[{"data":1,"prerenderedAt":1499},["ShallowReactive",2],{"blog-self-learning-knowledge-base":3,"all-blogs":777,"content-query-wzPU7Rarqh":933},{"_path":4,"_dir":5,"_draft":6,"_partial":6,"_locale":7,"title":8,"description":9,"image":10,"type":5,"author":11,"readingTime":12,"createdAt":13,"updatedAt":14,"faqs":15,"body":34,"_type":770,"_id":771,"_source":772,"_file":773,"_stem":774,"_extension":775,"sitemap":776},"\u002Fblog\u002Fself-learning-knowledge-base","blog",false,"","Self-Learning Knowledge Base: What It Really Means (Hype vs. Practice)","What a self-learning knowledge base actually does: feedback loops from real user questions to content gaps to reviewed drafts, and exactly where automation stops.","\u002Fimages\u002Fblog\u002Fheaders\u002Fai-kb-roadmap.png","Sarah Chen","11 min read","2026-07-06","2026-08-08",[16,19,22,25,28,31],{"question":17,"answer":18},"What is a self-learning knowledge base?","A self-learning knowledge base is a documentation system that improves through structured feedback loops driven by real usage: it logs the questions users actually ask, identifies which ones the content could not answer, ranks those gaps by frequency, and helps draft the missing content for human review. The 'learning' is the loop from user behavior to content improvement, not an AI silently rewriting your documentation on its own.",{"question":20,"answer":21},"How is a self-learning knowledge base different from a self-updating one?","They are complementary loops. Self-updating focuses on keeping existing content accurate: syncing with source documents, flagging stale pages, tracking release changes. Self-learning focuses on growing the right content: mining real user questions to discover what is missing, then closing those gaps. A mature intelligent knowledge base runs both: updating keeps what you have true, learning ensures you have what users need.",{"question":23,"answer":24},"Does a self-learning knowledge base write its own content?","It drafts; it does not publish. An AI agent can read the gap signals and produce a first draft grounded in your related existing content and the actual user questions that revealed the gap. A subject-matter expert then reviews, corrects, and publishes. Systems that publish unreviewed AI content into a knowledge base risk feeding errors back into their own AI answers, so credible platforms keep a human approval step.",{"question":26,"answer":27},"Can an AI agent write directly into Dewstack?","Yes, into drafts only. You connect Claude to a Dewstack workspace through Settings > Connectors, or connect a coding agent such as Claude Code or Cursor with an API key that has Read and write scope. ChatGPT supports this as a developer mode connector only, because OpenAI limits write-capable connectors to developer mode. The agent can create collections, create articles as drafts, and edit drafts. It cannot publish, unpublish, or touch a live page.",{"question":29,"answer":30},"What is an intelligent knowledge base?","An intelligent knowledge base combines three capabilities: semantic search that understands questions in natural language rather than matching keywords, grounded AI answers with citations back to source documents, and feedback loops that turn usage data, unanswered questions, failed searches, ratings, into a prioritized content improvement queue. The intelligence is in the retrieval and the loop, while content accuracy remains a human responsibility.",{"question":32,"answer":33},"How long does it take for a self-learning knowledge base to improve?","The loop produces its first useful output within days, because unanswered questions start accumulating as soon as real users interact with the system. Teams that run a weekly review of unanswered questions see the answer rate climb as the highest-frequency gaps get filled first. The improvement compounds, but it is not automatic; it depends on someone acting on the queue each week.",{"type":35,"children":36,"toc":747},"root",[37,46,56,61,66,73,78,83,95,105,123,133,151,157,162,171,178,183,189,203,209,214,220,225,230,236,241,246,252,264,270,275,281,295,307,313,334,340,369,374,382,388,393,403,413,423,433,439,444,450,455,571,591,610,616,621,629,702,708,720,725,730,742],{"type":38,"tag":39,"props":40,"children":42},"element","h1",{"id":41},"self-learning-knowledge-base-what-it-really-means",[43],{"type":44,"value":45},"text","Self-Learning Knowledge Base: What It Really Means",{"type":38,"tag":47,"props":48,"children":49},"p",{},[50],{"type":38,"tag":51,"props":52,"children":53},"strong",{},[54],{"type":44,"value":55},"A self-learning knowledge base is a documentation system that improves through feedback loops driven by real usage: it captures the questions users actually ask, detects which ones the content failed to answer, ranks those gaps by frequency, and drafts the missing content for human review. The learning is in the loop, not in an AI mystically rewriting your docs while you sleep.",{"type":38,"tag":47,"props":57,"children":58},{},[59],{"type":44,"value":60},"\"Self-learning\" is one of the most abused phrases in the knowledge management market, so this article does two jobs: it separates the marketing version from the mechanism that actually exists, and it shows you how to run that mechanism so your knowledge base genuinely gets better every week.",{"type":38,"tag":47,"props":62,"children":63},{},[64],{"type":44,"value":65},"It also does something the earlier version of this article could not do. One step of the loop, the drafting step, is now automated end to end in a real product. That makes the boundary easier to describe, not fuzzier. You can point at the exact place where software stops and a person starts.",{"type":38,"tag":67,"props":68,"children":70},"h2",{"id":69},"the-hype-version-vs-the-real-version",[71],{"type":44,"value":72},"The Hype Version vs. the Real Version",{"type":38,"tag":47,"props":74,"children":75},{},[76],{"type":44,"value":77},"The hype version goes like this: install an intelligent knowledge base, and AI will autonomously learn your business, write your documentation, and keep it current forever. No such product exists, and Dewstack is not one either. AI cannot know that your refund window changed last Tuesday, or what the workaround for the new SSO bug is. That knowledge lives in your team's heads and systems, and someone has to put it into words at least once. Nothing about agent-capable tooling changes that. An agent that writes into your workspace is still writing from what you already published, plus what your users asked, plus whatever a person tells it in the prompt.",{"type":38,"tag":47,"props":79,"children":80},{},[81],{"type":44,"value":82},"What has changed is narrower and worth stating precisely. The drafting step used to require a human to read a gap report, open an editor, and retype the answer. That handoff is now genuinely automated: an agent can read the gap signals directly, write a draft into your workspace, and stop there. The review step is unchanged and mandatory. Software got one step further down the loop. It did not get to the end of it.",{"type":38,"tag":47,"props":84,"children":85},{},[86,88,93],{"type":44,"value":87},"The real version is less magical and more useful: ",{"type":38,"tag":51,"props":89,"children":90},{},[91],{"type":44,"value":92},"a knowledge base instrumented so that every user interaction teaches you something, and the lessons arrive as a ranked to-do list instead of anecdotes.",{"type":44,"value":94}," Three kinds of learning actually happen in practice:",{"type":38,"tag":47,"props":96,"children":97},{},[98,103],{"type":38,"tag":51,"props":99,"children":100},{},[101],{"type":44,"value":102},"1. Retrieval learning.",{"type":44,"value":104}," Semantic search and AI answering understand meaning, so the system handles phrasings nobody wrote articles for. Ask \"why did my card get declined\" and it finds your payments troubleshooting page even though those words never appear in it. This is genuine machine learning, baked into the models. You get it on day one without training anything.",{"type":38,"tag":47,"props":106,"children":107},{},[108,113,115,121],{"type":38,"tag":51,"props":109,"children":110},{},[111],{"type":44,"value":112},"2. Gap learning.",{"type":44,"value":114}," Every question the AI assistant cannot answer, and every search that returns nothing, is logged in the user's own words. Clustered and counted, these become a precise, demand-ranked map of what your documentation is missing. This is the heart of the self-learning loop, and it is data you have never had before: traditional knowledge bases only told you what people read, never what they ",{"type":38,"tag":116,"props":117,"children":118},"em",{},[119],{"type":44,"value":120},"needed and did not find",{"type":44,"value":122},".",{"type":38,"tag":47,"props":124,"children":125},{},[126,131],{"type":38,"tag":51,"props":127,"children":128},{},[129],{"type":44,"value":130},"3. Quality learning.",{"type":44,"value":132}," Ratings, citation click-throughs, rephrased follow-up questions, and rising-traffic-falling-satisfaction patterns flag content that exists but is not working: articles that are outdated, ambiguous, or answering the wrong version of the question.",{"type":38,"tag":47,"props":134,"children":135},{},[136,138,143,145,150],{"type":44,"value":137},"What none of these do is remove the human from publishing. The system learns ",{"type":38,"tag":116,"props":139,"children":140},{},[141],{"type":44,"value":142},"what to fix",{"type":44,"value":144},". People still decide ",{"type":38,"tag":116,"props":146,"children":147},{},[148],{"type":44,"value":149},"what is true",{"type":44,"value":122},{"type":38,"tag":67,"props":152,"children":154},{"id":153},"the-loop-step-by-step",[155],{"type":44,"value":156},"The Loop, Step by Step",{"type":38,"tag":47,"props":158,"children":159},{},[160],{"type":44,"value":161},"Here is the full cycle a self-learning knowledge base runs, and where automation ends and judgment begins:",{"type":38,"tag":47,"props":163,"children":164},{},[165],{"type":38,"tag":166,"props":167,"children":170},"img",{"alt":168,"src":169},"A left-to-right loop: a user asks, a miss is detected, the gap is ranked, the agent drafts, and an approval gate sits before a human reviews and publishes. The answer then feeds back to the start.","\u002Fimages\u002Fblog\u002Ffigures\u002Fselflearn-loop-with-gate.webp",[],{"type":38,"tag":172,"props":173,"children":175},"h3",{"id":174},"step-1-capture-real-questions",[176],{"type":44,"value":177},"Step 1: Capture real questions",{"type":38,"tag":47,"props":179,"children":180},{},[181],{"type":44,"value":182},"Users ask questions through the docs site's AI assistant or an embedded widget. Unlike search keywords (\"sso saml error\"), conversational questions are complete thoughts (\"how do I set up SAML SSO for a workspace that already uses Google sign-in?\"). This phrasing is gold: it tells you not just the topic but the exact scenario and vocabulary your users have.",{"type":38,"tag":172,"props":184,"children":186},{"id":185},"step-2-detect-the-misses",[187],{"type":44,"value":188},"Step 2: Detect the misses",{"type":38,"tag":47,"props":190,"children":191},{},[192,194,201],{"type":44,"value":193},"The system separates questions it answered with grounded, cited responses from ones it declined or answered poorly. The misses go to a log. (How grounded answering and declining work is covered in ",{"type":38,"tag":195,"props":196,"children":198},"a",{"href":197},"\u002Fblog\u002Fai-answers-from-your-docs",[199],{"type":44,"value":200},"AI answers from your docs",{"type":44,"value":202},".)",{"type":38,"tag":172,"props":204,"children":206},{"id":205},"step-3-rank-the-gaps",[207],{"type":44,"value":208},"Step 3: Rank the gaps",{"type":38,"tag":47,"props":210,"children":211},{},[212],{"type":44,"value":213},"Recurring questions cluster together. Forty users asking variations of the same unanswered question outranks one user asking something exotic. Your content backlog is now ordered by demonstrated demand, not by whoever argued loudest in the planning meeting.",{"type":38,"tag":172,"props":215,"children":217},{"id":216},"step-4-draft-grounded-in-what-exists",[218],{"type":44,"value":219},"Step 4: Draft, grounded in what exists",{"type":38,"tag":47,"props":221,"children":222},{},[223],{"type":44,"value":224},"For each gap, a draft article or FAQ entry gets written using your related existing content plus the real user questions as the brief. The draft arrives structured and phrased in the vocabulary users actually used in Step 1.",{"type":38,"tag":47,"props":226,"children":227},{},[228],{"type":44,"value":229},"This is the step that changed. Until recently, \"AI drafting\" meant a human copied a gap into a chat window, got prose back, and pasted it into an editor. Now an agent can read the gap list and write the draft into the workspace itself, with no copying in either direction. The next section covers exactly how that works and what it refuses to do.",{"type":38,"tag":172,"props":231,"children":233},{"id":232},"step-5-human-review-and-publish",[234],{"type":44,"value":235},"Step 5: Human review and publish",{"type":38,"tag":47,"props":237,"children":238},{},[239],{"type":44,"value":240},"A subject-matter expert corrects, completes, and approves. This step is non-negotiable for a reason beyond quality: in an AI-answering system, published content becomes the source for future answers. Publishing unreviewed AI drafts means the system starts learning from itself, and errors compound instead of correcting.",{"type":38,"tag":47,"props":242,"children":243},{},[244],{"type":44,"value":245},"Automating the drafting step makes this step more important, not less. When drafts are cheap, more of them arrive, and the reviewer becomes the only thing standing between a plausible-sounding paragraph and your customers. Budget review time in proportion to draft volume.",{"type":38,"tag":172,"props":247,"children":249},{"id":248},"step-6-verify-the-loop-closed",[250],{"type":44,"value":251},"Step 6: Verify the loop closed",{"type":38,"tag":47,"props":253,"children":254},{},[255,257,262],{"type":44,"value":256},"Next cycle, the questions from Step 1 that drove the new content should now get grounded answers. Answer rate ticks up, and the log fills with the ",{"type":38,"tag":116,"props":258,"children":259},{},[260],{"type":44,"value":261},"next",{"type":44,"value":263}," layer of gaps. That measurable, repeating improvement is what \"self-learning\" legitimately means.",{"type":38,"tag":67,"props":265,"children":267},{"id":266},"where-the-automation-actually-reaches",[268],{"type":44,"value":269},"Where the Automation Actually Reaches",{"type":38,"tag":47,"props":271,"children":272},{},[273],{"type":44,"value":274},"Dewstack ships both halves of the boundary: an agent that can write, and a set of rules that stop it at review. Here is the mechanism in full, so you can judge it rather than take a claim on faith.",{"type":38,"tag":172,"props":276,"children":278},{"id":277},"connecting-an-agent",[279],{"type":44,"value":280},"Connecting an agent",{"type":38,"tag":47,"props":282,"children":283},{},[284,286,293],{"type":44,"value":285},"You connect Claude to a Dewstack workspace once. Open Settings > Connectors > Add custom connector, paste the Write connector URL, which ends in ",{"type":38,"tag":287,"props":288,"children":290},"code",{"className":289},[],[291],{"type":44,"value":292},"\u002Fmcp-rw",{"type":44,"value":294},", sign in, and press Allow on an approval screen that names the workspace. There are no keys to manage and no code to write. Only account Owners and Admins can approve a connection, so a contributor cannot quietly wire an agent into your docs.",{"type":38,"tag":47,"props":296,"children":297},{},[298,300,305],{"type":44,"value":299},"ChatGPT supports this as a ",{"type":38,"tag":51,"props":301,"children":302},{},[303],{"type":44,"value":304},"developer mode",{"type":44,"value":306}," connector only, because OpenAI limits write-capable connectors to developer mode. ChatGPT's regular connector surface can read your docs but cannot write to them. Coding agents such as Claude Code and Cursor connect differently: they use an API key with Read and write scope rather than OAuth, which fits how they already authenticate.",{"type":38,"tag":172,"props":308,"children":310},{"id":309},"what-the-agent-can-read",[311],{"type":44,"value":312},"What the agent can read",{"type":38,"tag":47,"props":314,"children":315},{},[316,318,324,326,332],{"type":44,"value":317},"Two signal tools give the agent the same view of the loop that you have. ",{"type":38,"tag":287,"props":319,"children":321},{"className":320},[],[322],{"type":44,"value":323},"list_content_gaps",{"type":44,"value":325}," returns what readers searched for and did not find, plus the questions the assistant could not answer, with counts. ",{"type":38,"tag":287,"props":327,"children":329},{"className":328},[],[330],{"type":44,"value":331},"list_stale_pages",{"type":44,"value":333}," returns published pages untouched past a threshold you set. Those two lists are Steps 2 and 3 of the loop, handed to the agent as structured data instead of a dashboard a person has to interpret.",{"type":38,"tag":172,"props":335,"children":337},{"id":336},"what-the-agent-can-write",[338],{"type":44,"value":339},"What the agent can write",{"type":38,"tag":47,"props":341,"children":342},{},[343,345,351,353,359,361,367],{"type":44,"value":344},"Three tools, and their limits are the interesting part. ",{"type":38,"tag":287,"props":346,"children":348},{"className":347},[],[349],{"type":44,"value":350},"create_collection",{"type":44,"value":352}," makes a collection. ",{"type":38,"tag":287,"props":354,"children":356},{"className":355},[],[357],{"type":44,"value":358},"create_article",{"type":44,"value":360}," creates a page as a draft. ",{"type":38,"tag":287,"props":362,"children":364},{"className":363},[],[365],{"type":44,"value":366},"update_article",{"type":44,"value":368}," edits a draft, and the previous version goes to Page History before the change lands. That is the whole write surface.",{"type":38,"tag":47,"props":370,"children":371},{},[372],{"type":44,"value":373},"Agents also receive Dewstack's docs dialect when they connect, so what comes back has callouts, numbered step blocks, tables, and video embeds rather than a wall of prose. A draft that already looks like documentation is faster to review than one you have to reformat before you can even judge it.",{"type":38,"tag":47,"props":375,"children":376},{},[377],{"type":38,"tag":166,"props":378,"children":381},{"alt":379,"src":380},"Two columns. An agent can create a collection, create a draft article, edit a draft, and read gap and stale-page signals. An agent cannot publish, unpublish, edit a live page, or bypass Page History.","\u002Fimages\u002Fblog\u002Ffigures\u002Fselflearn-agent-can-cannot.webp",[],{"type":38,"tag":172,"props":383,"children":385},{"id":384},"the-four-rules",[386],{"type":44,"value":387},"The four rules",{"type":38,"tag":47,"props":389,"children":390},{},[391],{"type":44,"value":392},"These hold with no exceptions, and they are the reason the debunk at the top of this article survives contact with the product.",{"type":38,"tag":47,"props":394,"children":395},{},[396,401],{"type":38,"tag":51,"props":397,"children":398},{},[399],{"type":44,"value":400},"Drafts only.",{"type":44,"value":402}," An agent cannot publish, cannot unpublish, and cannot touch a live page. If you want an agent to rework something already published, a human unpublishes it first. Otherwise the agent is refused.",{"type":38,"tag":47,"props":404,"children":405},{},[406,411],{"type":38,"tag":51,"props":407,"children":408},{},[409],{"type":44,"value":410},"History first.",{"type":44,"value":412}," The previous version is saved to Page History before any change. Page History attributes agent edits to the connection's key name, for example \"Claude connector - agent update\", and one click restores the previous version.",{"type":38,"tag":47,"props":414,"children":415},{},[416,421],{"type":38,"tag":51,"props":417,"children":418},{},[419],{"type":44,"value":420},"Sanitised on the way in.",{"type":44,"value":422}," Script handlers and unsafe embeds are stripped from anything an agent writes. A model that has been talked into emitting something hostile does not get to store it in your docs.",{"type":38,"tag":47,"props":424,"children":425},{},[426,431],{"type":38,"tag":51,"props":427,"children":428},{},[429],{"type":44,"value":430},"Visible and revocable.",{"type":44,"value":432}," The connection appears under API keys. Revoking it disconnects the agent immediately.",{"type":38,"tag":172,"props":434,"children":436},{"id":435},"the-operational-limits",[437],{"type":44,"value":438},"The operational limits",{"type":38,"tag":47,"props":440,"children":441},{},[442],{"type":44,"value":443},"Each connection or key is capped at 500 writes per day. Write access requires the workspace to be Public. Every write, and every refused write, is recorded against the key, so the audit trail includes the attempts that did not succeed. That last detail matters more than it sounds: a log of refusals is how you notice an agent repeatedly trying to edit live pages, which usually means someone's prompt is wrong rather than that your policy is.",{"type":38,"tag":67,"props":445,"children":447},{"id":446},"self-learning-vs-self-updating-two-loops-one-system",[448],{"type":44,"value":449},"Self-Learning vs. Self-Updating: Two Loops, One System",{"type":38,"tag":47,"props":451,"children":452},{},[453],{"type":44,"value":454},"These terms get conflated, but they solve different decay problems:",{"type":38,"tag":456,"props":457,"children":458},"table",{},[459,481],{"type":38,"tag":460,"props":461,"children":462},"thead",{},[463],{"type":38,"tag":464,"props":465,"children":466},"tr",{},[467,471,476],{"type":38,"tag":468,"props":469,"children":470},"th",{},[],{"type":38,"tag":468,"props":472,"children":473},{},[474],{"type":44,"value":475},"Self-learning loop",{"type":38,"tag":468,"props":477,"children":478},{},[479],{"type":44,"value":480},"Self-updating loop",{"type":38,"tag":482,"props":483,"children":484},"tbody",{},[485,517,535,553],{"type":38,"tag":464,"props":486,"children":487},{},[488,494,506],{"type":38,"tag":489,"props":490,"children":491},"td",{},[492],{"type":44,"value":493},"Question it answers",{"type":38,"tag":489,"props":495,"children":496},{},[497,499,504],{"type":44,"value":498},"What content is ",{"type":38,"tag":116,"props":500,"children":501},{},[502],{"type":44,"value":503},"missing",{"type":44,"value":505},"?",{"type":38,"tag":489,"props":507,"children":508},{},[509,511,516],{"type":44,"value":510},"Is existing content still ",{"type":38,"tag":116,"props":512,"children":513},{},[514],{"type":44,"value":515},"true",{"type":44,"value":505},{"type":38,"tag":464,"props":518,"children":519},{},[520,525,530],{"type":38,"tag":489,"props":521,"children":522},{},[523],{"type":44,"value":524},"Primary signals",{"type":38,"tag":489,"props":526,"children":527},{},[528],{"type":44,"value":529},"Unanswered questions, zero-result searches",{"type":38,"tag":489,"props":531,"children":532},{},[533],{"type":44,"value":534},"Source-document changes, release notes, stale-page flags",{"type":38,"tag":464,"props":536,"children":537},{},[538,543,548],{"type":38,"tag":489,"props":539,"children":540},{},[541],{"type":44,"value":542},"Output",{"type":38,"tag":489,"props":544,"children":545},{},[546],{"type":44,"value":547},"New articles and FAQs",{"type":38,"tag":489,"props":549,"children":550},{},[551],{"type":44,"value":552},"Revisions and re-syncs",{"type":38,"tag":464,"props":554,"children":555},{},[556,561,566],{"type":38,"tag":489,"props":557,"children":558},{},[559],{"type":44,"value":560},"Failure if absent",{"type":38,"tag":489,"props":562,"children":563},{},[564],{"type":44,"value":565},"Coverage gaps, repetitive tickets",{"type":38,"tag":489,"props":567,"children":568},{},[569],{"type":44,"value":570},"Confidently wrong answers",{"type":38,"tag":47,"props":572,"children":573},{},[574,576,582,584,590],{"type":44,"value":575},"A knowledge base running only the learning loop grows while its old content rots. One running only the updating loop stays accurate about an ever-smaller slice of what users ask. Run both. The updating side is covered in our guide to ",{"type":38,"tag":195,"props":577,"children":579},{"href":578},"\u002Fblog\u002Fself-updating-knowledge-base",[580],{"type":44,"value":581},"self-updating knowledge bases",{"type":44,"value":583},", and the surrounding tooling in ",{"type":38,"tag":195,"props":585,"children":587},{"href":586},"\u002Fblog\u002Fknowledge-base-automation",[588],{"type":44,"value":589},"knowledge base automation",{"type":44,"value":122},{"type":38,"tag":47,"props":592,"children":593},{},[594,596,601,603,608],{"type":44,"value":595},"Agent access serves both loops, because ",{"type":38,"tag":287,"props":597,"children":599},{"className":598},[],[600],{"type":44,"value":323},{"type":44,"value":602}," feeds the learning side and ",{"type":38,"tag":287,"props":604,"children":606},{"className":605},[],[607],{"type":44,"value":331},{"type":44,"value":609}," feeds the updating side. The output is the same shape either way: a draft waiting for a person.",{"type":38,"tag":67,"props":611,"children":613},{"id":612},"how-to-evaluate-self-learning-claims",[614],{"type":44,"value":615},"How to Evaluate \"Self-Learning\" Claims",{"type":38,"tag":47,"props":617,"children":618},{},[619],{"type":44,"value":620},"When a vendor says their product learns, ask these questions:",{"type":38,"tag":47,"props":622,"children":623},{},[624],{"type":38,"tag":166,"props":625,"children":628},{"alt":626,"src":627},"A table of six questions to ask a vendor, each with a credible answer and the warning sign that should worry you.","\u002Fimages\u002Fblog\u002Ffigures\u002Fselflearn-vendor-questions.webp",[],{"type":38,"tag":630,"props":631,"children":632},"ul",{},[633,644,654,672,682,692],{"type":38,"tag":634,"props":635,"children":636},"li",{},[637,642],{"type":38,"tag":51,"props":638,"children":639},{},[640],{"type":44,"value":641},"\"Show me the unanswered-questions log.\"",{"type":44,"value":643}," If the product cannot show you exactly which questions it failed on, in the user's words, there is no learning loop, just a chatbot.",{"type":38,"tag":634,"props":645,"children":646},{},[647,652],{"type":38,"tag":51,"props":648,"children":649},{},[650],{"type":44,"value":651},"\"What does the system do with a gap?\"",{"type":44,"value":653}," The credible answer involves ranking, drafting, and a review queue. The suspicious answer is \"the AI handles it.\"",{"type":38,"tag":634,"props":655,"children":656},{},[657,662,664,670],{"type":38,"tag":51,"props":658,"children":659},{},[660],{"type":44,"value":661},"\"Does anything publish without human approval?\"",{"type":44,"value":663}," For ",{"type":38,"tag":195,"props":665,"children":667},{"href":666},"\u002Ffeatures\u002Fproduct-documentation",[668],{"type":44,"value":669},"product documentation",{"type":44,"value":671}," that customers rely on, the answer should be no. Ask what enforces that. A policy in a help article is not the same as a write API that refuses.",{"type":38,"tag":634,"props":673,"children":674},{},[675,680],{"type":38,"tag":51,"props":676,"children":677},{},[678],{"type":44,"value":679},"\"If an agent can write, what exactly can it write to?\"",{"type":44,"value":681}," Ask for the tool list and the refusals. A vendor that can name the operations an agent is blocked from has thought about the boundary. A vendor that answers \"it is fully autonomous\" has told you the risk, not the feature.",{"type":38,"tag":634,"props":683,"children":684},{},[685,690],{"type":38,"tag":51,"props":686,"children":687},{},[688],{"type":44,"value":689},"\"Can I see and undo what the agent did?\"",{"type":44,"value":691}," Look for version history that names the agent connection and restores in one step, plus an audit log that includes refused attempts.",{"type":38,"tag":634,"props":693,"children":694},{},[695,700],{"type":38,"tag":51,"props":696,"children":697},{},[698],{"type":44,"value":699},"\"How do I measure improvement?\"",{"type":44,"value":701}," Look for answer rate over time, gap-closure tracking, and per-article quality signals.",{"type":38,"tag":67,"props":703,"children":705},{"id":704},"running-the-loop-on-dewstack",[706],{"type":44,"value":707},"Running the Loop on Dewstack",{"type":38,"tag":47,"props":709,"children":710},{},[711,713,718],{"type":44,"value":712},"Dewstack implements this loop end to end. SmartDocs, the AI assistant, answers user questions from your imported content with citations, and its conversation logs surface every question it could not answer. Search analytics add zero-result queries. Those two streams are what ",{"type":38,"tag":287,"props":714,"children":716},{"className":715},[],[717],{"type":44,"value":323},{"type":44,"value":719}," exposes to a connected agent, and what your team sees in the dashboard.",{"type":38,"tag":47,"props":721,"children":722},{},[723],{"type":44,"value":724},"From there you have two ways to close a gap. A person can write the article. Or you can point a connected agent at the gap list and let it produce drafts, which land in your workspace unpublished, formatted in the docs dialect, and attributed in Page History. Either way, someone with the authority to be right about your product opens the draft, fixes what is wrong, and presses publish.",{"type":38,"tag":47,"props":726,"children":727},{},[728],{"type":44,"value":729},"Because content imports and syncs from Notion, Google Docs, Confluence, and PDF, the loop runs on your real, current material rather than a stale copy.",{"type":38,"tag":47,"props":731,"children":732},{},[733,735,741],{"type":44,"value":734},"Start small. Enable the assistant, let a week of real questions accumulate, and fill the top three gaps. Then connect one agent, in one workspace, with one reviewer who owns the queue. Watch the answer rate and the refusal log together for a month before you widen access. For the broader architecture this loop lives inside, covering ingestion, embeddings, retrieval, and grounding, see the ",{"type":38,"tag":195,"props":736,"children":738},{"href":737},"\u002Fblog\u002Fai-knowledge-base-a-detailed-guide",[739],{"type":44,"value":740},"complete guide to AI knowledge bases",{"type":44,"value":122},{"type":38,"tag":47,"props":743,"children":744},{},[745],{"type":44,"value":746},"The honest summary is the same one this article opened with, just with a sharper edge. Nothing here learns your business on its own. What it does is watch what your readers fail to find, hand that list to something that can write a first pass, and then stop and wait for you.",{"title":7,"searchDepth":748,"depth":748,"links":749},2,[750,751,760,767,768,769],{"id":69,"depth":748,"text":72},{"id":153,"depth":748,"text":156,"children":752},[753,755,756,757,758,759],{"id":174,"depth":754,"text":177},3,{"id":185,"depth":754,"text":188},{"id":205,"depth":754,"text":208},{"id":216,"depth":754,"text":219},{"id":232,"depth":754,"text":235},{"id":248,"depth":754,"text":251},{"id":266,"depth":748,"text":269,"children":761},[762,763,764,765,766],{"id":277,"depth":754,"text":280},{"id":309,"depth":754,"text":312},{"id":336,"depth":754,"text":339},{"id":384,"depth":754,"text":387},{"id":435,"depth":754,"text":438},{"id":446,"depth":748,"text":449},{"id":612,"depth":748,"text":615},{"id":704,"depth":748,"text":707},"markdown","content:blog:self-learning-knowledge-base.md","content","blog\u002Fself-learning-knowledge-base.md","blog\u002Fself-learning-knowledge-base","md",{"loc":4,"lastmod":14},[778,782,789,794,799,803,804,808,814,821,828,834,840,846,852,858,865,871,878,885,893,900,906,913,919,926],{"_path":197,"title":779,"description":780,"image":781,"readingTime":12,"createdAt":13,"updatedAt":14},"AI Answers From Your Docs: Grounded Answers, Citations & Hallucination Control","How an AI chatbot for technical documentation gives grounded answers with source citations, controls hallucinations, and turns unanswered questions into drafted pages.","\u002Fimages\u002Fblog\u002Fheaders\u002Fai-answers-grounded.png",{"_path":783,"title":784,"description":785,"image":786,"readingTime":787,"createdAt":788,"updatedAt":14},"\u002Fblog\u002Fai-improve-existing-documentation","How to Use AI to Improve Existing Documentation","Use AI to audit, rewrite, and translate the documentation you already have, and let a connected agent draft improvements to stale pages for a human to publish.","\u002Fimages\u002Fblog\u002Fheaders\u002Fimprove-existing-docs.png","19 min read","2026-06-05",{"_path":790,"title":791,"description":792,"image":793,"readingTime":787,"createdAt":788,"updatedAt":14},"\u002Fblog\u002Fautomate-sop-creation-ai","How to Automate SOP Creation with AI","Capture a workflow, let AI draft the SOP, review it, and publish it to a searchable knowledge base your team can query around the clock.","\u002Fimages\u002Fblog\u002Fheaders\u002Fautomate-sop-ai.png",{"_path":795,"title":796,"description":797,"image":798,"readingTime":787,"createdAt":788,"updatedAt":14},"\u002Fblog\u002Fbest-documentation-platform-saas-companies","Best Documentation Platform for SaaS Companies in 2026","Compare documentation platforms for SaaS teams. See how Dewstack's agent write path, SmartDocs AI, and 20+ imports cut support tickets. Start free today.","\u002Fimages\u002Fblog\u002Fheaders\u002Fbest-docs-platform-saas.png",{"_path":586,"title":800,"description":801,"image":802,"readingTime":12,"createdAt":13,"updatedAt":14},"Knowledge Base Automation: What to Automate (and What Not To)","A practical guide to knowledge base automation: content sync from Notion and Confluence, AI answering, gap detection, agent-written drafts, and the analytics loop.","\u002Fimages\u002Fblog\u002Fheaders\u002Fkb-automation.png",{"_path":4,"title":8,"description":9,"image":10,"readingTime":12,"createdAt":13,"updatedAt":14},{"_path":578,"title":805,"description":806,"image":807,"readingTime":12,"createdAt":13,"updatedAt":14},"Self-Updating Knowledge Base: How to Keep Documentation Up to Date","Why knowledge bases go stale, what stale docs really cost, and how automated update signals plus connected agents plus human review keep documentation accurate over time.","\u002Fimages\u002Fblog\u002Fheaders\u002Fself-updating-kb.png",{"_path":737,"title":809,"description":810,"image":811,"readingTime":812,"createdAt":813,"updatedAt":13},"AI Knowledge Base: What It Is & How to Build One (2026)","What an AI knowledge base is, how it works, and how to build one step by step. Includes examples, ROI benchmarks and a free knowledge base template.","\u002Fimages\u002Fblog\u002Fheaders\u002Fai-knowledge-base.png","15 min read","2024-12-01",{"_path":815,"title":816,"description":817,"image":818,"readingTime":819,"createdAt":820,"updatedAt":13},"\u002Fblog\u002Fcomprehensive-guide-to-effective-sops-in-manufacturing","Manufacturing SOPs: Guide, Examples & Template (2026)","Create manufacturing SOPs for quality, safety and maintenance. Practical guide with examples and a free SOP template. Boost efficiency by up to 25%.","\u002Fimages\u002Fblog\u002Fheaders\u002Fmanufacturing-sops.png","12 min read","2024-11-24",{"_path":822,"title":823,"description":824,"image":825,"readingTime":826,"createdAt":827,"updatedAt":13},"\u002Fblog\u002Feffective-sops-in-finance-a-comprehensive-guide","Finance SOPs: Guide, Examples & Free Template (2026)","Write audit-ready finance SOPs with real examples and a free template. Covers SOX, GAAP and reconciliation procedures that cut errors by up to 67%.","\u002Fimages\u002Fblog\u002Fheaders\u002Ffinance-sops.png","14 min read","2024-11-17",{"_path":829,"title":830,"description":831,"image":832,"readingTime":826,"createdAt":833,"updatedAt":13},"\u002Fblog\u002Fguide-to-effective-sops-in-healthcare","Healthcare SOPs: Guide, Examples & Free Template (2026)","Create healthcare SOPs that meet HIPAA, OSHA and Joint Commission standards. Step-by-step guide with real examples and a free SOP template to download.","\u002Fimages\u002Fblog\u002Fheaders\u002Fhealthcare-sops.png","2024-11-03",{"_path":835,"title":836,"description":837,"image":838,"readingTime":826,"createdAt":839,"updatedAt":13},"\u002Fblog\u002Fguide-to-effective-sops-in-technology","IT & Technology SOPs: Guide, Examples & Template (2026)","Build SOPs for DevOps, incident response and IT operations. Real examples plus a free SOP template and incident runbook you can download and adapt.","\u002Fimages\u002Fblog\u002Fheaders\u002Ftech-sops.png","2024-10-27",{"_path":841,"title":842,"description":843,"image":844,"readingTime":819,"createdAt":845,"updatedAt":13},"\u002Fblog\u002Fmastering-user-manual-step-by-step-guide","How to Write a User Manual: Steps + Free Template (2026)","Write a user manual people actually read: a step-by-step guide with examples, checklists and a free Word template. Cut support tickets by up to 50%.","\u002Fimages\u002Fblog\u002Fheaders\u002Fuser-manual-guide.png","2024-10-06",{"_path":847,"title":848,"description":849,"image":850,"createdAt":851,"updatedAt":13,"readingTime":826},"\u002Fblog\u002Fthe-importance-of-sops-in-project-management-a-comprehensive-guide","Project Management SOPs: Guide & Free Templates (2026)","Run smoother projects with SOPs. Practical guide with examples plus free templates for kickoff meetings, risk assessment and status reporting.","\u002Fimages\u002Fblog\u002Fheaders\u002Fpm-sops.png","2024-09-01",{"_path":853,"title":854,"description":855,"image":856,"readingTime":819,"createdAt":857,"updatedAt":13},"\u002Fblog\u002Fthe-ultimate-troubleshooting-guide-tips-and-tricks-for-success","Troubleshooting Guide: Steps, Examples & Playbook (2026)","A proven 6-step troubleshooting framework with real IT, manufacturing and support examples, plus a free playbook with decision trees and 5 Whys.","\u002Fimages\u002Fblog\u002Fheaders\u002Ftroubleshooting-guide.png","2024-08-11",{"_path":859,"title":860,"description":861,"image":862,"readingTime":826,"createdAt":863,"updatedAt":864},"\u002Fblog\u002Fa-comprehensive-guide-to-creating-and-managing-effective-documentation-for-your-knowledge-base","A Comprehensive Guide to Creating and Managing Effective Documentation for Your Knowledge Base","Learn how to create and manage knowledge base documentation that reduces support costs by up to 50%. Discover proven strategies, real-world examples, and expert tips for building documentation that users actually use.","\u002Fimages\u002Fblog\u002Fheaders\u002Fknowledge-base-guide.png","2024-12-08","2026-01-20",{"_path":866,"title":867,"description":868,"image":869,"readingTime":12,"createdAt":870,"updatedAt":864},"\u002Fblog\u002Frevolutionize-your-documentation-with-the-ultimate-import-tool","Revolutionize Your Documentation with the Ultimate Import Tool","Learn how to overcome documentation fragmentation, migrate content effectively, and build a unified knowledge base. Complete guide with migration strategies, platform comparisons, and best practices.","\u002Fimages\u002Fblog\u002Fheaders\u002Fmigration-guide.png","2024-09-29",{"_path":872,"title":873,"description":874,"image":875,"readingTime":826,"createdAt":876,"updatedAt":877},"\u002Fblog\u002Fthe-ultimate-guide-to-documentation-how-to-create-user-friendly-and-effective-documentation","The Ultimate Guide to Documentation: How to Create User-Friendly and Effective User Documentation","Learn how to create user-friendly documentation that boosts customer satisfaction by 47% and reduces support costs. Complete guide with templates, frameworks, accessibility best practices, and proven strategies for effective technical documentation.","\u002Fimages\u002Fblog\u002Fheaders\u002Fultimate-docs-guide.png","2024-08-18","2026-01-03",{"_path":879,"title":880,"description":881,"image":882,"readingTime":819,"createdAt":883,"updatedAt":884},"\u002Fblog\u002F10-reasons-why-your-documentation-is-killing-your-user-experience","10 Reasons Why Your Documentation is Killing Your User Experience","Discover the 10 critical documentation mistakes that drive users away, backed by research from Forrester, Nielsen Norman Group, and more. Learn actionable fixes for each issue.","\u002Fimages\u002Fblog\u002Fheaders\u002Fdocumentation-mistakes.png","2024-12-15","2025-12-31",{"_path":886,"title":887,"description":888,"image":889,"readingTime":890,"createdAt":891,"updatedAt":892},"\u002Fblog\u002Fthe-power-of-knowledge-why-user-guides-and-documentation-are-essential-for-your-business","The Power of Knowledge: Why User Documentation and Guides are Essential for Your Business","Discover how user guides, documentation, and knowledge bases drive business growth by reducing support costs by 30-50%, improving customer retention, and meeting the 67% of customers who prefer self-service solutions.","\u002Fimages\u002Fblog\u002Fheaders\u002Fknowledge-power.png","10 min read","2024-08-25","2025-12-30",{"_path":894,"title":895,"description":896,"image":897,"readingTime":826,"createdAt":898,"updatedAt":899},"\u002Fblog\u002Fguide-to-creating-and-managing-a-sop","The Ultimate Guide to Creating and Managing a SOP (Standard Operating Procedure)","Learn how to create effective SOPs that boost operational efficiency by up to 30%. This comprehensive SOP guide covers standard operating procedure templates, best practices for SOP management, implementation strategies, and free downloadable SOP templates.","\u002Fimages\u002Fblog\u002Fheaders\u002Fsop-creation-guide.png","2024-11-10","2025-12-28",{"_path":901,"title":902,"description":903,"image":904,"createdAt":905,"updatedAt":899,"readingTime":819},"\u002Fblog\u002Fsubtle-art-of-content-creation-with-ai-based-content-creation-tool","The Subtle Art of AI-Powered Content Creation: A Complete Guide to Transforming Your Documentation Workflow","Discover how AI content creation tools are revolutionizing documentation workflows. Learn best practices for human-AI collaboration, productivity gains, and practical strategies to create high-quality content faster with AI writing assistants.","\u002Fimages\u002Fblog\u002Fheaders\u002Fai-content-creation.png","2024-09-22",{"_path":907,"title":908,"description":909,"image":910,"readingTime":819,"createdAt":911,"updatedAt":912},"\u002Fblog\u002Fthe-benefits-of-online-documentation-and-how-to-get-started","The Benefits of Online Documentation and How to Get Started","Discover how online documentation reduces support costs by 45%, improves customer satisfaction, and drives digital transformation. Complete guide with ROI calculations, migration strategies, and implementation best practices for 2025.","\u002Fimages\u002Fblog\u002Fheaders\u002Fonline-docs-benefits.png","2024-09-15","2025-12-27",{"_path":914,"title":915,"description":916,"image":917,"readingTime":12,"createdAt":918,"updatedAt":912},"\u002Fblog\u002Fthe-importance-of-documentation","The Importance of Documentation: Why It's Your Business's Most Undervalued Asset","Discover why documentation is critical to business success. Learn how proper documentation reduces costs, preserves knowledge, improves efficiency, and drives growth with data-backed insights and actionable strategies.","\u002Fimages\u002Fblog\u002Fheaders\u002Fdocumentation-roi.png","2024-09-08",{"_path":920,"title":921,"description":922,"image":923,"readingTime":890,"createdAt":924,"updatedAt":925},"\u002Fblog\u002Fhow-neglecting-user-guides-can-sink-your-product","How Neglecting User Guides Can Sink Your Product","Discover why neglecting user guides can sink your product. Learn how poor documentation leads to user frustration, churn, and bad reviews - and how to fix it.","\u002Fimages\u002Fblog\u002Fheaders\u002Fuser-guide-neglect.png","2024-10-20","2025-12-26",{"_path":927,"title":928,"description":929,"image":930,"readingTime":819,"createdAt":931,"updatedAt":932},"\u002Fblog\u002Fhow-to-write-faqs-effectively","How to Write FAQs That Actually Help Customers: A Complete Guide","Master the art of writing effective FAQs that reduce support tickets by up to 70%, improve customer satisfaction, and boost SEO rankings. Learn proven strategies, see before\u002Fafter examples, and discover how to implement FAQ schema markup.","\u002Fimages\u002Fblog\u002Fheaders\u002Ffaq-writing.png","2024-10-13","2025-12-25",{"_path":4,"_dir":5,"_draft":6,"_partial":6,"_locale":7,"title":8,"description":9,"image":10,"type":5,"author":11,"readingTime":12,"createdAt":13,"updatedAt":14,"faqs":934,"body":941,"_type":770,"_id":771,"_source":772,"_file":773,"_stem":774,"_extension":775,"sitemap":1498},[935,936,937,938,939,940],{"question":17,"answer":18},{"question":20,"answer":21},{"question":23,"answer":24},{"question":26,"answer":27},{"question":29,"answer":30},{"question":32,"answer":33},{"type":35,"children":942,"toc":1477},[943,947,954,958,962,966,970,974,983,991,1004,1012,1026,1030,1034,1040,1044,1048,1052,1061,1065,1069,1073,1077,1081,1085,1089,1093,1097,1106,1110,1114,1118,1128,1137,1141,1157,1161,1183,1187,1193,1197,1201,1209,1217,1225,1233,1237,1241,1245,1249,1342,1356,1372,1376,1380,1386,1442,1446,1456,1460,1464,1473],{"type":38,"tag":39,"props":944,"children":945},{"id":41},[946],{"type":44,"value":45},{"type":38,"tag":47,"props":948,"children":949},{},[950],{"type":38,"tag":51,"props":951,"children":952},{},[953],{"type":44,"value":55},{"type":38,"tag":47,"props":955,"children":956},{},[957],{"type":44,"value":60},{"type":38,"tag":47,"props":959,"children":960},{},[961],{"type":44,"value":65},{"type":38,"tag":67,"props":963,"children":964},{"id":69},[965],{"type":44,"value":72},{"type":38,"tag":47,"props":967,"children":968},{},[969],{"type":44,"value":77},{"type":38,"tag":47,"props":971,"children":972},{},[973],{"type":44,"value":82},{"type":38,"tag":47,"props":975,"children":976},{},[977,978,982],{"type":44,"value":87},{"type":38,"tag":51,"props":979,"children":980},{},[981],{"type":44,"value":92},{"type":44,"value":94},{"type":38,"tag":47,"props":984,"children":985},{},[986,990],{"type":38,"tag":51,"props":987,"children":988},{},[989],{"type":44,"value":102},{"type":44,"value":104},{"type":38,"tag":47,"props":992,"children":993},{},[994,998,999,1003],{"type":38,"tag":51,"props":995,"children":996},{},[997],{"type":44,"value":112},{"type":44,"value":114},{"type":38,"tag":116,"props":1000,"children":1001},{},[1002],{"type":44,"value":120},{"type":44,"value":122},{"type":38,"tag":47,"props":1005,"children":1006},{},[1007,1011],{"type":38,"tag":51,"props":1008,"children":1009},{},[1010],{"type":44,"value":130},{"type":44,"value":132},{"type":38,"tag":47,"props":1013,"children":1014},{},[1015,1016,1020,1021,1025],{"type":44,"value":137},{"type":38,"tag":116,"props":1017,"children":1018},{},[1019],{"type":44,"value":142},{"type":44,"value":144},{"type":38,"tag":116,"props":1022,"children":1023},{},[1024],{"type":44,"value":149},{"type":44,"value":122},{"type":38,"tag":67,"props":1027,"children":1028},{"id":153},[1029],{"type":44,"value":156},{"type":38,"tag":47,"props":1031,"children":1032},{},[1033],{"type":44,"value":161},{"type":38,"tag":47,"props":1035,"children":1036},{},[1037],{"type":38,"tag":166,"props":1038,"children":1039},{"alt":168,"src":169},[],{"type":38,"tag":172,"props":1041,"children":1042},{"id":174},[1043],{"type":44,"value":177},{"type":38,"tag":47,"props":1045,"children":1046},{},[1047],{"type":44,"value":182},{"type":38,"tag":172,"props":1049,"children":1050},{"id":185},[1051],{"type":44,"value":188},{"type":38,"tag":47,"props":1053,"children":1054},{},[1055,1056,1060],{"type":44,"value":193},{"type":38,"tag":195,"props":1057,"children":1058},{"href":197},[1059],{"type":44,"value":200},{"type":44,"value":202},{"type":38,"tag":172,"props":1062,"children":1063},{"id":205},[1064],{"type":44,"value":208},{"type":38,"tag":47,"props":1066,"children":1067},{},[1068],{"type":44,"value":213},{"type":38,"tag":172,"props":1070,"children":1071},{"id":216},[1072],{"type":44,"value":219},{"type":38,"tag":47,"props":1074,"children":1075},{},[1076],{"type":44,"value":224},{"type":38,"tag":47,"props":1078,"children":1079},{},[1080],{"type":44,"value":229},{"type":38,"tag":172,"props":1082,"children":1083},{"id":232},[1084],{"type":44,"value":235},{"type":38,"tag":47,"props":1086,"children":1087},{},[1088],{"type":44,"value":240},{"type":38,"tag":47,"props":1090,"children":1091},{},[1092],{"type":44,"value":245},{"type":38,"tag":172,"props":1094,"children":1095},{"id":248},[1096],{"type":44,"value":251},{"type":38,"tag":47,"props":1098,"children":1099},{},[1100,1101,1105],{"type":44,"value":256},{"type":38,"tag":116,"props":1102,"children":1103},{},[1104],{"type":44,"value":261},{"type":44,"value":263},{"type":38,"tag":67,"props":1107,"children":1108},{"id":266},[1109],{"type":44,"value":269},{"type":38,"tag":47,"props":1111,"children":1112},{},[1113],{"type":44,"value":274},{"type":38,"tag":172,"props":1115,"children":1116},{"id":277},[1117],{"type":44,"value":280},{"type":38,"tag":47,"props":1119,"children":1120},{},[1121,1122,1127],{"type":44,"value":285},{"type":38,"tag":287,"props":1123,"children":1125},{"className":1124},[],[1126],{"type":44,"value":292},{"type":44,"value":294},{"type":38,"tag":47,"props":1129,"children":1130},{},[1131,1132,1136],{"type":44,"value":299},{"type":38,"tag":51,"props":1133,"children":1134},{},[1135],{"type":44,"value":304},{"type":44,"value":306},{"type":38,"tag":172,"props":1138,"children":1139},{"id":309},[1140],{"type":44,"value":312},{"type":38,"tag":47,"props":1142,"children":1143},{},[1144,1145,1150,1151,1156],{"type":44,"value":317},{"type":38,"tag":287,"props":1146,"children":1148},{"className":1147},[],[1149],{"type":44,"value":323},{"type":44,"value":325},{"type":38,"tag":287,"props":1152,"children":1154},{"className":1153},[],[1155],{"type":44,"value":331},{"type":44,"value":333},{"type":38,"tag":172,"props":1158,"children":1159},{"id":336},[1160],{"type":44,"value":339},{"type":38,"tag":47,"props":1162,"children":1163},{},[1164,1165,1170,1171,1176,1177,1182],{"type":44,"value":344},{"type":38,"tag":287,"props":1166,"children":1168},{"className":1167},[],[1169],{"type":44,"value":350},{"type":44,"value":352},{"type":38,"tag":287,"props":1172,"children":1174},{"className":1173},[],[1175],{"type":44,"value":358},{"type":44,"value":360},{"type":38,"tag":287,"props":1178,"children":1180},{"className":1179},[],[1181],{"type":44,"value":366},{"type":44,"value":368},{"type":38,"tag":47,"props":1184,"children":1185},{},[1186],{"type":44,"value":373},{"type":38,"tag":47,"props":1188,"children":1189},{},[1190],{"type":38,"tag":166,"props":1191,"children":1192},{"alt":379,"src":380},[],{"type":38,"tag":172,"props":1194,"children":1195},{"id":384},[1196],{"type":44,"value":387},{"type":38,"tag":47,"props":1198,"children":1199},{},[1200],{"type":44,"value":392},{"type":38,"tag":47,"props":1202,"children":1203},{},[1204,1208],{"type":38,"tag":51,"props":1205,"children":1206},{},[1207],{"type":44,"value":400},{"type":44,"value":402},{"type":38,"tag":47,"props":1210,"children":1211},{},[1212,1216],{"type":38,"tag":51,"props":1213,"children":1214},{},[1215],{"type":44,"value":410},{"type":44,"value":412},{"type":38,"tag":47,"props":1218,"children":1219},{},[1220,1224],{"type":38,"tag":51,"props":1221,"children":1222},{},[1223],{"type":44,"value":420},{"type":44,"value":422},{"type":38,"tag":47,"props":1226,"children":1227},{},[1228,1232],{"type":38,"tag":51,"props":1229,"children":1230},{},[1231],{"type":44,"value":430},{"type":44,"value":432},{"type":38,"tag":172,"props":1234,"children":1235},{"id":435},[1236],{"type":44,"value":438},{"type":38,"tag":47,"props":1238,"children":1239},{},[1240],{"type":44,"value":443},{"type":38,"tag":67,"props":1242,"children":1243},{"id":446},[1244],{"type":44,"value":449},{"type":38,"tag":47,"props":1246,"children":1247},{},[1248],{"type":44,"value":454},{"type":38,"tag":456,"props":1250,"children":1251},{},[1252,1269],{"type":38,"tag":460,"props":1253,"children":1254},{},[1255],{"type":38,"tag":464,"props":1256,"children":1257},{},[1258,1261,1265],{"type":38,"tag":468,"props":1259,"children":1260},{},[],{"type":38,"tag":468,"props":1262,"children":1263},{},[1264],{"type":44,"value":475},{"type":38,"tag":468,"props":1266,"children":1267},{},[1268],{"type":44,"value":480},{"type":38,"tag":482,"props":1270,"children":1271},{},[1272,1297,1312,1327],{"type":38,"tag":464,"props":1273,"children":1274},{},[1275,1279,1288],{"type":38,"tag":489,"props":1276,"children":1277},{},[1278],{"type":44,"value":493},{"type":38,"tag":489,"props":1280,"children":1281},{},[1282,1283,1287],{"type":44,"value":498},{"type":38,"tag":116,"props":1284,"children":1285},{},[1286],{"type":44,"value":503},{"type":44,"value":505},{"type":38,"tag":489,"props":1289,"children":1290},{},[1291,1292,1296],{"type":44,"value":510},{"type":38,"tag":116,"props":1293,"children":1294},{},[1295],{"type":44,"value":515},{"type":44,"value":505},{"type":38,"tag":464,"props":1298,"children":1299},{},[1300,1304,1308],{"type":38,"tag":489,"props":1301,"children":1302},{},[1303],{"type":44,"value":524},{"type":38,"tag":489,"props":1305,"children":1306},{},[1307],{"type":44,"value":529},{"type":38,"tag":489,"props":1309,"children":1310},{},[1311],{"type":44,"value":534},{"type":38,"tag":464,"props":1313,"children":1314},{},[1315,1319,1323],{"type":38,"tag":489,"props":1316,"children":1317},{},[1318],{"type":44,"value":542},{"type":38,"tag":489,"props":1320,"children":1321},{},[1322],{"type":44,"value":547},{"type":38,"tag":489,"props":1324,"children":1325},{},[1326],{"type":44,"value":552},{"type":38,"tag":464,"props":1328,"children":1329},{},[1330,1334,1338],{"type":38,"tag":489,"props":1331,"children":1332},{},[1333],{"type":44,"value":560},{"type":38,"tag":489,"props":1335,"children":1336},{},[1337],{"type":44,"value":565},{"type":38,"tag":489,"props":1339,"children":1340},{},[1341],{"type":44,"value":570},{"type":38,"tag":47,"props":1343,"children":1344},{},[1345,1346,1350,1351,1355],{"type":44,"value":575},{"type":38,"tag":195,"props":1347,"children":1348},{"href":578},[1349],{"type":44,"value":581},{"type":44,"value":583},{"type":38,"tag":195,"props":1352,"children":1353},{"href":586},[1354],{"type":44,"value":589},{"type":44,"value":122},{"type":38,"tag":47,"props":1357,"children":1358},{},[1359,1360,1365,1366,1371],{"type":44,"value":595},{"type":38,"tag":287,"props":1361,"children":1363},{"className":1362},[],[1364],{"type":44,"value":323},{"type":44,"value":602},{"type":38,"tag":287,"props":1367,"children":1369},{"className":1368},[],[1370],{"type":44,"value":331},{"type":44,"value":609},{"type":38,"tag":67,"props":1373,"children":1374},{"id":612},[1375],{"type":44,"value":615},{"type":38,"tag":47,"props":1377,"children":1378},{},[1379],{"type":44,"value":620},{"type":38,"tag":47,"props":1381,"children":1382},{},[1383],{"type":38,"tag":166,"props":1384,"children":1385},{"alt":626,"src":627},[],{"type":38,"tag":630,"props":1387,"children":1388},{},[1389,1397,1405,1418,1426,1434],{"type":38,"tag":634,"props":1390,"children":1391},{},[1392,1396],{"type":38,"tag":51,"props":1393,"children":1394},{},[1395],{"type":44,"value":641},{"type":44,"value":643},{"type":38,"tag":634,"props":1398,"children":1399},{},[1400,1404],{"type":38,"tag":51,"props":1401,"children":1402},{},[1403],{"type":44,"value":651},{"type":44,"value":653},{"type":38,"tag":634,"props":1406,"children":1407},{},[1408,1412,1413,1417],{"type":38,"tag":51,"props":1409,"children":1410},{},[1411],{"type":44,"value":661},{"type":44,"value":663},{"type":38,"tag":195,"props":1414,"children":1415},{"href":666},[1416],{"type":44,"value":669},{"type":44,"value":671},{"type":38,"tag":634,"props":1419,"children":1420},{},[1421,1425],{"type":38,"tag":51,"props":1422,"children":1423},{},[1424],{"type":44,"value":679},{"type":44,"value":681},{"type":38,"tag":634,"props":1427,"children":1428},{},[1429,1433],{"type":38,"tag":51,"props":1430,"children":1431},{},[1432],{"type":44,"value":689},{"type":44,"value":691},{"type":38,"tag":634,"props":1435,"children":1436},{},[1437,1441],{"type":38,"tag":51,"props":1438,"children":1439},{},[1440],{"type":44,"value":699},{"type":44,"value":701},{"type":38,"tag":67,"props":1443,"children":1444},{"id":704},[1445],{"type":44,"value":707},{"type":38,"tag":47,"props":1447,"children":1448},{},[1449,1450,1455],{"type":44,"value":712},{"type":38,"tag":287,"props":1451,"children":1453},{"className":1452},[],[1454],{"type":44,"value":323},{"type":44,"value":719},{"type":38,"tag":47,"props":1457,"children":1458},{},[1459],{"type":44,"value":724},{"type":38,"tag":47,"props":1461,"children":1462},{},[1463],{"type":44,"value":729},{"type":38,"tag":47,"props":1465,"children":1466},{},[1467,1468,1472],{"type":44,"value":734},{"type":38,"tag":195,"props":1469,"children":1470},{"href":737},[1471],{"type":44,"value":740},{"type":44,"value":122},{"type":38,"tag":47,"props":1474,"children":1475},{},[1476],{"type":44,"value":746},{"title":7,"searchDepth":748,"depth":748,"links":1478},[1479,1480,1488,1495,1496,1497],{"id":69,"depth":748,"text":72},{"id":153,"depth":748,"text":156,"children":1481},[1482,1483,1484,1485,1486,1487],{"id":174,"depth":754,"text":177},{"id":185,"depth":754,"text":188},{"id":205,"depth":754,"text":208},{"id":216,"depth":754,"text":219},{"id":232,"depth":754,"text":235},{"id":248,"depth":754,"text":251},{"id":266,"depth":748,"text":269,"children":1489},[1490,1491,1492,1493,1494],{"id":277,"depth":754,"text":280},{"id":309,"depth":754,"text":312},{"id":336,"depth":754,"text":339},{"id":384,"depth":754,"text":387},{"id":435,"depth":754,"text":438},{"id":446,"depth":748,"text":449},{"id":612,"depth":748,"text":615},{"id":704,"depth":748,"text":707},{"loc":4,"lastmod":14},1786373779699]