,<\/p>\n\n\n\n<div style=\"height:30px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<ul class=\"wp-block-list\">\n<li>&quot;What are some examples related to process X?&quot;<\/li>\n\n\n\n<li>&quot;What are some examples related to process Y?&quot;<\/li>\n\n\n\n<li>&quot;What are some examples of how to use Equipment A?&quot;<\/li>\n\n\n\n<li>&quot;What are some examples of nighttime work?&quot;<\/li>\n\n\n\n<li>&quot;What are some examples in situations like XX?&quot;<\/li>\n<\/ul>\n\n\n\n<div style=\"height:30px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p class=\"wp-block-paragraph\">\u2014And so, you need to repeat the question many times.<strong>Leaks will occur, and if you consider combinations of perspectives (process X \u00d7 nighttime \u00d7 equipment A), the number of combinations will explode.<\/strong>.Also,<strong>The answers to each question are patched together from different reports, resulting in inconsistent solutions.<\/strong>It&#039;s possible.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Now, let&#039;s change our perspective.<strong>Instead of constructing a question, use tomorrow&#039;s work plan (document) itself as the starting point for your search.<\/strong>Therefore, the work plan already contains the process, target, conditions, and anticipated hazards.<strong>It mechanically breaks it down, retrieves past examples from each perspective, and returns them all together.<\/strong>\u2014This is &quot;<strong>Document vs. Document<\/strong>This is a custom search for &quot;[...]&quot;.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">To summarize the differences from the regular RAG, it&#039;s as follows:<\/p>\n\n\n\n<div style=\"height:30px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<figure class=\"wp-block-table\"><table class=\"has-fixed-layout\"><thead><tr><th>perspective<\/th><th>Standard RAG (Question \u2192 Answer)<\/th><th>Document vs. Document pipeline<\/th><\/tr><\/thead><tbody><tr><td>input<\/td><td>Questions in natural language<\/td><td>Business documents (work plans, etc.)<\/td><\/tr><tr><td>Input Interpretation<\/td><td>Search by meaning of a single question<\/td><td>Document<strong>Multiple perspectives<\/strong>Decompose into<\/td><\/tr><tr><td>search<\/td><td>1 search<\/td><td>Each perspective<strong>Multi-stage search<\/strong><\/td><\/tr><tr><td>output<\/td><td>One answer<\/td><td>Different perspectives<strong>Related Documents<\/strong><\/td><\/tr><tr><td>Suitable tasks<\/td><td>Individual inquiries<\/td><td>Risk assessment and preparation<\/td><\/tr><\/tbody><\/table><\/figure>\n\n\n\n<div style=\"height:30px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p class=\"wp-block-paragraph\">&quot;<strong>The question is in writing.<\/strong>This difference fundamentally changes how you structure your search.<strong>The system analyzes input documents, breaks them down into different perspectives, constructs searches for each perspective, and integrates the results.<\/strong>\u2014This series of processes constitutes the &quot;document vs. document&quot; search pipeline.<\/p>\n\n\n\n<div style=\"height:50px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<h2 id=\"aioseo-\u691c\u7d22\u30a2\u30fc\u30ad\u30c6\u30af\u30c1\u30e3-\u6587\u66f8-vs-\u6587\u66f8\u306e\u30ab\u30b9\u30bf\u30e0\u30d1\u30a4\u30d7\u30e9\u30a4\u30f3-43\" class=\"wp-block-heading\"><a>Search Architecture \u2014 Custom &quot;Document vs. Document&quot; Pipelines<\/a><\/h2>\n\n\n\n<p class=\"wp-block-paragraph\">I will draw a typical pipeline that applies to Ms. M&#039;s work.<\/p>\n\n\n\n<div style=\"height:30px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<figure class=\"wp-block-image size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"407\" src=\"https:\/\/cinnamon.ai\/wp-content\/uploads\/2026\/07\/03_pipeline_overview-1024x407.png\" alt=\"\" class=\"wp-image-5115\" srcset=\"https:\/\/cinnamon.ai\/wp-content\/uploads\/2026\/07\/03_pipeline_overview-1024x407.png 1024w, https:\/\/cinnamon.ai\/wp-content\/uploads\/2026\/07\/03_pipeline_overview-300x119.png 300w, https:\/\/cinnamon.ai\/wp-content\/uploads\/2026\/07\/03_pipeline_overview-768x305.png 768w, https:\/\/cinnamon.ai\/wp-content\/uploads\/2026\/07\/03_pipeline_overview-18x7.png 18w, https:\/\/cinnamon.ai\/wp-content\/uploads\/2026\/07\/03_pipeline_overview.png 1410w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><\/figure>\n\n\n\n<p class=\"wp-block-paragraph\"><em>The overall picture of the &quot;document vs. document&quot; pipeline.<\/em><\/p>\n\n\n\n<div style=\"height:30px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Step 1: Enter the work plan.<\/strong> Users submit their work plan for the next day (Word\/Excel\/PDF, etc.) to the system. For example,<strong>November 15th, late night shift, manufacturing line AA, container replacement work for product B, 3 workers, tools used: chain hoist, special note: scaffolding was damp.<\/strong>This is a business document like the one shown.<\/p>\n\n\n\n<div style=\"height:30px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Step 2: Preprocessing (Super RAG API)<\/strong> Super RAG preprocesses the entered work plan. Specifically,<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>POST \/api\/v3.3\/actions\/document-extract\/<strong>Text conversion while preserving structure<\/strong>(DocReader)<\/li>\n\n\n\n<li>POST \/api\/v3.3\/actions\/chunking\/<strong>Division into meaningful units<\/strong><\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\">At this point, the work plan is structured as a collection of sections such as &quot;Process Schedule,&quot; &quot;Condition Description,&quot; and &quot;Special Notes.&quot;<strong>Units that machines can handle<\/strong>It will be organized as follows.<\/p>\n\n\n\n<div style=\"height:30px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Step 3: Queryset generation (the process of assembling it on the API frontend)<\/strong> This is the &quot;document vs. document&quot; pipeline.<strong>Responsibilities of the API frontend<\/strong>That&#039;s it. From the pre-processed work plan,<strong>We construct search queries based on different perspectives.<\/strong>In terms of safety management, typical perspectives include the following:<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>Process perspective<\/strong>&quot;Container replacement work,&quot; &quot;Work using chain hoists&quot;<\/li>\n\n\n\n<li><strong>Target perspective<\/strong>&quot;Manufacturing Line AA&quot; &quot;Handling of Product B&quot;<\/li>\n\n\n\n<li><strong>Conditional perspective<\/strong>&quot;Nighttime work,&quot; &quot;Wet scaffolding environment,&quot; &quot;Three-person team&quot;<\/li>\n\n\n\n<li><strong>Risk perspective<\/strong>&quot;Risk of falling,&quot; &quot;Risk of exposure to heat and chemicals,&quot; &quot;Risk of reduced visibility&quot;<\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\">These are each a set of independent search queries.<strong>The perspective and axis are determined by the specific task.<\/strong>Therefore, this is set via the prompt.<\/p>\n\n\n\n<div style=\"height:30px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Step 4: Multi-stage search (calling a Super RAG search)<\/strong> The assembled queries for each perspective are sent to the Super RAG search endpoint (POST \/api\/v3\/workflows\/retrieve\/). The search targets are folders containing past accident reports and near-miss incidents. For each perspective, a list of document fragments is returned in order of relevance.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Here, we introduced in the fifth installment<strong>Hybrid search (based on both meaning and word)<\/strong>This works.<strong>The scaffolding is wet.<\/strong>Situational descriptions (more semantic) like this and &quot;<strong>Manufacturing line AA<\/strong>Even in searches that include both proper nouns (word-like terms) such as &quot;[...]&quot;, running both axes simultaneously reduces the chances of missing results.<\/p>\n\n\n\n<div style=\"height:30px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Step 5: Integrating Results (The process of assembling them on the API frontend)<\/strong> The examples that came back from each perspective,<strong>Duplicate removal, score consolidation, and re-evaluation of relevance.<\/strong>Integrating with,<strong>Final reference case file list<\/strong>This is how it is presented to users. If the same case is relevant from multiple perspectives, it can be weighted here to be placed higher as a particularly important case.<\/p>\n\n\n\n<div style=\"height:30px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Step 6: Present with evidence<\/strong> The final result is,<strong>Case title, related perspectives, quoted section from the original document, source.<\/strong>It will be presented along with this. At the KY meeting, Ms. M said,<strong>This case was a hit from both a process and a conditions perspective. The original accident report can be found here.<\/strong>You can share it with supporting evidence, like this:<\/p>\n\n\n\n<div style=\"height:30px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p class=\"wp-block-paragraph\">This pipeline<strong>The part that Super RAG is responsible for<\/strong>This involves preprocessing in step 2 and searching in step 4.<strong>The part handled by the API frontend<\/strong>This involves step 1 data input, step 3 query decomposition, step 5 integration, and step 6 result display. Super RAG is &quot;<strong>parts<\/strong>This involves incorporating it as a search engine and building a business-specific search flow in-house\u2014this is the structure of Pattern \u2461 (search-focused).<\/p>\n\n\n\n<hr class=\"wp-block-separator has-alpha-channel-opacity\"\/>\n\n\n\n<p class=\"wp-block-paragraph\">\ud83d\udcfa <strong>Demo video<\/strong>You can see how the pipeline described so far actually works in this video. It&#039;s a demo application built using the Super RAG API.<strong>The entire process from uploading a work plan to the presentation of related case studies.<\/strong>, and<strong>Accuracy measurement using LLM rerank (Hit@1: 84.6%)<\/strong>You can check everything up to this point.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">The video shows you four screens in order: defining relevance, setting up search parameters, measuring accuracy, and actually executing a search. This is an experimental implementation of the API frontend by our company. While the article focuses on explaining the concepts, the video also includes specific numerical data (such as running searches with a maximum of 8 chunks, and accuracy improvement from 69.2% before reranking to 84.6% after reranking).<\/p>\n\n\n\n<div style=\"height:30px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<figure class=\"wp-block-embed aligncenter is-type-video is-provider-youtube wp-block-embed-youtube wp-embed-aspect-16-9 wp-has-aspect-ratio\"><div class=\"wp-block-embed__wrapper\">\n<iframe loading=\"lazy\" title=\"Super RAG API Application Example: Hazard Prediction \u2013 Searching past accidents from work plans to identify risks.\" width=\"500\" height=\"281\" src=\"https:\/\/www.youtube.com\/embed\/iX-Bjj4rrB8?feature=oembed\" frameborder=\"0\" allow=\"accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share\" referrerpolicy=\"strict-origin-when-cross-origin\" allowfullscreen><\/iframe>\n<\/div><\/figure>\n\n\n\n<hr class=\"wp-block-separator has-alpha-channel-opacity\"\/>\n\n\n\n<div style=\"height:70px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<h2 id=\"aioseo-\u306a\u305c\u6587\u66f8-vs-\u6587\u66f8\u691c\u7d22\u306f-super-rag\u306e\u524d\u51e6\u7406\u3092\u5fc5\u8981\u3068\u3059\u308b\u304b-70\" class=\"wp-block-heading\"><a>Why does &quot;document vs. document&quot; search require Super RAG preprocessing?<\/a><\/h2>\n\n\n\n<p class=\"wp-block-paragraph\">Here, I will explain the core differentiating factor of this article. Custom search for &quot;document vs. document&quot; is...<strong>Both the input document and the search target document are business documents.<\/strong>The point is a major feature. And both are<strong>Japanese + complex layout + tables + diagrams<\/strong>&quot; is mixed in,<strong>Pre-processing is difficult.<\/strong>These are the materials.<\/p>\n\n\n\n<div style=\"height:30px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<figure class=\"wp-block-image size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"553\" src=\"https:\/\/cinnamon.ai\/wp-content\/uploads\/2026\/07\/04_why_super_rag_preprocessing-1024x553.png\" alt=\"\" class=\"wp-image-5116\" srcset=\"https:\/\/cinnamon.ai\/wp-content\/uploads\/2026\/07\/04_why_super_rag_preprocessing-1024x553.png 1024w, https:\/\/cinnamon.ai\/wp-content\/uploads\/2026\/07\/04_why_super_rag_preprocessing-300x162.png 300w, https:\/\/cinnamon.ai\/wp-content\/uploads\/2026\/07\/04_why_super_rag_preprocessing-768x415.png 768w, https:\/\/cinnamon.ai\/wp-content\/uploads\/2026\/07\/04_why_super_rag_preprocessing-18x10.png 18w, https:\/\/cinnamon.ai\/wp-content\/uploads\/2026\/07\/04_why_super_rag_preprocessing.png 1174w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><\/figure>\n\n\n\n<p class=\"wp-block-paragraph\"><em>Why document vs. document search relies on Super RAG preprocessing.<\/em><\/p>\n\n\n\n<div style=\"height:30px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Contents of the input side (work plan)<\/strong><\/p>\n\n\n\n<p class=\"wp-block-paragraph\">A work plan may look simple, but its content is surprisingly multi-layered.<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>process chart<\/strong>A table listing the steps of the work in order (Step 1, Step 2, ...). Cell merging may be used to list &quot;Tools Used&quot; across multiple steps.<\/li>\n\n\n\n<li><strong>Table of working conditions<\/strong>The date, time, location, person in charge, and any special notes are written in a table format.<\/li>\n\n\n\n<li><strong>Equipment diagrams and work area diagrams<\/strong>: Drawings of the equipment to be worked on and a floor plan showing the work area are attached.<\/li>\n\n\n\n<li><strong>Main text<\/strong>Notes, special information, and reference information are written in text.<\/li>\n<\/ul>\n\n\n\n<div style=\"height:30px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p class=\"wp-block-paragraph\">These are<strong>It cannot be read in meaningful units.<\/strong>&quot; and Step 3&#039;s perspective decomposition cannot be performed. The merged cells in the table break down, the correspondence between &quot;Tools Used&quot; and &quot;Procedure&quot; is broken, and the figure caption (&quot;<strong>Figure 1: Working area<\/strong>If something like this is separated from the text or chunked in an inappropriate location,<strong>Extraction will no longer be possible along the perspective axis.<\/strong>.<\/p>\n\n\n\n<div style=\"height:30px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Content of the search target (accident reports, near misses)<\/strong><\/p>\n\n\n\n<p class=\"wp-block-paragraph\">The search criteria themselves are just as complex.<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>Section structure of the accident report<\/strong>The chapter structure is often fixed, including sections such as &quot;Accident Overview,&quot; &quot;Circumstances of Occurrence,&quot; &quot;Direct Cause,&quot; &quot;Root Cause,&quot; &quot;Corrective Measures,&quot; and &quot;Measures to Prevent Recurrence.&quot;<\/li>\n\n\n\n<li><strong>On-site photos and diagrams<\/strong>This includes photographs of the accident scene, diagrams illustrating the extent of the damage, and flowcharts of the work procedures.<\/li>\n\n\n\n<li><strong>Table of damage status<\/strong>: Damaged parts, extent, and repair costs are organized in a table format.<\/li>\n\n\n\n<li><strong>Handwritten elements<\/strong>: A scanned handwritten document containing a near-miss report filled out on-site.<\/li>\n<\/ul>\n\n\n\n<div style=\"height:30px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p class=\"wp-block-paragraph\">These<strong>The structure is preserved while being included in the search.<\/strong>This is the premise for finding &quot;similar cases that occurred under the same working conditions.&quot; For example,<strong>The accident occurred during container replacement work, and the root cause was moisture on the scaffolding.<\/strong>When you want to search using a combination of conditions like this, if the section structure is broken, the search will mix all the text together, and the results will be buried in noise.<\/p>\n\n\n\n<div style=\"height:30px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>The three pillars of the 4th to 6th installments support the document quality on both sides.<\/strong><\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Here, the three pillars of Super RAG that we introduced in parts 4-6 come into play as the foundation supporting the document quality on both sides.<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>Part 4 (Structure preservation using DocReader)<\/strong>Table cell merging, figure captions, chapter structure\u2014because these remain intact, both the input work plan and the searchable accident report are preserved.<strong>Both can be mapped at the same granularity.<\/strong><\/li>\n\n\n\n<li><strong>Fifth installment (Hybrid Search)<\/strong>Even for search queries that mix both &quot;wet scaffolding&quot; (meaning-focused) and &quot;manufacturing line AA&quot; (word-focused), Super RAG can handle both axes with a single call.<\/li>\n\n\n\n<li><strong>6th installment (Dictionary preprocessing)<\/strong>Internal company terminology and industry jargon such as &quot;Line AA,&quot; &quot;TBM,&quot; &quot;KYT,&quot; and &quot;PRTR substances&quot; are standardized in a glossary before being searchable.<\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\">In other words, custom searches for &quot;document vs. document&quot; are:<strong>It won&#039;t work if the in-house development team simply handles query decomposition and integration.<\/strong>.<strong>Both documents have been pre-processed to a high quality.<\/strong>The premise is that Super RAG provides the pre-processing layer for that purpose.<\/p>\n\n\n\n<div style=\"height:50px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Differentiation from individual LLM development<\/strong><\/p>\n\n\n\n<p class=\"wp-block-paragraph\">To be frank,<strong>By combining the OpenAI GPT API with your own vector database, you can create the basic structure of a search pipeline.<\/strong>Frameworks like LangChain allow you to write code for perspective decomposition, multi-stage search, and integration. The implementation patterns themselves are now something anyone can write.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">However, when trying to create a search pipeline that works in the field,<strong>I&#039;m stuck on implementing the preprocessing layer.<\/strong>.<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Reading data while preserving merged cells in a table is difficult for general-purpose OCR and PyMuPDF.<\/li>\n\n\n\n<li>Converting handwritten near-miss reports from scanned PDFs into text and chunking them is a very difficult process to implement from scratch.<\/li>\n\n\n\n<li>Maintaining the link between figure captions and the text requires accurate layout analysis.<\/li>\n\n\n\n<li>Building a pre-processing engine for business documents that handle &quot;Japanese language + complex layout + tables + diagrams + handwritten elements&quot; in-house with sufficient quality for business use, and then continuing to maintain it afterward, is...<strong>Specialized team and ongoing investment<\/strong>Required<\/li>\n<\/ul>\n\n\n\n<div style=\"height:30px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p class=\"wp-block-paragraph\">Super RAG is this<strong>Providing a preprocessing layer and a search infrastructure.<\/strong>So, the customer is searching for &quot;document vs. document&quot;.<strong>You can focus on the higher layers (query decomposition, perspective design, and integration).<\/strong>Making it so\u2014this is a clear distinction from individual LLM development.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Furthermore, Super RAG publishes its specifications in OpenAPI format (JSON file), and if it is loaded into a coding agent (AI-based development support) that has become popular in recent years,<strong>Humans<\/strong><strong>You can build business-specific custom pipelines in a short amount of time without needing to understand the API in detail.<\/strong>This is well-suited to a modern development style. The demo application in this article was also built in just about three days by combining the Streamlit and Super RAG APIs. The philosophy of Pattern \u2461, which involves &quot;using Super RAG as a component and assembling business-specific features in-house,&quot; is well-suited to this kind of rapid development.<\/p>\n\n\n\n<div style=\"height:50px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<h2 id=\"aioseo-\u3053\u306e\u30a2\u30fc\u30ad\u30c6\u30af\u30c1\u30e3\u306f\u4f55\u3092\u610f\u5473\u3059\u308b\u304b-\u7b2c2\u56de\u7b2c6\u56de\u306e\u8a2d\u8a08\u5224\u65ad\u3068\u306e\u6574\u5408-107\" class=\"wp-block-heading\"><a>What does this architecture mean? \u2014 Consistency with the design decisions in Parts 2 and 6.<\/a><\/h2>\n\n\n\n<p class=\"wp-block-paragraph\">Here, we&#039;ll tie up the loose ends from the series.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Full-scale utilization of Pattern \u2461 (Search-focused)<\/strong><\/p>\n\n\n\n<p class=\"wp-block-paragraph\">In the second installment, when we introduced the &quot;three embedded patterns,&quot; pattern \u2461 was &quot;<strong>Preprocessing and search are handled by Super RAG, while higher-level orchestration is done in-house.<\/strong>He introduced it by saying, &quot;This case is exactly<strong>Full-scale application of pattern \u2461<\/strong>And further<strong>We control the decomposition and integration logic of business-specific queries.<\/strong>This enables &quot;document vs. document&quot; searching.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">In the second installment, we illustrated the expense review process for Dify integration as a representative example of pattern \u2461.<strong>Even with the same pattern \u2461, there are multiple options for how to construct the higher-level orchestration.<\/strong>Using a general-purpose workflow platform like Dify, or, as in this case study...<strong>Build business-specific search logic using our own code.<\/strong>Both forms are included in pattern \u2461. In cases where the process is highly specific to the task, the latter is easier to optimize for the task.<\/p>\n\n\n\n<div style=\"height:30px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>The sixth division of labor design and alignment<\/strong><\/p>\n\n\n\n<p class=\"wp-block-paragraph\">In the 6th installment,<strong>Multi-hop inference orchestration is a division of labor with the agent-based architecture.<\/strong>The idea put forward as &quot;&quot; is, in this case, &quot;<strong>Orchestration of custom search<\/strong>This is how it appears: Super RAG provides knowledge source understanding and search quality, while the in-house development team handles the decomposition and integration of business-specific queries.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">In the sixth installment, we cited &quot;agent-based infrastructure such as Dify for multi-hop needs&quot; as an example.<strong>In the demo, we built a general-purpose pipeline, but ideally, we want to inject business-specific details to optimize it.<\/strong>This is a domain. The &quot;design of perspective axes,&quot; &quot;logic of query decomposition,&quot; and &quot;weighting of integration&quot; are unique to each business, and accuracy can be further improved by optimizing at the code level, not just by adjusting prompts. In this case, a method of building the search logic with a coding agent is suitable. Even within the same pattern \u2461, the choice is between using Dify, which is more stable and visible, or prioritizing the efficiency and flexibility of the coding agent and building a custom logic, even if it reduces visibility.<strong>High degree of job specificity<\/strong>You should choose based on that.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">If your company lacks an in-house development team or requires specialized knowledge for search pipeline design, you can choose to collaborate with a Cinnamon AI partner for design and implementation. More details will be provided in a later article, but if you are interested, please feel free to contact us using the inquiry form at the bottom of the page.<\/p>\n\n\n\n<div style=\"height:50px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<h2 id=\"aioseo-\u6a2a\u5c55\u958b\u8ef8a-\u88fd\u9020\u696d\u5185\u3067\u306e\u4ed6\u306e\u6587\u66f8-vs-\u6587\u66f8\u30b7\u30fc\u30f3-116\" class=\"wp-block-heading\"><a>Horizontal Expansion\/Axis A \u2014 Other &quot;Document vs. Document&quot; Scenarios within Manufacturing<\/a><\/h2>\n\n\n\n<p class=\"wp-block-paragraph\">The &quot;document vs. document&quot; pipeline isn&#039;t limited to KY (Kiken Yochi - Hazard Prediction) activities.<strong>Even if you only look at the manufacturing industry<\/strong>There are many tasks that can be designed using the same pattern.<\/p>\n\n\n\n<div style=\"height:30px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<figure class=\"wp-block-image size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"446\" src=\"https:\/\/cinnamon.ai\/wp-content\/uploads\/2026\/07\/05a_manufacturing_horizontal-1024x446.png\" alt=\"\" class=\"wp-image-5117\" srcset=\"https:\/\/cinnamon.ai\/wp-content\/uploads\/2026\/07\/05a_manufacturing_horizontal-1024x446.png 1024w, https:\/\/cinnamon.ai\/wp-content\/uploads\/2026\/07\/05a_manufacturing_horizontal-300x131.png 300w, https:\/\/cinnamon.ai\/wp-content\/uploads\/2026\/07\/05a_manufacturing_horizontal-768x335.png 768w, https:\/\/cinnamon.ai\/wp-content\/uploads\/2026\/07\/05a_manufacturing_horizontal-18x8.png 18w, https:\/\/cinnamon.ai\/wp-content\/uploads\/2026\/07\/05a_manufacturing_horizontal.png 1209w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><\/figure>\n\n\n\n<p class=\"wp-block-paragraph\"><em>Proposed methods for utilizing &quot;document vs. document&quot; within the manufacturing industry.<\/em><\/p>\n\n\n\n<div style=\"height:50px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Design Review<\/strong><\/p>\n\n\n\n<p class=\"wp-block-paragraph\">In design reviews for new products and new equipment,<strong>Design drawings, specifications, and FMEA (Failure Mode and Effects Analysis) reports.<\/strong>Enter this as input.<strong>Past trouble cases with similar designs, design change history, and parts recall information.<\/strong>There is a need to extract this. The perspective axis is &quot;<strong>Functions, components, conditions, and failure modes<\/strong>For example, it can be used to ensure that designers and reviewers thoroughly review past cases when preparing for design review meetings.<\/p>\n\n\n\n<div style=\"height:30px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Maintenance work<\/strong><\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Using signs of equipment failure (such as abnormal vibration noises) obtained from daily inspections as input,<strong>Past similar failure cases, repair procedures, and replacement parts information<\/strong>There is a need to extract this. The perspective axis is &quot;<strong>Equipment type, symptoms, operating conditions, age<\/strong>These are used for initial response to sudden breakdowns and for preliminary planning of regular maintenance schedules.<\/p>\n\n\n\n<div style=\"height:30px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Patent search<\/strong><\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Enter the proposed configuration and technical specifications for the newly developed system.<strong>Past similar patents, technical papers, and internal technical reports<\/strong>There is a need to extract this. The perspective axis is &quot;<strong>Technical fields, components, effects, and application areas<\/strong>These are used for invention discovery and prior art searches before filing a patent application. Pre-processing that allows you to read patent drawings (flowcharts, structural diagrams) is also effective here.<\/p>\n\n\n\n<div style=\"height:30px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>quality control<\/strong><\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Input the inspection report of the defective product, the actual product photos, the analysis data of the defective sample, etc.<strong>Past similar failure cases, cause analysis reports, and countermeasures.<\/strong>There is a need to extract this. The perspective axis is &quot;<strong>Product type, symptoms, process of occurrence, causal chain<\/strong>These are examples of how they are used in handling quality complaints and in improving manufacturing lines.<\/p>\n\n\n\n<div style=\"height:30px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>The more clearly defined the perspective and framework of a task, the more likely it is to benefit from a pipeline.<\/strong><\/p>\n\n\n\n<p class=\"wp-block-paragraph\">What these manufacturing scenarios have in common is<strong>The perspective and framework are clearly defined as part of the work.<\/strong>For example, design reviews focus on &quot;functions, components, and failure modes,&quot; maintenance work focuses on &quot;equipment type, symptoms, and operating conditions,&quot; and quality control focuses on &quot;product type, symptoms, and process of occurrence.&quot;<strong>The axis of disassembly that veterans use in their minds<\/strong>However, the more a task can be articulated as operational know-how, the easier it is to create a &quot;document vs. document&quot; pipeline.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Conversely, the first step in designing a &quot;document vs. document&quot; pipeline is:<strong>Interview the person in charge of the work to extract their perspective.<\/strong>That&#039;s the point. Understanding the business operations is the starting point, before getting into technical details.<\/p>\n\n\n\n<div style=\"height:50px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<h2 id=\"aioseo-\u6a2a\u5c55\u958b\u8ef8b-\u696d\u7a2e\u3092\u8d85\u3048\u308b\u30c8\u30e9\u30d6\u30eb\u691c\u7d22\u30d1\u30bf\u30fc\u30f3-131\" class=\"wp-block-heading\"><a>Horizontal Expansion\/Axis B \u2014 Trouble Search Patterns Across Industries<\/a><\/h2>\n\n\n\n<p class=\"wp-block-paragraph\">The &quot;document vs. document&quot; pattern extends beyond manufacturing.<strong>Cross-industry &quot;troubleshooting&quot; patterns<\/strong>It can also be viewed as such.<\/p>\n\n\n\n<div style=\"height:30px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<figure class=\"wp-block-image size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"342\" src=\"https:\/\/cinnamon.ai\/wp-content\/uploads\/2026\/07\/05b_cross_industry_horizontal-1024x342.png\" alt=\"\" class=\"wp-image-5118\" srcset=\"https:\/\/cinnamon.ai\/wp-content\/uploads\/2026\/07\/05b_cross_industry_horizontal-1024x342.png 1024w, https:\/\/cinnamon.ai\/wp-content\/uploads\/2026\/07\/05b_cross_industry_horizontal-300x100.png 300w, https:\/\/cinnamon.ai\/wp-content\/uploads\/2026\/07\/05b_cross_industry_horizontal-768x257.png 768w, https:\/\/cinnamon.ai\/wp-content\/uploads\/2026\/07\/05b_cross_industry_horizontal-18x6.png 18w, https:\/\/cinnamon.ai\/wp-content\/uploads\/2026\/07\/05b_cross_industry_horizontal.png 1200w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><\/figure>\n\n\n\n<p class=\"wp-block-paragraph\"><em>Cross-industry troubleshooting patterns<\/em><\/p>\n\n\n\n<div style=\"height:30px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p class=\"wp-block-paragraph\">Let&#039;s take a quick overview.<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>IT<\/strong><strong>Operation<\/strong>: Log of incident alerts \u2192 Past similar incidents and response procedures, post-incident evaluation<\/li>\n\n\n\n<li><strong>Legal<\/strong>Documents subject to contract review \u2192 Past similar contracts, revision history, and relevant laws and regulations<\/li>\n\n\n\n<li><strong>Support work (call center, etc.)<\/strong>Complaint details \u2192 Past similar complaints and response records\/success stories<\/li>\n\n\n\n<li><strong>medical care<\/strong>Today&#039;s incident report \u2192 Past similar cases, cause analysis, and preventive measures<\/li>\n<\/ul>\n\n\n\n<div style=\"height:30px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p class=\"wp-block-paragraph\">both,&quot;<strong>I want to retrieve other related documents based on a specific business document.<\/strong>This is the pattern. Even if the industries are different,<strong>At the core of our work is the need to &quot;extract past experiences as organizational knowledge.&quot;<\/strong>As long as it&#039;s within the same &quot;document vs. document&quot; pipeline, it can be designed.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">What&#039;s important here is,<strong>The perspective changes depending on the industry.<\/strong>For example, in IT operations, it would be &quot;type of failure, scope of impact, and related services&quot;; in legal matters, it would be &quot;contract type, clauses, and risk level&quot;; and in call centers, it would be &quot;product, inquiry type, and customer attributes&quot;\u2014and so on.<strong>Understanding the business is the starting point.<\/strong>The resulting composition is the same as the manufacturing scene in Axis A.<\/p>\n\n\n\n<div style=\"height:50px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<h2 id=\"aioseo-\u30b9\u30c8\u30fc\u30ea\u30fc\u306e\u7d9a\u304d-super-rag\u3092\u5165\u308c\u305f\u5f8c\u306e-m\u3055\u3093\u306e\u696d\u52d9-143\" class=\"wp-block-heading\"><a>Continuing the story \u2014 M&#039;s work after installing Super RAG<\/a><\/h2>\n\n\n\n<p class=\"wp-block-paragraph\">Let&#039;s return to the work of Ms. M, who was introduced at the beginning. After implementing Super RAG in the Pattern \u2461 configuration, how will Ms. M&#039;s monthly work change?<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">The preparation for the pre-risk assessment, which had previously taken a lot of time,<strong>Significantly shortened<\/strong>It will be done. When you enter tomorrow&#039;s work plan,<strong>List of related case studies organized by perspective<\/strong>However, the response comes back with supporting evidence. Ms. M reviews it visually and then prepares it as material for the KY meeting\u2014that&#039;s the current process.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">The content of the KY meeting has also changed. The part that used to rely on the veterans&#039; memories has changed.<strong>Referencing past cases as an organization<\/strong>This change allows everyone, including new employees, to discuss things from the same premise.<strong>This case is similar to one that happened three years ago.<\/strong>&quot;<strong>The countermeasures at that time were written as follows:<\/strong>This kind of exchange proceeds with supporting evidence. You&#039;ll also find examples that serve as the basis for the veteran&#039;s advice.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">And Ms. M was also able to reallocate her own work time. She was able to use the half day she had been spending searching for case studies.<strong>Review of business process improvements, training of new employees, on-site inspections.<\/strong>such as,<strong>Higher value-added activities<\/strong>It will be redirected to this. This is because AI<strong>Replacing people<\/strong>rather than<strong>People can focus on more essential work.<\/strong>This is one form of modern AI utilization, aiming to achieve this goal.<\/p>\n\n\n\n<div style=\"height:30px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p class=\"wp-block-paragraph\">Furthermore, at the meeting,<strong>What were the final improvements made to the measures taken in this case?<\/strong>\u201d, \u201c<strong>What aspects did you use as evidence to determine that they were highly relevant?<\/strong>There is also a scene where M asks additional questions to Super RAG in a dialogue format. What is important here is,<strong>You can first narrow down similar cases and &quot;fix the document,&quot; and then repeatedly ask questions about that fixed document.<\/strong>This is the point. In a typical RAG, the documents referenced change every time the question is changed, making consistent analysis difficult, but in this pipeline...<strong>Based on a carefully selected set of case studies, we will delve into the causes, preventive measures, and the effectiveness of countermeasures from multiple perspectives.<\/strong>This is possible. This is what we introduced in the 6th installment.<strong>Simple multi-stage inference that can be covered by repeating single-hop steps<\/strong>This shows how it&#039;s being used in document-fixed mode. Here we see how pre-risk assessment and additional questions using document-fixed mode are being effectively combined in practice.<\/p>\n\n\n\n<div style=\"height:50px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<h2 id=\"aioseo-\u6301\u3061\u5e30\u308a\u691c\u8a0e\u6750\u6599\u81ea\u696d\u52d9\u306e\u6587\u66f8-vs-\u6587\u66f8\u30b7\u30fc\u30f3\u68da\u5378\u3057\u30d5\u30ed\u30fc-149\" class=\"wp-block-heading\"><a>[Things to take home and consider] Inventory of &quot;document vs. document&quot; scenarios in your own work<\/a>flow<\/h2>\n\n\n\n<p class=\"wp-block-paragraph\">Thank you for reading this far,<strong>Are there similar situations in our work where we have to &quot;search for documents within documents&quot;?<\/strong>You might be thinking, &quot;How do I figure that out?&quot; We&#039;ve prepared a simple sheet to help you get a better idea.<\/p>\n\n\n\n<div style=\"height:30px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<figure class=\"wp-block-image size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"366\" src=\"https:\/\/cinnamon.ai\/wp-content\/uploads\/2026\/07\/06_takeaway_sheet-1024x366.png\" alt=\"\" class=\"wp-image-5119\" srcset=\"https:\/\/cinnamon.ai\/wp-content\/uploads\/2026\/07\/06_takeaway_sheet-1024x366.png 1024w, https:\/\/cinnamon.ai\/wp-content\/uploads\/2026\/07\/06_takeaway_sheet-300x107.png 300w, https:\/\/cinnamon.ai\/wp-content\/uploads\/2026\/07\/06_takeaway_sheet-768x274.png 768w, https:\/\/cinnamon.ai\/wp-content\/uploads\/2026\/07\/06_takeaway_sheet-18x6.png 18w, https:\/\/cinnamon.ai\/wp-content\/uploads\/2026\/07\/06_takeaway_sheet.png 1428w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><\/figure>\n\n\n\n<p class=\"wp-block-paragraph\"><em>Inventory sheet for &quot;document vs. document&quot; scenarios in your own work<\/em><\/p>\n\n\n\n<div style=\"height:30px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p class=\"wp-block-paragraph\">From five perspectives, each is:<strong>Tasks\/Deliverables\/Tips<\/strong>We will focus on this point.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Perspective 1: Identify &quot;document \u2192 document&quot; scenarios.<\/strong><\/p>\n\n\n\n<p class=\"wp-block-paragraph\">In my work,<strong>Using one document as a starting point, search for other related documents.<\/strong>We will identify 3 to 5 such scenes. These include the KY activities, design reviews, quality control, and maintenance work introduced in this article.<strong>Searching for and cross-referencing information in multiple parts of one document with information from another document.<\/strong>Look for the work scenes.<strong>Tips<\/strong>&quot;<strong>What the veteran is thinking in that scene<\/strong>Observing this will make the scene easier to visualize.<\/p>\n\n\n\n<div style=\"height:30px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Perspective 2: Location and format of reference documents<\/strong><\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Past documents that you want to refer to in each scene (accident reports, past projects, design review records, claim handling records, etc.)<strong>Where and in what format is it saved?<\/strong>We will check.<strong>Tips<\/strong>Understanding the proportions of Excel, Word, PDF, scanned PDF, and handwritten documents will help identify areas where Super RAG preprocessing is particularly effective (complex layouts, scanned documents, and handwritten documents).<\/p>\n\n\n\n<div style=\"height:30px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Perspective 3: Identifying the perspective axes<\/strong><\/p>\n\n\n\n<p class=\"wp-block-paragraph\">In each scene<strong>The axis of alignment that the person in charge of the business uses in their mind<\/strong>We will extract this information through business interviews.<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>KY (Hazard Prediction) Activities \u2192 Process, Target, Conditions, and Hazard Perspectives<\/li>\n\n\n\n<li>Design Review \u2192 Function, Components, Conditions, Failure Modes<\/li>\n\n\n\n<li>Quality control \u2192 Product, symptoms, process, cause chain<\/li>\n\n\n\n<li>Maintenance work \u2192 Equipment type, symptoms, operating conditions, age<\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>The clearer the perspective and framework of a task, the more likely it is to benefit from a &quot;document vs. document&quot; pipeline.<\/strong>\u2014This is one of the key messages of this article.<strong>Tips<\/strong>The perspective axis doesn&#039;t need to be perfect.<strong>With 3-4 axes, it will almost certainly work.<\/strong>.<\/p>\n\n\n\n<div style=\"height:30px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Perspective 4: Confirmation of API usage patterns<\/strong><\/p>\n\n\n\n<p class=\"wp-block-paragraph\">We will now decide which of the three built-in patterns introduced in the second installment will be used for implementation.<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>Pattern \u2460 (Super RAG center)<\/strong>: When it is sufficient to be able to make inquiries using the standard UI in an interactive format.<\/li>\n\n\n\n<li><strong>Pattern \u2461 (Search-focused)<\/strong>: If you want to build business-specific orchestration on top of Super RAG preprocessing and searching using a workflow platform such as Dify or your own code (<strong>The example in this article is here.<\/strong>)<\/li>\n\n\n\n<li><strong>Pattern 3 (Specialized for pre-processing)<\/strong>: For cases where you want to use Super RAG for preprocessing only and complete everything in-house, from vector database to search infrastructure and inference.<\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Tips<\/strong>The &quot;document vs. document&quot; scenario basically falls under pattern \u2461. Both using a general-purpose workflow platform like Dify and building it with your own code, as described in this article, are within pattern \u2461. However, if you use your own infrastructure (vector DB storage, search, and inference), it can also be achieved with pattern \u2462. You should choose between \u2461 and \u2462 depending on the degree of business specificity.<\/p>\n\n\n\n<div style=\"height:30px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Perspective 5: Consultation on verification during the trial.<\/strong><\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Once the inventory is complete, it&#039;s essential to verify whether you can actually use the Super RAG API with real data to perform document searches that suit your business needs. Specific parameters such as scope, duration, and number of target files will vary depending on your business requirements. While it&#039;s possible to implement this in-house using a coding agent, if that proves difficult, we also offer the option of assisting you with a free or paid Proof of Concept (PoC).<strong>Please use the contact form at the end of the article.<\/strong>Please feel free to contact us for advice.<strong>Tips<\/strong>If you could summarize and share the results from perspectives 1-4, it would greatly streamline the process of designing what to test in the initial trial.<\/p>\n\n\n\n<div style=\"height:50px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<h2 id=\"aioseo-\u6b21\u56de\u4e88\u544a-175\" class=\"wp-block-heading\"><a>Next episode preview<\/a><\/h2>\n\n\n\n<p class=\"wp-block-paragraph\">Next time (Episode 8),<strong>Use Case Example 2: Support\/FAQ\/Glossary Management<\/strong>This article will cover the world of custom search pipelines specific to your business, specifically &quot;document vs. document&quot; searches. Next time, we&#039;ll shift our perspective and...<strong>Development and operation of FAQs and glossaries that can be used for handling inquiries in support operations.<\/strong>This article will address how to organize the large volume of documents generated daily for inquiries using a Super RAG (One Paper, One Telegraph) approach.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">especially,<strong>A system for semi-automatically generating FAQ data from existing document sets.<\/strong>or,<strong>Maintain an internal glossary to maximize the effectiveness of dictionary preprocessing.<\/strong>This article is a counterpart to the one presented in the sixth installment, as it provides a practical application of the dictionary preprocessing techniques introduced in the first installment.<\/p>\n\n\n\n<div style=\"height:50px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<h2 id=\"aioseo-\u307e\u3068\u3081-178\" class=\"wp-block-heading\"><a>summary<\/a><\/h2>\n\n\n\n<p class=\"wp-block-paragraph\">A regular RAG is &quot;<strong>For each natural language question, answer one question at a time using relevant documents.<\/strong>The main pattern is &quot;. On the other hand, the work on site is &quot;<strong>I want to use a certain business document as input and retrieve a related set of documents.<\/strong>&quot;\u2014Extracting past accidents and near misses from tomorrow&#039;s work plan, extracting past trouble cases from design drawings, extracting similar past defects from defective product inspection reports\u2014&quot;<strong>Document vs. Document<\/strong>There is a high demand for &quot;[this]&quot;.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">The solution to this is the &quot;<strong>Document vs. Document Custom Search Pipeline<\/strong>That was it. For the full-scale implementation of Pattern \u2461 (search-focused), which we introduced in the second installment, Super RAG will handle preprocessing and the search infrastructure, while the business-specific query decomposition and integration will be assembled by our in-house development team.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">The premise for this pipeline to work is,<strong>Both the input documents and the accident reports and past case studies being searched maintain their structure of &quot;Japanese + complex layout + tables + diagrams&quot; and are readable.<\/strong>This is where the three pillars of Super RAG\u2014structure preservation using DocReader in the fourth session, hybrid search in the fifth session, and dictionary preprocessing in the sixth session\u2014come together and become effective.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Within the manufacturing industry, it has the potential to be widely applied to tasks with clearly defined perspectives, such as KY activities, design reviews, maintenance work, patent searches, and quality control. Furthermore, it can be applied across industries, including IT operations, legal affairs, call centers, and healthcare.<strong>Searching for documents within documents<\/strong>The same pipeline can be applied to all aspects of your work. If you think, &quot;There might be similar &#039;document vs. document&#039; scenarios in our work,&quot; please consider the five perspectives outlined in this article for further consideration. If you would like to verify this with actual data,<strong>Please feel free to contact us using the inquiry form at the end of the article.<\/strong>With free and paid trials, you can verify the preprocessing quality and search accuracy of Super RAG using your own data.<\/p>\n\n\n\n<div style=\"height:50px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p class=\"wp-block-paragraph\">&lt;Articles in this series&gt;<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><a href=\"https:\/\/cinnamon.ai\/en\/ideas\/super-rag-tech-blog-01\/\" target=\"_blank\" rel=\"noopener\" title=\"[Part 1] Achieving a Quality Beyond Standard RAG \u2014 Technical Innovations Supporting Super RAG\">[Part 1] Achieving a Quality Beyond Standard RAG \u2014 Technical Innovations Supporting Super RAG<\/a><\/li>\n\n\n\n<li><a href=\"https:\/\/cinnamon.ai\/en\/ideas\/super-rag-tech-blog-02\/\" target=\"_blank\" rel=\"noopener\" title=\"[Part 2] A Thorough Explanation of Three Embedded Patterns \u2014 Analyzing API Call Flows and Dify Expense Review\">[Part 2] A Thorough Explanation of Three Embedded Patterns \u2014 Analyzing API Call Flows and Dify Expense Review<\/a><\/li>\n\n\n\n<li><a href=\"https:\/\/cinnamon.ai\/en\/ideas\/super-rag-tech-blog-03\/\" target=\"_blank\" rel=\"noopener\" title=\"[Part 3] API Function Catalog \u2014 What can be done, and what should be handled in-house?\">[Part 3] API Function Catalog \u2014 What can be done, and what should be handled in-house?<\/a><\/li>\n\n\n\n<li><a href=\"https:\/\/cinnamon.ai\/en\/ideas\/super-rag-tech-blog-04\/\" target=\"_blank\" rel=\"noopener\" title=\"[Part 4] The Contents of Document Extraction \u2014 How is the Engine Selected?\">[Part 4] The Contents of Document Extraction \u2014 How is the Engine Selected?<\/a><\/li>\n\n\n\n<li><a href=\"https:\/\/cinnamon.ai\/en\/ideas\/super-rag-tech-blog-05\/\" title=\"[Part 5] Hybrid Search and Reranking \u2014 Making the Most of Extracted Content by Balancing &quot;Meaning&quot; and &quot;Words&quot;\">[Part 5] Hybrid Search and Reranking \u2014 Making the Most of Extracted Content by Balancing &quot;Meaning&quot; and &quot;Words&quot;<\/a><\/li>\n\n\n\n<li><a href=\"https:\/\/cinnamon.ai\/en\/ideas\/super-rag-tech-blog-06\/\" title=\"[Part 6] Super RAG Answer Strategy \u2014 Design Decisions for Mastering Single-hop Answers and 3 Techniques to Support Answer Quality\">[Part 6] Super RAG Answer Strategy \u2014 Design Decisions for Mastering Single-hop Answers and 3 Techniques to Support Answer Quality<\/a><\/li>\n\n\n\n<li>[Episode 7] Super RAG API Use Case: &quot;Document vs. Document&quot; Search \u2014 Searching past accidents and near misses from work plans to predict hazards (Scheduled for release in August 2026)<\/li>\n<\/ul>\n\n\n\n<div style=\"height:80px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p class=\"wp-block-paragraph\"><\/p>\n\n\n\n<div class=\"wp-block-buttons is-content-justification-center is-layout-flex wp-container-core-buttons-is-layout-fe48e5de wp-block-buttons-is-layout-flex\">\n<div class=\"wp-block-button\"><a class=\"wp-block-button__link wp-element-button\" href=\"https:\/\/contents.cinnamon.ai\/contact\/inquiry_blog\" target=\"_blank\" rel=\"noreferrer noopener\">contact<\/a><\/div>\n\n\n\n<div class=\"wp-block-button\"><a class=\"wp-block-button__link wp-element-button\" href=\"https:\/\/contents.cinnamon.ai\/download\/wp_superrag_dl_blog\" target=\"_blank\" rel=\"noreferrer noopener\">Super RAG Document Download<\/a><\/div>\n<\/div>\n\n\n\n<div style=\"height:150px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>","protected":false},"excerpt":{"rendered":"<p>\u306f\u3058\u3081\u306b \u672c\u8a18\u4e8b\u306e\u30e1\u30c3\u30bb\u30fc\u30b8\uff1a\u901a\u5e38\u306eRAG\u306f\u300c\u8cea\u554f\u6587\u3092\u5165\u529b \u2192 \u95a2\u9023\u6587\u66f8\u3092\u691c\u7d22\u30fb\u53c2\u7167\u3057\u30661\u554f1\u7b54\u300d\u3068\u3044\u3046\u30d1\u30bf\u30fc\u30f3\u3067\u52d5\u304d\u307e\u3059\u3002\u4e00\u65b9\u3001\u73fe\u5834\u306e\u696d\u52d9\u306b\u306f\u300c\u6587\u66f8\u305d\u306e\u3082\u306e\u3092\u5165\u529b\u306b\u3057\u3066\u3001\u95a2\u9023\u3059\u308b\u5225\u306e\u6587\u66f8\u7fa4\u3092\u5f15\u304d\u51fa\u3057\u305f\u3044\u300d\u2014\u2014\u305f\u3068\u3048\u3070\u3001 [&hellip;]<\/p>","protected":false},"featured_media":5122,"template":"","ideas-cat":[],"class_list":["post-5106","ideas","type-ideas","status-publish","has-post-thumbnail","hentry"],"acf":[],"aioseo_notices":[],"_links":{"self":[{"href":"https:\/\/cinnamon.ai\/en\/wp-json\/wp\/v2\/ideas\/5106","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/cinnamon.ai\/en\/wp-json\/wp\/v2\/ideas"}],"about":[{"href":"https:\/\/cinnamon.ai\/en\/wp-json\/wp\/v2\/types\/ideas"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/cinnamon.ai\/en\/wp-json\/wp\/v2\/media\/5122"}],"wp:attachment":[{"href":"https:\/\/cinnamon.ai\/en\/wp-json\/wp\/v2\/media?parent=5106"}],"wp:term":[{"taxonomy":"ideas-cat","embeddable":true,"href":"https:\/\/cinnamon.ai\/en\/wp-json\/wp\/v2\/ideas-cat?post=5106"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}