[{"data":1,"prerenderedAt":1648},["ShallowReactive",2],{"site-nav-content":3,"blog:/blog/what-is-enterprise-rag":163,"blog-index-copy":687,"blog:/blog/what-is-enterprise-rag:surround":708,"hiring-banner-content":1617,"site-cta-content":1629},{"header":4,"productNav":9,"nav":42,"footer":61,"askAI":116,"id":147,"title":148,"archived":149,"authors":150,"badge":150,"body":151,"date":150,"department":150,"description":155,"extension":158,"eyebrow":150,"faqHeader":150,"faqs":150,"footerBand":150,"headline":150,"image":150,"industry":150,"jobType":150,"listed":149,"location":150,"navigation":115,"openRoles":150,"pageLayout":150,"path":159,"relatedHeading":150,"seo":160,"series":150,"sitemap":115,"status":150,"stem":161,"subhead":150,"tags":150,"video":150,"whyJoin":150,"workplaceType":150,"__hash__":162},{"productLabel":5,"loginLabel":6,"contactLabel":7,"contactSalesLabel":8},"Product","Log in","Contact","Get started for free",[10,14,18,22,26,30,34,38],{"label":11,"to":12,"description":13},"Overview","/overview","Seven layers. One closed loop.",{"label":15,"to":16,"description":17},"Conflux","/product/conflux","Where your team, workstreams, and agents meet.",{"label":19,"to":20,"description":21},"Agent Teams","/product/agent-teams","Specialist teams - governed from day one.",{"label":23,"to":24,"description":25},"Lifecycle Graph","/product/lifecycle-graph","Intelligence that compounds across every interaction.",{"label":27,"to":28,"description":29},"Company Wiki","/product/wiki","Playbooks and policies where expertise stays.",{"label":31,"to":32,"description":33},"Workstreams","/product/workstreams","From brief to signed-off deliverable on one canvas.",{"label":35,"to":36,"description":37},"Perception Console","/product/perception","Ask your whole business in plain English.",{"label":39,"to":40,"description":41},"Governance","/product/governance","Frontier AI you can actually sign off on.",[43,46,49,52,55,58],{"label":44,"to":45},"Models","/models",{"label":47,"to":48},"Pricing","/pricing",{"label":50,"to":51},"Integrations","/integrations",{"label":53,"to":54},"Security","/security",{"label":56,"to":57},"Partners","/partners",{"label":59,"to":60},"Insights","/blog",{"productHeading":5,"companyHeading":62,"legalHeading":63,"docsLabel":64,"docsUrl":65,"statementLines":66,"copyright":69,"companyLinks":70,"legalLinks":85,"socialLinks":95,"bottomLinks":105},"Company","Legal","Docs","https://docs.gonimbus.ai",[67,68],"Stop training someone else's model.","Control your AI.","© 2026 Nimbus Intelligence, Inc. All rights reserved.",[71,72,73,74,75,77,80,83],{"label":47,"to":48},{"label":50,"to":51},{"label":53,"to":54},{"label":59,"to":60},{"label":76,"to":57},"Partner Program",{"label":78,"to":79},"Careers","/careers",{"label":81,"to":82},"System status","/status",{"label":7,"to":84},"/contact",[86,89,92],{"label":87,"to":88},"Terms of Service","/terms",{"label":90,"to":91},"Privacy Policy","/privacy",{"label":93,"to":94},"Compliance","/compliance",[96,99,102],{"label":97,"href":98},"LinkedIn","https://www.linkedin.com/company/gonimbusai/",{"label":100,"href":101},"X","https://x.com/gonimbusai",{"label":103,"href":104},"Instagram","https://www.instagram.com/gonimbus_ai/",[106,108,110,111,112],{"label":107,"to":88},"Terms",{"label":109,"to":91},"Privacy",{"label":93,"to":94},{"label":81,"to":82},{"label":113,"to":114,"external":115},"LLMs.txt","/llms.txt",true,{"text":117,"prompt":118},"Ask AI about Nimbus",{"I'm researching enterprise intelligence platforms and want to know how Nimbus combines perception, collaboration, and autonomous agents to drive strategic decision-making":119,"platforms":121},{" Summarize the highlights from Nimbus's website":120},"https://gonimbus.ai",[122,127,132,137,142],{"name":123,"label":124,"icon":125,"hrefPrefix":126},"chatgpt","ChatGPT","simple-icons:openai","https://chatgpt.com/?prompt=",{"name":128,"label":129,"icon":130,"hrefPrefix":131},"perplexity","Perplexity","mdi:magnify","https://www.perplexity.ai/search/new?q=",{"name":133,"label":134,"icon":135,"hrefPrefix":136},"grok","Grok","simple-icons:x","https://x.com/i/grok?text=",{"name":138,"label":139,"icon":140,"hrefPrefix":141},"claude","Claude","simple-icons:anthropic","https://claude.ai/new?q=",{"name":143,"label":144,"icon":145,"hrefPrefix":146},"google-ai","Google AI","simple-icons:google","https://www.google.com/search?udm=50&aep=11&q=","content/shared/nav.md","Site navigation",false,null,{"type":152,"value":153,"toc":154},"minimark",[],{"title":155,"searchDepth":156,"depth":156,"links":157},"",2,[],"md","/shared/nav",{"title":148,"description":155},"shared/nav","p6IsjEfjcrwkjwvGsPcuEkpoJSeUn2vJXPTxLWQfw9M",{"id":164,"title":165,"archived":149,"authors":166,"badge":169,"body":171,"date":676,"department":150,"description":677,"extension":158,"eyebrow":150,"faqHeader":150,"faqs":150,"footerBand":150,"headline":150,"image":150,"industry":150,"jobType":150,"listed":149,"location":150,"navigation":115,"openRoles":150,"pageLayout":150,"path":678,"relatedHeading":150,"seo":679,"series":680,"sitemap":115,"status":150,"stem":681,"subhead":150,"tags":682,"video":150,"whyJoin":150,"workplaceType":150,"__hash__":686},"content/blog/what-is-enterprise-rag.md","What is Enterprise RAG",[167],{"name":168,"to":120},"Nimbus Research",{"label":170},"Explainer",{"type":152,"value":172,"toc":651},[173,186,197,208,217,220,225,294,301,305,308,311,336,343,348,354,360,366,372,383,387,393,403,409,415,421,427,435,442,446,458,465,477,481,485,488,492,495,499,510,514,517,521,526,530,538,542,545,549,557,561,574,578,581,585,594,598,606,610,618,622],[174,175,176,177,181,182,185],"p",{},"RAG stands for ",[178,179,180],"strong",{},"retrieval-augmented generation",". In plain language: ",[178,183,184],{},"look up, then answer",".",[174,187,188,189,196],{},"The model does not rely only on what it was trained on. It first fetches supporting documents from a company corpus, then writes the answer using those documents. The original research paper is ",[190,191,195],"a",{"href":192,"rel":193},"https://arxiv.org/abs/2005.11401",[194],"nofollow","Lewis et al., Retrieval-Augmented Generation (2020)",". The idea is older than ChatGPT: give the generator evidence at question time so it is less likely to invent.",[174,198,199,202,203,207],{},[178,200,201],{},"Enterprise RAG"," is that move with permissions respected. Search runs as a named person or team, not as an admin crawler of everything. Citations include a document, version, and date. It is how you reduce hallucination on ",[204,205,206],"em",{},"company"," facts.",[174,209,210,211,216],{},"It is not, by itself, an operating system, a write gate, or a memory of decisions. ",[190,212,215],{"href":213,"rel":214},"https://eur-lex.europa.eu/eli/reg/2016/679/oj",[194],"GDPR"," does not pause because the “user” of the files is an AI. Retrieval is still processing personal data.",[174,218,219],{},"The demo omitted the hard parts. A laptop search over a folder of PDFs is not enterprise RAG. Neither is a chatbot that sometimes browses the public web. Enterprise lookup has to survive access lists, freshness SLAs, poisoned documents, and the difference between “this file exists” and “this is policy.”",[221,222,224],"h2",{"id":223},"words-youll-hear","Words you’ll hear",[226,227,228,235,241,247,253,264,270,276,282],"ul",{},[229,230,231,234],"li",{},[178,232,233],{},"Corpus."," The set of files and records the AI is allowed to search. At work, this should be the job’s corpus, not the company’s entire Drive.",[229,236,237,240],{},[178,238,239],{},"Embedding / vector store."," A numerical fingerprint of text, used to find similar passages. Similarity search fails on invoice IDs and clause numbers unless you also use keywords. At work, “find contract 88421” is a keyword problem pretending to be a semantic one.",[229,242,243,246],{},[178,244,245],{},"Citation."," A clickable source Legal can check — not “according to our documents.” At work, the citation needs a version and a date, or it is a vibe.",[229,248,249,252],{},[178,250,251],{},"Hallucination."," Fluent invention. RAG reduces it on company facts. It does not eliminate it, and it does not stop an ungoverned write.",[229,254,255,258,259,263],{},[178,256,257],{},"Asserted policy."," What the company currently wants. That belongs in a ",[190,260,262],{"href":261},"what-is-a-company-wiki-for-ai-agents","company wiki",", not in whichever PDF sounded closest.",[229,265,266,269],{},[178,267,268],{},"Chunking."," Splitting files so search can retrieve a passage. Bad chunking is how a table’s header parts company from its numbers.",[229,271,272,275],{},[178,273,274],{},"Freshness."," When the index sees a change. At work, “we changed the vendor template yesterday” is an SLA question.",[229,277,278,281],{},[178,279,280],{},"Permission-aware search."," The retriever sees what the user (or the job) may see. A superuser crawler is not enterprise; it is a new data store.",[229,283,284,287,288,293],{},[178,285,286],{},"Prompt injection via documents."," Retrieved text that instructs the model to ignore policy. The ",[190,289,292],{"href":290,"rel":291},"https://genai.owasp.org/llm-top-10/",[194],"OWASP Top 10 for LLM applications"," treats that as a security surface.",[174,295,296,297,185],{},"What enterprise RAG is not: a chatbot that sometimes browses the public web; a dump of all tickets into a vector database; a replacement for official playbooks; or permission-aware search sold as a work OS. Search that respects permissions is still search. It does not gate a write. See ",[190,298,300],{"href":299},"/blog/nimbus-vs-glean","Nimbus vs Glean",[221,302,304],{"id":303},"why-you-should-care","Why you should care",[174,306,307],{},"It affects you if answers about policy, customers, or finance will be trusted — and if those answers later need a source you can click.",[174,309,310],{},"Enterprise lookup adds:",[226,312,313,319,324,330],{},[229,314,315,318],{},[178,316,317],{},"Permissions."," SharePoint, Salesforce, and Drive access lists still apply.",[229,320,321,323],{},[178,322,274],{}," If the index updates on Sundays, your SLA is weekly.",[229,325,326,329],{},[178,327,328],{},"Poisoned documents."," Retrieved text can instruct the model to ignore policy.",[229,331,332,335],{},[178,333,334],{},"Purpose."," Indexing everything “just in case” is a privacy and quality problem.",[174,337,338,339,342],{},"Treat RAG as ",[178,340,341],{},"infrastructure with an SLA",", not as a magic brain. Separate asserted versus retrieved. Scope retrieval to the job. Demand citations. Assign owners the way you would for a search service. GDPR erasure is harder if you forgot the index.",[344,345,347],"h3",{"id":346},"what-changes-by-role","What changes by role",[174,349,350,353],{},[178,351,352],{},"Finance."," Retrieval of last year’s close pack is not the close checklist. Numbers in retrieved slides go stale. Finance should insist that thresholds live in asserted wiki tables, and that RAG citations are dated. A fluent answer about recognition policy without a clickable source is not usable in a close.",[174,355,356,359],{},[178,357,358],{},"Legal."," Citations are the point. “According to our documents” is not reviewable. Legal also owns the processing question: indexing HR files into a shared vector store is a new copy of personal data. Erasure requests have to hit the index, not only the source system.",[174,361,362,365],{},[178,363,364],{},"Operations."," Freshness and owners. Ops should treat the retriever like any other search service: uptime, lag, and who gets paged when the wrong SOP is served. Chunking errors show up as “the agent missed the table.”",[174,367,368,371],{},[178,369,370],{},"Go-to-market."," Competitive decks and old playbooks are semantically close to this quarter’s question. Without a conflict rule that wiki wins, GTM will ship last year’s discount floor because it matched the query. RAG without assertion is folklore with better ranking.",[174,373,374,377,378,382],{},[178,375,376],{},"Security."," Superuser crawlers, poisoned documents, and a second store of sensitive text. Security should ask who the retriever authenticates as, whether ",[190,379,381],{"href":380},"what-is-model-context-protocol","MCP"," helpers search as a superuser, and whether prompt injection in a PDF can change tool behaviour. Network search products are not write gates.",[344,384,386],{"id":385},"what-people-get-wrong","What people get wrong",[174,388,389,392],{},[178,390,391],{},"Indexing everything."," Quality falls. Privacy rises. Purpose disappears.",[174,394,395,398,399,185],{},[178,396,397],{},"RAG as an OS."," Lookup does not isolate jobs, quote writes, or store decisions. See ",[190,400,402],{"href":401},"what-is-an-enterprise-ai-operating-system","What is an enterprise AI operating system",[174,404,405,408],{},[178,406,407],{},"RAG as the wiki."," Retrieved files are what exists. The wiki is what is in force.",[174,410,411,414],{},[178,412,413],{},"Citations without versions."," Legal cannot check “the wiki” or “our Drive.”",[174,416,417,420],{},[178,418,419],{},"Warehouse SQL as a substitute."," “What is our revenue recognition policy?” is retrieval. “What was Q4 revenue by region?” is structured query. Many jobs need both.",[174,422,423,426],{},[178,424,425],{},"Assuming hallucination is solved."," Missing files still produce fluent guesses. Ungoverned writes still land.",[174,428,429,430,434],{},"Good looks like: permission-aware retrieval scoped to the ",[190,431,433],{"href":432},"what-is-an-ai-workstream","workstream",", hybrid keyword plus similarity, dated citations, a wiki conflict rule, an index SLA, and a write gate that does not care how good the retrieval was. Failure looks like a tenant-wide vector lake labelled “the brain.”",[174,436,437,438,441],{},"Lewis et al. (2020) showed that lookup-then-answer reduces invention on facts in the corpus. Enterprise buyers still have to decide which corpus, whose permissions, and whether a retrieved PDF is allowed to outrank the ",[190,439,440],{"href":261},"wiki",". The paper does not answer those questions. Your runtime must.",[221,443,445],{"id":444},"how-this-shows-up-in-nimbus","How this shows up in Nimbus",[174,447,448,449,452,453,457],{},"Nimbus uses RAG-like retrieval ",[178,450,451],{},"inside"," a work OS, not as a standalone search SKU. Wiki is asserted policy. Connectors supply live context. Workstreams pre-scope the corpus. Governance still gates any write. The Lifecycle Graph stores which sources were used for a decision — retrieval becomes part of ",[190,454,456],{"href":455},"what-is-institutional-memory-in-enterprise-ai","institutional memory",", not a forgotten context window.",[174,459,460,461,464],{},"A common plug so AI apps can use the same tools — ",[190,462,463],{"href":380},"Model Context Protocol"," — can standardise access to repositories. It does not implement access lists for you. A tool that searches Drive as a superuser is still a superuser.",[174,466,467,468,471,472,474,475,185],{},"See ",[190,469,470],{"href":28},"Wiki",", ",[190,473,31],{"href":32},", and ",[190,476,39],{"href":40},[221,478,480],{"id":479},"questions-people-actually-ask","Questions people actually ask",[344,482,484],{"id":483},"will-rag-stop-the-model-making-things-up","Will RAG stop the model making things up?",[174,486,487],{},"It reduces invention on facts that exist in authorised files. It does not make the model honest about missing files, and it does not replace a person on a live-system change.",[344,489,491],{"id":490},"is-indexing-everything-just-in-case-a-good-idea","Is indexing everything “just in case” a good idea?",[174,493,494],{},"No. Indexing without a purpose is a privacy and quality problem. Scope the corpus to the job.",[344,496,498],{"id":497},"how-is-this-different-from-a-company-wiki","How is this different from a company wiki?",[174,500,501,502,505,506,509],{},"The wiki is what the company ",[204,503,504],{},"wants"," to be true. RAG is what ",[204,507,508],{},"exists"," in files. If they conflict, the wiki should win unless a human promotes a change.",[344,511,513],{"id":512},"can-warehouse-sql-replace-rag","Can warehouse SQL replace RAG?",[174,515,516],{},"They answer different questions. Policy prose is retrieval. Regional revenue is a query. Many real jobs need both.",[344,518,520],{"id":519},"is-glean-or-similar-an-enterprise-ai-os","Is Glean (or similar) an enterprise AI OS?",[174,522,523,524,185],{},"Permission-aware search is still search. It does not, by itself, quote a CRM write or bind a named signer. See ",[190,525,300],{"href":299},[344,527,529],{"id":528},"why-do-invoice-numbers-fail-in-vector-search","Why do invoice numbers fail in vector search?",[174,531,532,533,537],{},"Embeddings capture similarity of meaning, not identity of tokens. Hybrid search — keywords plus vectors — is how you find ",[534,535,536],"code",{},"INV-88421"," instead of a semantically nearby invoice.",[344,539,541],{"id":540},"how-fast-should-the-index-update","How fast should the index update?",[174,543,544],{},"As fast as the decision you are supporting. If a template changed yesterday and the agent still cites last month, your SLA is wrong. Publish the lag.",[344,546,548],{"id":547},"what-is-document-based-prompt-injection","What is document-based prompt injection?",[174,550,551,552,556],{},"A retrieved file that says, in effect, “ignore previous instructions.” Treat retrieved text as untrusted input. The ",[190,553,555],{"href":290,"rel":554},[194],"OWASP LLM list"," is the starting point. A wiki conflict rule and a write gate still matter.",[344,558,560],{"id":559},"does-gdpr-apply-to-the-vector-index","Does GDPR apply to the vector index?",[174,562,563,564,568,569,185],{},"Yes, if it holds personal data. The index is another copy. Erasure, purpose, and access control apply. See the ",[190,565,567],{"href":213,"rel":566},[194],"GDPR text"," and ",[190,570,573],{"href":571,"rel":572},"https://ico.org.uk/for-organisations/uk-gdpr-guidance-and-resources/artificial-intelligence/",[194],"ICO AI guidance",[344,575,577],{"id":576},"can-mcp-make-retrieval-respect-permissions","Can MCP make retrieval respect permissions?",[174,579,580],{},"Only if the helper is built that way. The protocol will happily pass superuser results.",[344,582,584],{"id":583},"should-customer-facing-chatbots-use-rag-on-the-public-website-plus-internal-policy","Should customer-facing chatbots use RAG on the public website plus internal policy?",[174,586,587,588,593],{},"Internal policy in a customer bot is how invented fares happen unless a human still owns the commitment. Air Canada’s case — ",[190,589,592],{"href":590,"rel":591},"https://www.cbc.ca/news/canada/british-columbia/air-canada-chatbot-lawsuit-1.7116416",[194],"CBC"," — is retrieval-plus-generation without a working gate.",[344,595,597],{"id":596},"how-do-we-know-which-sources-a-decision-used","How do we know which sources a decision used?",[174,599,600,601,605],{},"Record them on the ",[190,602,604],{"href":603},"what-is-a-lifecycle-graph","lifecycle graph",". A context window that evaporates is not memory.",[221,607,609],{"id":608},"related-reading","Related reading",[174,611,612,568,615,185],{},[190,613,614],{"href":261},"What is a company wiki for AI agents",[190,616,617],{"href":455},"What is institutional memory in enterprise AI",[221,619,621],{"id":620},"sources","Sources",[226,623,624,629,634,639,645],{},[229,625,626],{},[190,627,195],{"href":192,"rel":628},[194],[229,630,631],{},[190,632,215],{"href":213,"rel":633},[194],[229,635,636],{},[190,637,292],{"href":290,"rel":638},[194],[229,640,641],{},[190,642,644],{"href":571,"rel":643},[194],"ICO, AI and data protection",[229,646,647],{},[190,648,650],{"href":590,"rel":649},[194],"CBC, Air Canada chatbot lawsuit",{"title":155,"searchDepth":156,"depth":156,"links":652},[653,654,659,660,674,675],{"id":223,"depth":156,"text":224},{"id":303,"depth":156,"text":304,"children":655},[656,658],{"id":346,"depth":657,"text":347},3,{"id":385,"depth":657,"text":386},{"id":444,"depth":156,"text":445},{"id":479,"depth":156,"text":480,"children":661},[662,663,664,665,666,667,668,669,670,671,672,673],{"id":483,"depth":657,"text":484},{"id":490,"depth":657,"text":491},{"id":497,"depth":657,"text":498},{"id":512,"depth":657,"text":513},{"id":519,"depth":657,"text":520},{"id":528,"depth":657,"text":529},{"id":540,"depth":657,"text":541},{"id":547,"depth":657,"text":548},{"id":559,"depth":657,"text":560},{"id":576,"depth":657,"text":577},{"id":583,"depth":657,"text":584},{"id":596,"depth":657,"text":597},{"id":608,"depth":156,"text":609},{"id":620,"depth":156,"text":621},"2026-08-17","Enterprise RAG is looking up authorised company files before the AI answers, with permissions respected — lookup-then-answer, not an operating system.","/blog/what-is-enterprise-rag",{"title":165,"description":677},"explainer","blog/what-is-enterprise-rag",[680,683,684,685],"rag","retrieval","knowledge","0nzyVDNSVz_C1CRbLIQnUhimJK_EGwW9s_e9H6c8tyo",{"hero":688,"id":690,"title":691,"archived":149,"authors":150,"badge":150,"body":692,"date":150,"department":150,"description":696,"extension":158,"eyebrow":697,"faqHeader":150,"faqs":150,"footerBand":698,"headline":150,"image":150,"industry":150,"jobType":150,"listed":149,"location":150,"navigation":115,"openRoles":150,"pageLayout":150,"path":60,"relatedHeading":704,"seo":705,"series":150,"sitemap":115,"status":150,"stem":706,"subhead":150,"tags":150,"video":150,"whyJoin":150,"workplaceType":150,"__hash__":707},{"filename":689},"u2221455217_Flat_design_of_a_futuristic_minimalist_landscape__5d589295-cdea-4ea9-a262-be766881accf_1.png","content/blog/index.md","Exploring the future of intelligence.",{"type":152,"value":693,"toc":694},[],{"title":155,"searchDepth":156,"depth":156,"links":695},[],"Deep dives into pre-cognitive intelligence, sentient enterprises, and the evolving landscape of AI-driven business transformation.","Latest Research",{"headline":699,"description":700,"primaryLabel":701,"primaryTo":702,"secondaryLabel":703,"secondaryTo":12},"Stay at the frontier.","Subscribe for product updates and new insights.","Subscribe","/newsletter","Explore the platform","More research",{"title":691,"description":696},"blog/index","eK1RCXdDW8nfLSyKRXGB1mJm9FAmhAO6GWXwNKOMNVE",[709,1172],{"id":710,"title":711,"archived":149,"authors":712,"badge":714,"body":715,"date":676,"department":150,"description":1163,"extension":158,"eyebrow":150,"faqHeader":150,"faqs":150,"footerBand":150,"headline":150,"image":150,"industry":150,"jobType":150,"listed":149,"location":150,"navigation":115,"openRoles":150,"pageLayout":150,"path":1164,"relatedHeading":150,"seo":1165,"series":680,"sitemap":115,"status":150,"stem":1166,"subhead":150,"tags":1167,"video":150,"whyJoin":150,"workplaceType":150,"__hash__":1171},"content/blog/what-is-human-in-the-loop-ai.md","What is Human-in-the-Loop AI",[713],{"name":168,"to":120},{"label":170},{"type":152,"value":716,"toc":1139},[717,723,726,740,743,746,748,818,826,828,831,834,845,852,859,866,869,871,876,881,891,896,909,911,917,922,928,934,940,946,952,964,977,980,982,985,992,998,1005,1007,1011,1014,1018,1021,1025,1028,1032,1035,1039,1042,1046,1053,1057,1063,1067,1070,1074,1077,1081,1084,1088,1091,1095,1098,1100,1112,1114],[174,718,719,720,185],{},"Human-in-the-loop AI, in everyday language, means ",[178,721,722],{},"a person must approve before the AI can finish the job",[174,724,725],{},"Not “a human might read the chat.” Not a footer that says this content was generated. A gate the software cannot skip.",[174,727,728,729,733,734,739],{},"In February 2024, a British Columbia tribunal held Air Canada responsible for a chatbot that invented a bereavement-fare policy. ",[190,730,732],{"href":590,"rel":731},[194],"CBC reported"," that the airline’s argument — the chatbot is a separate legal entity — failed. The decision is ",[190,735,738],{"href":736,"rel":737},"https://decisions.civilresolutionbc.ca/crt/crtd/en/525448/1/document.do",[194],"Moffatt v. Air Canada",". A customer relied on the invented fare. A human did not catch the fiction before it became a commitment.",[174,741,742],{},"That is the class of failure this article is about.",[174,744,745],{},"The phrase is older than ChatGPT. Safety engineering already distinguished a signer on every payload from a supervisor with a kill switch. Generative AI borrowed the label and diluted it. Vendors now say “human in the loop” for a thumbs-up on a chat, a weekly review of logs, or a prompt that says “ask the user first.” Only one of those is a gate.",[221,747,224],{"id":223},[226,749,750,756,766,772,784,790,796,802,812],{},[229,751,752,755],{},[178,753,754],{},"In the loop."," The process cannot proceed past a gate without a human act. At work, the CRM write does not execute until a named person signs the quoted fields.",[229,757,758,761,762,765],{},[178,759,760],{},"On the loop."," The system runs; a human ",[204,763,764],{},"can"," stop it. Intervention is possible. It is not required per action. At work, this may be acceptable for read-only monitoring. It is not a write control.",[229,767,768,771],{},[178,769,770],{},"Theatre."," A checkbox “I understand this is AI,” or a prompt that says “ask the user first,” while the model may still act.",[229,773,774,777,778,783],{},[178,775,776],{},"Effective oversight."," ",[190,779,782],{"href":780,"rel":781},"https://eur-lex.europa.eu/eli/reg/2024/1689/oj",[194],"EU AI law"," Article 14: for higher-risk systems, people must be able to interpret outputs, stay aware that automation can lull them, and interrupt the system.",[229,785,786,789],{},[178,787,788],{},"Rubber stamp."," A gate that fires so often people auto-click. That is not oversight. It is fatigue.",[229,791,792,795],{},[178,793,794],{},"Named signer."," Identity bound to the decision. Shared inboxes destroy this.",[229,797,798,801],{},[178,799,800],{},"Quote / payload."," The exact change in the language of the live system — opportunity fields, journal lines, email body — not a wall of prompt text.",[229,803,804,807,808,185],{},[178,805,806],{},"Fail-closed."," Missing approval means nothing happens. See ",[190,809,811],{"href":810},"what-is-write-back-governance","What is write-back governance",[229,813,814,817],{},[178,815,816],{},"Maker-checker."," An older control: one person proposes, another authorises. HITL for AI is that instinct when the proposer is a model.",[174,819,820,825],{},[190,821,824],{"href":822,"rel":823},"https://www.reuters.com/legal/new-york-lawyers-sanctioned-using-fake-chatgpt-cases-legal-brief-2023-06-22/",[194],"Mata v. Avianca"," is the cousin case on the legal side: fluent fiction entered a court record because no working check caught invented citations. The loop failed before filing, not after.",[221,827,304],{"id":303},[174,829,830],{},"Enterprise buyers should demand a person at the gate for writes to live business systems and for customer-facing commitments. They may accept “on the loop” for read-only monitoring. They should reject theatre.",[174,832,833],{},"It affects you if AI can:",[226,835,836,839,842],{},[229,837,838],{},"change records or money",[229,840,841],{},"send a customer a message that asserts a policy, price, or term",[229,843,844],{},"affect employment, credit, or people’s rights",[174,846,847,848,851],{},"Place people where ",[178,849,850],{},"risk and reversibility"," change: before writes, before external messages, and at exception thresholds (amount, region, data class).",[174,853,854,855,858],{},"Do ",[178,856,857],{},"not"," put humans on every sentence. A gate that fires fifty times a day will be auto-clicked.",[174,860,861,862,865],{},"The person who already owns that class of change in the analogue process should sign it here. Inventing an “AI champion” who approves finance journals ",[204,863,864],{},"and"," legal emails is how you get a rubber stamp.",[174,867,868],{},"Show the change in the language of the live system. A person cannot oversee what they cannot parse.",[344,870,347],{"id":346},[174,872,873,875],{},[178,874,352],{}," Journals, forecast overrides, and material fields need the same owner who would sign in the analogue close. A champion who does not own the ledger will click through. Rejects are success: they prove the gate. A six-month zero reject rate is a finding.",[174,877,878,880],{},[178,879,358],{}," Customer commitments and filings need a signer who can interpret the payload. Air Canada is customer-facing fiction. Mata v. Avianca is professional fiction entering a record. Legal should also refuse “Act compliant” claims that rest only on a button. Article 14 is a bundle of duties, not a widget.",[174,882,883,885,886,890],{},[178,884,364],{}," Place gates at reversibility boundaries. Ops should measure time-to-approved-write and reject rate, and should treat human wait as a first-class ",[190,887,889],{"href":888},"what-is-an-agentic-workflow","workflow"," step, not a Slack nudge.",[174,892,893,895],{},[178,894,370],{}," Friction is real. The honest comparison is unreviewed mutation versus incident response, not versus a demo that writes instantly. GTM should not be asked to approve legal emails, and legal should not be asked to approve Amount.",[174,897,898,900,901,904,905,908],{},[178,899,376],{}," The gate must be unskippable by the model, including after prompt injection. A jailbreak can trick the model into ",[204,902,903],{},"requesting"," a bad write. It should not be able to ",[204,906,907],{},"execute"," without a quote and a signer. Identity binding matters: a generic “approve” in a shared inbox is not a control.",[344,910,386],{"id":385},[174,912,913,916],{},[178,914,915],{},"On the loop as in the loop."," A kill switch is not a per-action signer.",[174,918,919,921],{},[178,920,770],{}," Footers, checkboxes, and “shall I proceed?” in unbound chat.",[174,923,924,927],{},[178,925,926],{},"Too many gates."," Fatigue produces rubber stamps. Fewer gates, better quotes.",[174,929,930,933],{},[178,931,932],{},"Wrong human."," Whoever is online, or an AI champion spanning domains.",[174,935,936,939],{},[178,937,938],{},"Chat as the quote."," Prompt text is not field-level change.",[174,941,942,945],{},[178,943,944],{},"HITL as sufficient for the EU AI Act."," Oversight is necessary, not sufficient, for higher-risk systems.",[174,947,948,949,951],{},"Good looks like: read-only analysis without a click per sentence; quoted writes; named roles; fail-closed execution; rejects stored on the ",[190,950,604],{"href":603},"; metrics on reject rates. Failure looks like a prompt, a footer, and a customer who relied on the bot.",[174,953,954,955,958,959,963],{},"The person should sit at ",[178,956,957],{},"release",", not at every internal hand-off between ",[190,960,962],{"href":961},"what-is-multi-agent-ai","agent teams",". Internal critics can reduce garbage. They are not the signer.",[174,965,966,967,970,971,973,974,976],{},"Adjacent ideas are easy to mix. ",[190,968,969],{"href":810},"Write-back governance"," is the fail-closed property of the write. HITL is the human act that satisfies it. A ",[190,972,604],{"href":603}," is how you prove the act later. A ",[190,975,433],{"href":432}," is whose job the gate belongs to. None of those is a footer on a chatbot.",[174,978,979],{},"Fatigue is the operational enemy. If every sentence needs a click, people will click. If only irreversible steps need a click, people can still read. Design the quote so a finance owner can say yes or no in the language of the journal, and a legal owner can say yes or no in the language of the email body. Mixed payloads produce mixed, tired humans.",[221,981,445],{"id":444},[174,983,984],{},"Read-only connectors mean the loop can analyse without a human per sentence.",[174,986,987,988,991],{},"When a write is proposed, governance ",[178,989,990],{},"quotes"," it and stops. Named roles must sign. Agent teams can draft. They cannot waive the gate. Missing approval is fail-closed.",[174,993,994,995,997],{},"The ",[190,996,23],{"href":603}," stores the human act: who signed, what they saw, what happened next — including rejects. Perception can list rejected items.",[174,999,467,1000,1002,1003,185],{},[190,1001,39],{"href":40},". Companion: ",[190,1004,811],{"href":810},[221,1006,480],{"id":479},[344,1008,1010],{"id":1009},"isnt-this-just-slower-ai","Isn’t this just slower AI?",[174,1012,1013],{},"It is slower than ungoverned writes and faster than incident response. Invented policy is cheaper to catch in a quote than in a tribunal.",[344,1015,1017],{"id":1016},"who-should-be-the-human","Who should be the human?",[174,1019,1020],{},"The owner of the live-system change or the customer commitment, not “whoever is online.” Shared inboxes destroy accountability.",[344,1022,1024],{"id":1023},"does-a-person-at-the-gate-satisfy-eu-ai-law-by-itself","Does a person-at-the-gate satisfy EU AI law by itself?",[174,1026,1027],{},"No. Higher-risk systems have a bundle of duties. Oversight is necessary, not sufficient. Do not claim “Act compliant” because you have a button.",[344,1029,1031],{"id":1030},"how-do-we-stop-rubber-stamping","How do we stop rubber-stamping?",[174,1033,1034],{},"Fewer gates, better quotes, metrics on reject rates. A six-month zero reject rate on CRM writes is a finding: either you are perfect, or nobody is reading.",[344,1036,1038],{"id":1037},"is-a-chat-saying-shall-i-proceed-enough","Is a chat saying “shall I proceed?” enough?",[174,1040,1041],{},"Only if it is bound to identity, shows the payload, and cannot be skipped.",[344,1043,1045],{"id":1044},"what-is-the-difference-between-in-the-loop-and-on-the-loop","What is the difference between in the loop and on the loop?",[174,1047,1048,1049,1052],{},"In the loop: the job cannot finish the risky step without a human act. On the loop: a human ",[204,1050,1051],{},"may"," intervene. Vendors blur them because the second is cheaper to ship.",[344,1054,1056],{"id":1055},"can-agent-teams-approve-each-others-work","Can agent teams approve each other’s work?",[174,1058,1059,1060,185],{},"They can criticise drafts. Release still needs a named human. Multi-agent review is not a signer. See ",[190,1061,1062],{"href":961},"What is multi-agent AI",[344,1064,1066],{"id":1065},"do-read-only-jobs-need-a-person-every-time","Do read-only jobs need a person every time?",[174,1068,1069],{},"Usually not. That is the point of connectors defaulting to read-only. Put people where reversibility changes.",[344,1071,1073],{"id":1072},"how-does-this-relate-to-air-canada","How does this relate to Air Canada?",[174,1075,1076],{},"A customer-facing chatbot made a commitment with no working human catch. The tribunal did not treat the bot as a separate legal person. If your loop can send or display a policy, price, or term, you need a gate or you own the fiction.",[344,1078,1080],{"id":1079},"what-about-mata-v-avianca","What about Mata v. Avianca?",[174,1082,1083],{},"Lawyers filed invented case law from ChatGPT. The failure was the missing check before the record changed. The same pattern waits in CRM and ERP.",[344,1085,1087],{"id":1086},"can-we-batch-approve-200-records","Can we batch-approve 200 records?",[174,1089,1090],{},"Not as one click with no visible set. Bulk without inspection is a rubber stamp with worse radius. Show the set.",[344,1092,1094],{"id":1093},"does-logging-approvals-in-slack-count","Does logging approvals in Slack count?",[174,1096,1097],{},"Only if identity, payload, and outcome are bound and retained as a control record. A thumbs-up emoji is theatre.",[221,1099,609],{"id":608},[174,1101,1102,568,1106,1108,1109,1111],{},[190,1103,1105],{"href":1104},"what-is-ai-governance","What is AI governance",[190,1107,1062],{"href":961}," — the person should sit at ",[178,1110,957],{},", not at every internal hand-off.",[221,1113,621],{"id":620},[226,1115,1116,1121,1127,1133],{},[229,1117,1118],{},[190,1119,650],{"href":590,"rel":1120},[194],[229,1122,1123],{},[190,1124,1126],{"href":736,"rel":1125},[194],"Civil Resolution Tribunal, Moffatt v. Air Canada",[229,1128,1129],{},[190,1130,1132],{"href":780,"rel":1131},[194],"EU AI Act (Regulation 2024/1689), including Article 14",[229,1134,1135],{},[190,1136,1138],{"href":822,"rel":1137},[194],"Reuters, New York lawyers sanctioned for ChatGPT fake cases",{"title":155,"searchDepth":156,"depth":156,"links":1140},[1141,1142,1146,1147,1161,1162],{"id":223,"depth":156,"text":224},{"id":303,"depth":156,"text":304,"children":1143},[1144,1145],{"id":346,"depth":657,"text":347},{"id":385,"depth":657,"text":386},{"id":444,"depth":156,"text":445},{"id":479,"depth":156,"text":480,"children":1148},[1149,1150,1151,1152,1153,1154,1155,1156,1157,1158,1159,1160],{"id":1009,"depth":657,"text":1010},{"id":1016,"depth":657,"text":1017},{"id":1023,"depth":657,"text":1024},{"id":1030,"depth":657,"text":1031},{"id":1037,"depth":657,"text":1038},{"id":1044,"depth":657,"text":1045},{"id":1055,"depth":657,"text":1056},{"id":1065,"depth":657,"text":1066},{"id":1072,"depth":657,"text":1073},{"id":1079,"depth":657,"text":1080},{"id":1086,"depth":657,"text":1087},{"id":1093,"depth":657,"text":1094},{"id":608,"depth":156,"text":609},{"id":620,"depth":156,"text":621},"Human-in-the-loop AI means a person must approve before the AI can finish the job — seeing the exact change, signing with their identity, and leaving a record.","/blog/what-is-human-in-the-loop-ai",{"title":711,"description":1163},"blog/what-is-human-in-the-loop-ai",[680,1168,1169,1170],"human-in-the-loop","governance","approvals","8xJuZYtNYOnCYCsxxbZqk340T6yMdyoPDYPvvIta_v8",{"id":1173,"title":1174,"archived":149,"authors":1175,"badge":1177,"body":1178,"date":676,"department":150,"description":1608,"extension":158,"eyebrow":150,"faqHeader":150,"faqs":150,"footerBand":150,"headline":150,"image":150,"industry":150,"jobType":150,"listed":149,"location":150,"navigation":115,"openRoles":150,"pageLayout":150,"path":1609,"relatedHeading":150,"seo":1610,"series":680,"sitemap":115,"status":150,"stem":1611,"subhead":150,"tags":1612,"video":150,"whyJoin":150,"workplaceType":150,"__hash__":1616},"content/blog/what-is-causal-ai-for-operations.md","What is Causal AI for Operations",[1176],{"name":168,"to":120},{"label":170},{"type":152,"value":1179,"toc":1584},[1180,1186,1201,1208,1211,1214,1216,1273,1276,1291,1293,1296,1299,1323,1329,1343,1354,1356,1365,1370,1375,1380,1385,1387,1393,1399,1405,1411,1417,1420,1423,1436,1438,1447,1450,1460,1462,1466,1469,1473,1476,1480,1483,1487,1490,1494,1497,1501,1507,1511,1514,1518,1521,1525,1528,1532,1535,1539,1544,1548,1551,1553,1562,1564],[174,1181,1182,1183],{},"“Causal AI” is a phrase people type into ChatGPT when they mean: ",[178,1184,1185],{},"can we tell why something happened, or are we guessing?",[174,1187,1188,1189,1194,1195,1200],{},"In statistics, causality is a serious science. Did the discount cause the win, or did seasonality? That needs experiments and careful assumptions, not a model that says “because.” The ",[190,1190,1193],{"href":1191,"rel":1192},"https://plato.stanford.edu/entries/causal-models/",[194],"Stanford Encyclopedia of Philosophy’s entry on causal models"," is a fair orientation to that science. Judea Pearl’s overview, ",[190,1196,1199],{"href":1197,"rel":1198},"https://ftp.cs.ucla.edu/pub/stat_ser/r350.pdf",[194],"Causal inference in statistics",", is the technical companion: identification is a design problem, not a paragraph problem.",[174,1202,1203,1204,1207],{},"In operations, the question is more everyday and more urgent: ",[178,1205,1206],{},"why did this field, journal, or customer message change?"," If you cannot replay the brief, the sources, the named approval, and the live-system result, you have a dashboard, not a cause.",[174,1209,1210],{},"This article is about that second meaning. Nimbus does not claim to estimate market lift from a chatbot. It does claim you should be able to reconstruct the intervention.",[174,1212,1213],{},"Mixing the two meanings is how board decks get written. Two charts rose together; the model wrote a fluent “because”; finance cannot sample the journal. You can have excellent statistics in a notebook and still be unable to say who approved last night’s ERP write. You can have an excellent operations record and still be wrong about the market. Do not let one pretend to be the other.",[221,1215,224],{"id":223},[226,1217,1218,1224,1230,1236,1245,1257,1267],{},[229,1219,1220,1223],{},[178,1221,1222],{},"Cause vs correlation."," Two lines rising together is not proof that one caused the other. At work, AI usage and pipeline in the same quarter is a coincidence until you show the steps.",[229,1225,1226,1229],{},[178,1227,1228],{},"Intervention."," Something you actually did — an approval, a write, a refusal. At work, a fail-closed gate that blocked a write is an intervention with a known counterfactual: nothing would have changed.",[229,1231,1232,1235],{},[178,1233,1234],{},"Identification."," The statistics problem of isolating a true effect. Different from a work record. At work, this is “did signed next-step updates cause wins?” — a question for a designed comparison, not for Perception.",[229,1237,1238,1241,1242,185],{},[178,1239,1240],{},"Lifecycle graph."," The company’s chain of AI work: what was asked, who signed, what changed. See ",[190,1243,1244],{"href":603},"What is a lifecycle graph",[229,1246,1247,1250,1251,1256],{},[178,1248,1249],{},"Provenance."," Who, what, when, derived from what. ",[190,1252,1255],{"href":1253,"rel":1254},"https://www.w3.org/TR/prov-overview/",[194],"W3C PROV"," is the open vocabulary for that idea.",[229,1258,1259,1262,1263,185],{},[178,1260,1261],{},"Confounder."," In science, a hidden third factor. In operations, the hidden factor is often “a human pasted a consumer-model answer into CRM.” See ",[190,1264,1266],{"href":1265},"what-is-shadow-ai","What is shadow AI",[229,1268,1269,1272],{},[178,1270,1271],{},"Rationale."," The model’s English explanation. Often written after the fact. Not a recorded structure.",[174,1274,1275],{},"Keep two layers apart:",[1277,1278,1279,1285],"ol",{},[229,1280,1281,1284],{},[178,1282,1283],{},"Causal science."," Did the discount cause the win? Needs a design, not a fluent paragraph.",[229,1286,1287,1290],{},[178,1288,1289],{},"Causal operations."," Brief → sources → proposal → approval → write → system response. Needs a record.",[221,1292,304],{"id":303},[174,1294,1295],{},"Boards get briefed on “AI caused the pipeline jump” because both charts went up. Finance cannot sample a journal that only exists as a chat. Legal cannot explain a CRM exception that lived in someone’s personal account.",[174,1297,1298],{},"Causal operations affects you if you:",[226,1300,1301,1307,1317],{},[229,1302,1303,1306],{},[178,1304,1305],{},"Have to explain a change."," “Who caused this field to move, against which rule?”",[229,1308,1309,1316],{},[178,1310,1311,1312,1315],{},"Need to know what ",[204,1313,1314],{},"would"," have happened without approval."," In a real gate, the answer is nothing.",[229,1318,1319,1322],{},[178,1320,1321],{},"Are tempted to file a model’s “because” as truth."," Natural-language rationales are often written after the fact.",[174,1324,1325,1326,1328],{},"You do ",[178,1327,857],{}," need a data-science sprint to ask:",[226,1330,1331,1334,1337,1340],{},[229,1332,1333],{},"Why was this record changed?",[229,1335,1336],{},"Which policy version caused this refusal?",[229,1338,1339],{},"Did a spend cap stop the run?",[229,1341,1342],{},"Did analysis change the CRM, or only produce a draft?",[174,1344,1345,1346,1349,1350,1353],{},"You ",[178,1347,1348],{},"should"," need a statistician if you want to know whether signed next-step updates ",[204,1351,1352],{},"caused"," wins. The work record can attach “this account was treated.” Estimation is extra.",[344,1355,347],{"id":346},[174,1357,1358,1360,1361,1364],{},[178,1359,352],{}," Sampling a journal requires the chain, not a story. Spend caps that fire are causes of ",[204,1362,1363],{},"inaction",", which close packs also need to explain. Do not let “AI lift” into a board pack without either an experiment or an honest “we do not know.”",[174,1366,1367,1369],{},[178,1368,358],{}," Discovery and customer commitments need the payload the signer saw. A model rationale is advocacy, not evidence. Legal should also stop people treating a chatbot explanation as the company’s official why.",[174,1371,1372,1374],{},[178,1373,364],{}," This is the native question: why did this change, who signed, what was refused. Ops should keep BI for canonical metrics and the graph for AI-work lineage. Dumping bookings into the graph as a fake causal model is a mess.",[174,1376,1377,1379],{},[178,1378,370],{}," Forecast meetings will try to credit the copilot. GTM needs reconstructable interventions (which opportunities were touched, by which job) and should refuse market-lift claims without a design. Correlation slides train everyone to stop asking.",[174,1381,1382,1384],{},[178,1383,376],{}," Reconstructability is also incident response: which connector was read-only, which tool was called, whether a jailbreak requested a write that the gate refused. The refusal is a causal fact worth keeping.",[344,1386,386],{"id":385},[174,1388,1389,1392],{},[178,1390,1391],{},"The model’s “because.”"," Fluency is not identification and not a recorded structure.",[174,1394,1395,1398],{},[178,1396,1397],{},"Two rising lines."," Correlation. File it as a hypothesis.",[174,1400,1401,1404],{},[178,1402,1403],{},"Using the same AI that proposed the treatment to declare success."," That is marking your own homework.",[174,1406,1407,1410],{},[178,1408,1409],{},"Skipping the operations layer to buy a science platform."," Without reconstructable interventions, the science team inherits Slack folklore.",[174,1412,1413,1416],{},[178,1414,1415],{},"Using the graph as a BI tool."," Canonical commercial metrics stay in the warehouse. The graph answers mixed policy / approval / live-system questions.",[174,1418,1419],{},"Good looks like: a chain you can query, read-only analysis recorded as non-writes, named signers, spend stops as events, and a bright line before anyone claims lift. Failure looks like a dashboard, a chatbot paragraph, and a forecast that nobody can unwind.",[174,1421,1422],{},"Pearl’s identification problem and an operations reconstruction problem share a word and almost nothing else. Keep the word, split the buying decision. You can staff science later. You cannot reconstruct a write you never recorded.",[174,1424,1425,1426,1428,1429,1432,1433,1435],{},"Adjacent: ",[190,1427,604],{"href":603}," is the product shape of the operations layer. ",[190,1430,1431],{"href":455},"Institutional memory"," is what remains after people leave. ",[190,1434,969],{"href":810}," makes “nothing happened” a possible true answer.",[221,1437,445],{"id":444},[174,1439,1440,1441,1443,1444,185],{},"Nimbus implements causal operations as the ",[178,1442,23],{}," plus ",[178,1445,1446],{},"Perception",[174,1448,1449],{},"Work runs are chains you can query in ordinary language. The company wiki is often the parent of a refusal (“this playbook caused the flag”). Connectors default to read-only, which is itself a causal fact: analysis did not change the CRM. Spend quotes make cost an explicit stop, not an ambient cloud bill. Fail-closed writes mean a missing named signer is a recorded non-event with a known counterfactual.",[174,1451,467,1452,568,1454,1456,1457,185],{},[190,1453,23],{"href":24},[190,1455,1446],{"href":36},". Writes that cannot happen without a signer are ",[190,1458,1459],{"href":810},"write-back governance",[221,1461,480],{"id":479},[344,1463,1465],{"id":1464},"if-the-model-explains-why-is-that-causal-ai","If the model explains “why,” is that causal AI?",[174,1467,1468],{},"No. A fluent paragraph is not a recorded structure, and it is not a statistical identification.",[344,1470,1472],{"id":1471},"do-we-need-advanced-causal-statistics-to-buy-an-operating-layer","Do we need advanced causal statistics to buy an operating layer?",[174,1474,1475],{},"No. You need reconstructable interventions. If you later staff a science team, they will thank you for not storing decisions as Slack folklore.",[344,1477,1479],{"id":1478},"can-the-graph-estimate-lift","Can the graph estimate lift?",[174,1481,1482],{},"Only if you design an experiment or a credible comparison and collect the right outcomes. Beware of using the same AI that proposed the treatment to declare the treatment a success.",[344,1484,1486],{"id":1485},"where-does-this-end-and-a-bi-tool-start","Where does this end and a BI tool start?",[174,1488,1489],{},"The graph is for AI-work lineage and mixed policy / approval / live-system questions. BI remains for canonical commercial metrics. Dumping bookings into the graph as a fake causal model is a mess.",[344,1491,1493],{"id":1492},"what-is-an-intervention-in-this-sense","What is an intervention in this sense?",[174,1495,1496],{},"An approval, a write, a refusal, or a spend stop — something the company actually did (or refused to do) in software. Not a correlation on a slide.",[344,1498,1500],{"id":1499},"why-does-a-fail-closed-gate-matter-for-causality","Why does a fail-closed gate matter for causality?",[174,1502,1503,1504,1506],{},"Because the counterfactual is clean: without the named signer, the live system does not change. Fail-open systems cannot say what ",[204,1505,1314],{}," have happened; they can only hope someone noticed.",[344,1508,1510],{"id":1509},"is-w3c-prov-the-same-as-a-lifecycle-graph","Is W3C PROV the same as a lifecycle graph?",[174,1512,1513],{},"PROV is a standard for provenance concepts. A lifecycle graph is an operational record of AI-mediated work. You can be inspired by PROV without claiming a full W3C implementation.",[344,1515,1517],{"id":1516},"can-we-reconstruct-causes-from-crm-field-history-plus-slack","Can we reconstruct causes from CRM field history plus Slack?",[174,1519,1520],{},"Field history says the value changed. Slack may contain a rumour. Neither joins playbook version, quoted payload, and signer identity as a single chain.",[344,1522,1524],{"id":1523},"does-causal-ai-mean-the-model-uses-causal-graphs-internally","Does “causal AI” mean the model uses causal graphs internally?",[174,1526,1527],{},"Sometimes, in research marketing. In this article it means operations can answer why a change happened. Ask vendors which meaning they are selling.",[344,1529,1531],{"id":1530},"how-should-we-talk-to-the-board","How should we talk to the board?",[174,1533,1534],{},"Separate “we can reconstruct what we did” from “we can estimate market lift.” The first is a control. The second is a study.",[344,1536,1538],{"id":1537},"where-does-the-wiki-fit","Where does the wiki fit?",[174,1540,1541,1542,185],{},"Asserted policy is often the parent of a refusal or a draft. “Which playbook version caused this flag?” is a causal-operations question. See ",[190,1543,614],{"href":261},[344,1545,1547],{"id":1546},"how-is-this-different-from-audit-logging","How is this different from audit logging?",[174,1549,1550],{},"Audit logs are often thin events. Causal operations needs the join: job, sources, proposal, person, system response. A log that cannot join is a pile.",[221,1552,609],{"id":608},[174,1554,1555,471,1557,474,1560,185],{},[190,1556,1244],{"href":603},[190,1558,1559],{"href":432},"What is an AI workstream",[190,1561,617],{"href":455},[221,1563,621],{"id":620},[226,1565,1566,1572,1578],{},[229,1567,1568],{},[190,1569,1571],{"href":1191,"rel":1570},[194],"Stanford Encyclopedia of Philosophy, Causal Models",[229,1573,1574],{},[190,1575,1577],{"href":1197,"rel":1576},[194],"Pearl, Causal inference in statistics: An overview",[229,1579,1580],{},[190,1581,1583],{"href":1253,"rel":1582},[194],"W3C PROV overview",{"title":155,"searchDepth":156,"depth":156,"links":1585},[1586,1587,1591,1592,1606,1607],{"id":223,"depth":156,"text":224},{"id":303,"depth":156,"text":304,"children":1588},[1589,1590],{"id":346,"depth":657,"text":347},{"id":385,"depth":657,"text":386},{"id":444,"depth":156,"text":445},{"id":479,"depth":156,"text":480,"children":1593},[1594,1595,1596,1597,1598,1599,1600,1601,1602,1603,1604,1605],{"id":1464,"depth":657,"text":1465},{"id":1471,"depth":657,"text":1472},{"id":1478,"depth":657,"text":1479},{"id":1485,"depth":657,"text":1486},{"id":1492,"depth":657,"text":1493},{"id":1499,"depth":657,"text":1500},{"id":1509,"depth":657,"text":1510},{"id":1516,"depth":657,"text":1517},{"id":1523,"depth":657,"text":1524},{"id":1530,"depth":657,"text":1531},{"id":1537,"depth":657,"text":1538},{"id":1546,"depth":657,"text":1547},{"id":608,"depth":156,"text":609},{"id":620,"depth":156,"text":621},"Causal AI for operations means you can answer “why did this change happen?” with the actual steps and approval — not a guess that “the chatbot caused a lift.”","/blog/what-is-causal-ai-for-operations",{"title":1174,"description":1608},"blog/what-is-causal-ai-for-operations",[680,1613,1614,1615],"causal-ai","lifecycle-graph","operations","4CuAcrDUdv9JN5HzLCAh75Dgw_LrX9degw6PWZh30DA",{"enabled":149,"message":1618,"linkLabel":78,"linkHref":79,"id":1619,"title":1620,"archived":149,"authors":150,"badge":150,"body":1621,"date":150,"department":150,"description":155,"extension":158,"eyebrow":150,"faqHeader":150,"faqs":150,"footerBand":150,"headline":150,"image":150,"industry":150,"jobType":150,"listed":149,"location":150,"navigation":115,"openRoles":150,"pageLayout":150,"path":1625,"relatedHeading":150,"seo":1626,"series":150,"sitemap":115,"status":150,"stem":1627,"subhead":150,"tags":150,"video":150,"whyJoin":150,"workplaceType":150,"__hash__":1628},"We're hiring! Join the team building the Sentient Enterprise.","content/shared/hiring.md","Hiring banner",{"type":152,"value":1622,"toc":1623},[],{"title":155,"searchDepth":156,"depth":156,"links":1624},[],"/shared/hiring",{"title":1620,"description":155},"shared/hiring","-6bioYD7lKYokGUVU3ff4hHTvB-sDyOMuCptKHnojfk",{"fold":1630,"id":1634,"title":1635,"archived":149,"authors":150,"badge":150,"body":1636,"date":150,"department":150,"description":155,"extension":158,"eyebrow":150,"faqHeader":150,"faqs":150,"footerBand":1640,"headline":150,"image":150,"industry":150,"jobType":150,"listed":149,"location":150,"navigation":115,"openRoles":150,"pageLayout":150,"path":1644,"relatedHeading":150,"seo":1645,"series":150,"sitemap":115,"status":150,"stem":1646,"subhead":150,"tags":150,"video":150,"whyJoin":150,"workplaceType":150,"__hash__":1647},{"headline":1631,"description":1632,"primaryLabel":8,"primaryTo":1633,"secondaryLabel":703,"secondaryTo":12},"Run frontier AI your business actually owns.","Governed agent swarms, 2,000+ integrations, and a knowledge graph that stays inside your walls. Free 7-day trial.","/checkout","content/shared/cta.md","Site CTAs",{"type":152,"value":1637,"toc":1638},[],{"title":155,"searchDepth":156,"depth":156,"links":1639},[],{"headline":1641,"description":1642,"primaryLabel":8,"primaryTo":1633,"secondaryLabel":1643,"secondaryTo":84},"See what governed AI looks like on your stack.","Connect your tools, run a workstream, and keep every decision on your ledger - free for 7 days.","Talk to our team","/shared/cta",{"title":1635,"description":155},"shared/cta","YHK6Fb8AvCPR1zZq7R_xiXUG0hwhP5UxHA8Ix52JQp4",1787194073857]