[{"data":1,"prerenderedAt":1655},["ShallowReactive",2],{"site-nav-content":3,"hiring-banner-content":163,"blog:/blog/what-is-institutional-memory-in-enterprise-ai":175,"blog-index-copy":678,"blog:/blog/what-is-institutional-memory-in-enterprise-ai:surround":699,"site-cta-content":1636},{"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",{"enabled":149,"message":164,"linkLabel":78,"linkHref":79,"id":165,"title":166,"archived":149,"authors":150,"badge":150,"body":167,"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":171,"relatedHeading":150,"seo":172,"series":150,"sitemap":115,"status":150,"stem":173,"subhead":150,"tags":150,"video":150,"whyJoin":150,"workplaceType":150,"__hash__":174},"We're hiring! Join the team building the Sentient Enterprise.","content/shared/hiring.md","Hiring banner",{"type":152,"value":168,"toc":169},[],{"title":155,"searchDepth":156,"depth":156,"links":170},[],"/shared/hiring",{"title":166,"description":155},"shared/hiring","-6bioYD7lKYokGUVU3ff4hHTvB-sDyOMuCptKHnojfk",{"id":176,"title":177,"archived":149,"authors":178,"badge":181,"body":183,"date":667,"department":150,"description":668,"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":669,"relatedHeading":150,"seo":670,"series":671,"sitemap":115,"status":150,"stem":672,"subhead":150,"tags":673,"video":150,"whyJoin":150,"workplaceType":150,"__hash__":677},"content/blog/what-is-institutional-memory-in-enterprise-ai.md","What is Institutional Memory in Enterprise AI",[179],{"name":180,"to":120},"Nimbus Research",{"label":182},"Explainer",{"type":152,"value":184,"toc":642},[185,189,192,195,206,211,214,275,278,281,307,314,318,321,324,344,347,352,363,375,381,387,397,401,407,413,419,429,435,441,444,465,478,481,485,488,495,513,517,521,524,528,531,535,540,544,547,551,557,561,564,568,571,575,578,582,585,589,592,596,599,603,606,610,619,623],[186,187,188],"p",{},"Institutional memory is what the company still knows after the person who did the work leaves — after the chat vendor changes, after the model version rolls.",[186,190,191],{},"Individual memory is a hallway conversation and a personal ChatGPT thread. Company memory is playbooks, signed decisions, and live systems, with access control.",[186,193,194],{},"Organisations have always had memory: filing cabinets, shared drives, ERP history, “ask the person who was here last year.” Generative AI created a new amnesia: high-value reasoning happens in disposable threads, on personal accounts, in tools with the wrong retention, or in a vendor’s silo the company cannot query.",[186,196,197,198,205],{},"This is an evidence topic, not a nostalgia topic. Financial reporting changes have needed reconstructable authorisation for decades. Records-management programmes ask for metadata and assigned responsibility. None of those regimes is satisfied by a personal chat thread the predecessor took with them. ",[199,200,204],"a",{"href":201,"rel":202},"https://ico.org.uk/for-organisations/uk-gdpr-guidance-and-resources/artificial-intelligence/",[203],"nofollow","UK ICO guidance on AI and data protection"," still wants purpose and retention thinking when the “user” is a model.",[207,208,210],"h2",{"id":209},"words-youll-hear","Words you’ll hear",[186,212,213],{},"Keep four kinds of memory separate on purpose:",[215,216,217,235,245,260],"ul",{},[218,219,220,224,225,229,230,234],"li",{},[221,222,223],"strong",{},"Asserted policy."," What we ",[226,227,228],"em",{},"want"," to be true: playbooks, guardrails, approved language. See ",[199,231,233],{"href":232},"what-is-a-company-wiki-for-ai-agents","What is a company wiki for AI agents",". At work, this is the current discount floor, not last year’s slide.",[218,236,237,240,241,244],{},[221,238,239],{},"Systems of record."," What ",[226,242,243],{},"is"," true in operations: CRM, ERP, HR, the warehouse. These are memory of the business, not of AI work. At work, the opportunity Amount is here. The reason it changed may not be.",[218,246,247,250,251,254,255,259],{},[221,248,249],{},"Decision memory."," Why we ",[226,252,253],{},"changed"," something with AI in the loop: briefs, approvals, rejected options, source versions. A ",[199,256,258],{"href":257},"what-is-a-lifecycle-graph","lifecycle graph",". At work, this is “who signed this exception, against which playbook version.”",[218,261,262,265,266,269,270,274],{},[221,263,264],{},"Retrieved knowledge."," Documents we ",[226,267,268],{},"might"," use. That is ",[199,271,273],{"href":272},"what-is-enterprise-rag","enterprise RAG",". Lookup without policy and decisions is a search engine, not memory.",[186,276,277],{},"If you collapse all four into “one vector store” — a database of text fingerprints used to find similar documents — you get sludge that cannot tell policy from a brainstorm. You also get a new store of sensitive data.",[186,279,280],{},"Other terms:",[215,282,283,289,295,301],{},[218,284,285,288],{},[221,286,287],{},"Hallway knowledge."," The unofficial version of the rule. It leaves with people. Agents will invent a cousin if it is not asserted.",[218,290,291,294],{},[221,292,293],{},"Provider logs."," The vendor’s artefact. Not scoped to your jobs, not your access-controlled ledger.",[218,296,297,300],{},[221,298,299],{},"Retention."," How long a class of record is kept. Completeness is reconstructability, not hoarding.",[218,302,303,306],{},[221,304,305],{},"Perception."," Asking that memory in ordinary language, with permissions still applied.",[186,308,309,313],{},[199,310,312],{"href":311},"what-is-causal-ai-for-operations","Causal operations"," is the “why did this change?” slice of decision memory. It is not a claim about market lift.",[207,315,317],{"id":316},"why-you-should-care","Why you should care",[186,319,320],{},"It affects you the first Monday after someone leaves, and the first time an auditor asks “why is this exception in the CRM when the playbook still says otherwise?”",[186,322,323],{},"Three verbs:",[215,325,326,332,338],{},[218,327,328,331],{},[221,329,330],{},"Assert."," Put the rule into a controlled surface. If it only lives in a slide, agents will invent a cousin.",[218,333,334,337],{},[221,335,336],{},"Record."," Store the decision chain when AI is in the loop — not every token, the links that let you reconstruct a change.",[218,339,340,343],{},[221,341,342],{},"Ask."," Let the next operator query that memory in ordinary language, with permissions still applied.",[186,345,346],{},"Causal operations questions (“why did this change?”) need decision memory. Remember outcomes, quotes, approvals, and citations — not every failed token. Wiki needs owners; memory without freshness is last year’s discount floor.",[348,349,351],"h3",{"id":350},"what-changes-by-role","What changes by role",[186,353,354,357,358,362],{},[221,355,356],{},"Finance."," Close packs inherit exceptions. Finance needs the playbook version and the signer, not a rumour that “we always accrue this way.” Provider ChatGPT exports are not a SOX-style trail. Spend history also belongs in memory: which job consumed the units, which run stopped on a cap. See ",[199,359,361],{"href":360},"what-is-ai-token-economics","What is AI token economics",".",[186,364,365,368,369,374],{},[221,366,367],{},"Legal."," Discovery, customer commitments, and erasure. Legal should insist that decision memory points at systems of record rather than duplicating them, and that retention is typed. Infinite chat fails a privacy review. ",[199,370,373],{"href":371,"rel":372},"https://eur-lex.europa.eu/eli/reg/2016/679/oj",[203],"GDPR"," erasure is harder if you indexed everything into sludge.",[186,376,377,380],{},[221,378,379],{},"Operations."," Handoffs. The next shift should query “why did this pause?” without reconstructing Slack. Ops should refuse a design that stores every token “because AI” and then cannot delete it.",[186,382,383,386],{},[221,384,385],{},"Go-to-market."," Win/loss reasons and discount exceptions walk out the door with account owners. GTM should put asserted playbooks in the wiki and signed exceptions on the graph — not in a personal Claude project.",[186,388,389,392,393,362],{},[221,390,391],{},"Security."," Memory is a sensitive store. Access control on the graph and wiki is as important as on the CRM. Shadow AI is amnesia by design: the work happened on an account the company cannot query. See ",[199,394,396],{"href":395},"what-is-shadow-ai","What is shadow AI",[348,398,400],{"id":399},"what-people-get-wrong","What people get wrong",[186,402,403,406],{},[221,404,405],{},"CRM as sufficient memory."," CRM remembers the current field. It does not remember which playbook version, which AI run, or which person signed the exception.",[186,408,409,412],{},[221,410,411],{},"Exporting ChatGPT threads."," Vendor artefact. Wrong scope. Wrong access control.",[186,414,415,418],{},[221,416,417],{},"One vector store for everything."," Policy, brainstorms, tickets, and decisions become an undifferentiated similarity soup.",[186,420,421,424,425,428],{},[221,422,423],{},"A business knowledge graph as a substitute."," That graph models customers and products. Institutional memory for AI work models ",[221,426,427],{},"what we did with models"," — and why.",[186,430,431,434],{},[221,432,433],{},"Keeping everything forever."," Hoarding is not completeness. It is a privacy and cost failure.",[186,436,437,440],{},[221,438,439],{},"Remembering every token."," Reconstruct the change. Do not archive the model’s scratch reasoning by default.",[186,442,443],{},"Good looks like four layers kept apart, owners on wiki pages, a lifecycle graph of decisions, permissions on ask, typed retention, and pointers to live systems. Failure looks like a personal thread, a vendor log, and a vector lake.",[186,445,446,447,449,450,453,454,456,457,460,461,464],{},"Adjacent concepts: ",[199,448,273],{"href":272}," is lookup, not memory of what we decided. A ",[199,451,452],{"href":232},"company wiki"," is asserted policy, which goes stale without owners. A ",[199,455,258],{"href":257}," is decision memory of AI-mediated work. ",[199,458,459],{"href":311},"Causal AI for operations"," is the “why did this change?” question that memory should be able to answer. ",[199,462,463],{"href":395},"Shadow AI"," is how memory never starts.",[186,466,467,468,471,472,477],{},"Do not confuse this with a second CRM. Point at the opportunity; do not copy the pipeline. Copies become conflicting official numbers and an erasure problem under ",[199,469,373],{"href":371,"rel":470},[203],". The ",[199,473,476],{"href":474,"rel":475},"https://www.w3.org/TR/prov-overview/",[203],"W3C PROV"," idea — entities, activities, agents — is the right instinct for the decision layer: enough structure to reconstruct, not a lake of tokens.",[186,479,480],{},"A Monday-morning test is enough. Can the next operator, with the right permissions, find the playbook version, the signed exception, and the live field — without the predecessor’s laptop? If the answer depends on a personal chat vendor, you do not have institutional memory. You have a coincidence that the person has not left yet.",[207,482,484],{"id":483},"how-this-shows-up-in-nimbus","How this shows up in Nimbus",[186,486,487],{},"Nimbus combines wiki (asserted policy), connectors (systems of record), the Lifecycle Graph (decision memory), Perception (ask), and workstream scoping (who may see what).",[186,489,490,491,494],{},"Connectors default to read-only, so analysis can be remembered as ",[226,492,493],{},"not"," having written. Named signers and fail-closed writes make refusals part of memory, not missing events.",[186,496,497,498,500,501,504,505,508,509,362],{},"Product: ",[199,499,23],{"href":24},", ",[199,502,503],{"href":28},"Wiki",", and ",[199,506,507],{"href":36},"Perception",". The job boundary is a ",[199,510,512],{"href":511},"what-is-an-ai-workstream","workstream",[207,514,516],{"id":515},"questions-people-actually-ask","Questions people actually ask",[348,518,520],{"id":519},"isnt-crm-already-our-memory","Isn’t CRM already our memory?",[186,522,523],{},"CRM remembers the current field. It does not remember which playbook version, which AI run, or which person signed the exception.",[348,525,527],{"id":526},"can-we-just-export-chatgpt-threads","Can we just export ChatGPT threads?",[186,529,530],{},"Provider logs are the vendor’s artefact. They are not scoped to your jobs, and they are not your access-controlled ledger.",[348,532,534],{"id":533},"how-is-this-different-from-a-knowledge-graph-of-customers-and-products","How is this different from a knowledge graph of customers and products?",[186,536,537,538,428],{},"That graph models the business domain. Institutional memory for AI work models ",[221,539,427],{},[348,541,543],{"id":542},"does-this-mean-storing-everything-forever","Does this mean storing everything forever?",[186,545,546],{},"No. Retention follows the type of record. Completeness is reconstructability, not hoarding.",[348,548,550],{"id":549},"how-is-this-different-from-enterprise-rag","How is this different from enterprise RAG?",[186,552,553,554,362],{},"RAG retrieves what exists. Memory of work is what we asserted, what we decided, and what the live system holds. Retrieval without those layers is search. See ",[199,555,556],{"href":272},"What is enterprise RAG",[348,558,560],{"id":559},"what-should-we-remember-from-a-run","What should we remember from a run?",[186,562,563],{},"The brief, sources (including wiki version), quoted payload, named signer, live-system result, and spend stop if any. Not every failed token, and not secrets in transcripts by default.",[348,565,567],{"id":566},"how-do-we-stop-last-years-policy-living-forever","How do we stop last year’s policy living forever?",[186,569,570],{},"Owners and review cadence on the wiki. Archives must not win retrieval against current policy. Freshness is part of memory, not a nice-to-have.",[348,572,574],{"id":573},"can-perception-see-other-departments-decisions","Can Perception see other departments’ decisions?",[186,576,577],{},"Only with the same least privilege as the workstream. A go-to-market question should not surface People Ops briefs.",[348,579,581],{"id":580},"is-hallway-knowledge-always-bad","Is hallway knowledge always bad?",[186,583,584],{},"It is how work actually happens until you assert it. The failure is leaving it only in hallways once agents are in the loop.",[348,586,588],{"id":587},"how-does-switching-model-vendors-affect-memory","How does switching model vendors affect memory?",[186,590,591],{},"If memory lived in the vendor’s chat product, you lost it. If it lived in your wiki, graph, and systems of record, you kept it. That is a buying criterion.",[348,593,595],{"id":594},"where-does-shadow-ai-fit","Where does shadow AI fit?",[186,597,598],{},"Personal accounts are institutional amnesia: the company cannot assert, record, or ask. Substitution onto a governed path is how memory starts.",[348,600,602],{"id":601},"do-we-need-a-data-team-to-ask-the-memory","Do we need a data team to ask the memory?",[186,604,605],{},"Not if the product has an ordinary-language query surface over the graph and wiki, with permissions. That is Perception in Nimbus. A data team is still right for warehouse metrics.",[207,607,609],{"id":608},"related-reading","Related reading",[186,611,612,500,615,504,617,362],{},[199,613,614],{"href":257},"What is a lifecycle graph",[199,616,556],{"href":272},[199,618,233],{"href":232},[207,620,622],{"id":621},"sources","Sources",[215,624,625,631,636],{},[218,626,627],{},[199,628,630],{"href":201,"rel":629},[203],"ICO, AI and data protection",[218,632,633],{},[199,634,373],{"href":371,"rel":635},[203],[218,637,638],{},[199,639,641],{"href":474,"rel":640},[203],"W3C PROV overview",{"title":155,"searchDepth":156,"depth":156,"links":643},[644,645,650,651,665,666],{"id":209,"depth":156,"text":210},{"id":316,"depth":156,"text":317,"children":646},[647,649],{"id":350,"depth":648,"text":351},3,{"id":399,"depth":648,"text":400},{"id":483,"depth":156,"text":484},{"id":515,"depth":156,"text":516,"children":652},[653,654,655,656,657,658,659,660,661,662,663,664],{"id":519,"depth":648,"text":520},{"id":526,"depth":648,"text":527},{"id":533,"depth":648,"text":534},{"id":542,"depth":648,"text":543},{"id":549,"depth":648,"text":550},{"id":559,"depth":648,"text":560},{"id":566,"depth":648,"text":567},{"id":573,"depth":648,"text":574},{"id":580,"depth":648,"text":581},{"id":587,"depth":648,"text":588},{"id":594,"depth":648,"text":595},{"id":601,"depth":648,"text":602},{"id":608,"depth":156,"text":609},{"id":621,"depth":156,"text":622},"2026-08-17","Institutional memory is what the company still knows after the person who did the work leaves: official playbooks, signed decisions, and live systems — with access control.","/blog/what-is-institutional-memory-in-enterprise-ai",{"title":177,"description":668},"explainer","blog/what-is-institutional-memory-in-enterprise-ai",[671,674,675,676],"institutional-memory","lifecycle-graph","wiki","QKskx6GjKfWSaN9E9kbMHU2XCu29nlgWTKO656jg8aI",{"hero":679,"id":681,"title":682,"archived":149,"authors":150,"badge":150,"body":683,"date":150,"department":150,"description":687,"extension":158,"eyebrow":688,"faqHeader":150,"faqs":150,"footerBand":689,"headline":150,"image":150,"industry":150,"jobType":150,"listed":149,"location":150,"navigation":115,"openRoles":150,"pageLayout":150,"path":60,"relatedHeading":695,"seo":696,"series":150,"sitemap":115,"status":150,"stem":697,"subhead":150,"tags":150,"video":150,"whyJoin":150,"workplaceType":150,"__hash__":698},{"filename":680},"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":684,"toc":685},[],{"title":155,"searchDepth":156,"depth":156,"links":686},[],"Deep dives into pre-cognitive intelligence, sentient enterprises, and the evolving landscape of AI-driven business transformation.","Latest Research",{"headline":690,"description":691,"primaryLabel":692,"primaryTo":693,"secondaryLabel":694,"secondaryTo":12},"Stay at the frontier.","Subscribe for product updates and new insights.","Subscribe","/newsletter","Explore the platform","More research",{"title":682,"description":687},"blog/index","eK1RCXdDW8nfLSyKRXGB1mJm9FAmhAO6GWXwNKOMNVE",[700,1178],{"id":701,"title":702,"archived":149,"authors":703,"badge":705,"body":706,"date":667,"department":150,"description":1170,"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":1171,"relatedHeading":150,"seo":1172,"series":671,"sitemap":115,"status":150,"stem":1173,"subhead":150,"tags":1174,"video":150,"whyJoin":150,"workplaceType":150,"__hash__":1177},"content/blog/what-is-model-context-protocol.md","What is Model Context Protocol",[704],{"name":180,"to":120},{"label":182},{"type":152,"value":707,"toc":1146},[708,711,735,742,751,754,756,807,818,820,823,829,850,853,867,870,872,881,886,891,896,907,909,915,921,927,937,945,954,957,978,985,987,996,999,1007,1009,1013,1016,1020,1023,1027,1033,1037,1040,1044,1047,1051,1054,1058,1061,1065,1068,1072,1077,1081,1086,1090,1098,1102,1105,1107,1114,1116],[186,709,710],{},"USB did not create a data-governance programme. A common plug let keyboards, cameras, and drives talk to any computer. It did not decide who may copy the finance drive, or whether a change to the ledger needs a signer.",[186,712,713,716,717,722,723,728,729,734],{},[221,714,715],{},"Model Context Protocol (MCP)"," is the same kind of open standard for AI. ",[199,718,721],{"href":719,"rel":720},"https://www.anthropic.com/news/model-context-protocol",[203],"Anthropic announced it"," as a ",[199,724,727],{"href":725,"rel":726},"https://modelcontextprotocol.io/docs/2026-07-28/getting-started/intro",[203],"common plug"," so AI apps can use the same tools and files, instead of every vendor inventing a one-off connection. The ",[199,730,733],{"href":731,"rel":732},"https://modelcontextprotocol.io/specification/2025-11-25/index",[203],"specification"," standardises how a host calls tools and reads resources.",[186,736,737,738,741],{},"In one sentence: MCP is ",[221,739,740],{},"plumbing, not a company strategy",". It does not decide who may update Salesforce.",[186,743,744,745,750],{},"Developers already know this pattern from the ",[199,746,749],{"href":747,"rel":748},"https://microsoft.github.io/language-server-protocol/",[203],"Language Server Protocol",": one language server, many editors, instead of rewriting autocomplete for every IDE. MCP is that idea for tools an AI can call. LSP made language servers interchangeable. It did not make every language server a safe place for customer lists.",[186,752,753],{},"Before this standard, every AI product invented its own way to “use a tool.” Teams spent months redoing the same wiring. That cost was real. So is the over-read: “we support MCP” is not “we have enterprise governance.”",[207,755,210],{"id":209},[215,757,758,764,770,776,782,788,798],{},[218,759,760,763],{},[221,761,762],{},"Protocol / standard."," Agreed wiring so products can interoperate. At work, this is the USB cable, not the access-control list on the share.",[218,765,766,769],{},[221,767,768],{},"MCP server / helper."," A small programme that says “here are the actions I can take, and here are the files I can show you.” At work, a helper that searches Drive as a superuser is still a superuser.",[218,771,772,775],{},[221,773,774],{},"Host / client."," The AI application that calls the helper. At work, several hosts can speak MCP and still have completely different write gates — or none.",[218,777,778,781],{},[221,779,780],{},"Tool call."," The AI asking that helper to search a folder, look up a ticket, post a message, or query a database.",[218,783,784,787],{},[221,785,786],{},"Resource."," A file or record the helper can expose for reading.",[218,789,790,793,794,362],{},[221,791,792],{},"Connector (Nimbus)."," A supported, company-controlled integration to a live system — OAuth, scoped to the job, read-only by default. That is the operator-facing story. MCP may sit at a developer edge. It is not a substitute for connectors plus ",[199,795,797],{"href":796},"what-is-ai-governance","governance",[218,799,800,803,804,806],{},[221,801,802],{},"Least privilege."," Which tools this ",[199,805,512],{"href":511}," may call. Importing every available helper is how a demo becomes one actor with every production login.",[186,808,809,810,813,814,362],{},"A tool call can still change production data. The protocol will happily pass that change along. The company still has to decide whether that is allowed. Fail-closed writes, named signers, and quoted payloads live ",[226,811,812],{},"above"," the plug. See ",[199,815,817],{"href":816},"what-is-write-back-governance","What is write-back governance",[207,819,317],{"id":316},[186,821,822],{},"The plug is useful. It is also easy to over-read.",[186,824,825,826,828],{},"MCP does ",[221,827,493],{}," decide:",[215,830,831,834,837,840,843],{},[218,832,833],{},"whose login is used",[218,835,836],{},"whether the AI may only read, or also change a live system",[218,838,839],{},"who must approve a change",[218,841,842],{},"how the company remembers what happened",[218,844,845,846],{},"which model is used for the step — see ",[199,847,849],{"href":848},"what-is-model-routing","What is model routing",[186,851,852],{},"Choosing a model is choosing a brain. This standard is choosing hands. A cheap model with dangerous tools is worse than a strong model with none. Decide them separately.",[186,854,855,856,860,861,866],{},"An ",[199,857,859],{"href":858},"what-is-an-agentic-workflow","agentic workflow"," that imports every available tool is a confused workflow. Plumbing is not a stop condition. ",[199,862,865],{"href":863,"rel":864},"https://www.anthropic.com/engineering/building-effective-agents",[203],"Anthropic’s note on building effective agents"," is about bounding tools and stops, not about collecting helpers.",[186,868,869],{},"It affects you if a vendor says “we support MCP” and you hear “we have enterprise governance.” Those are different sentences.",[348,871,351],{"id":350},[186,873,874,876,877,880],{},[221,875,356],{}," A helper that can post a journal is a write path, protocol or not. Finance should ask whether the host quotes the payload and requires a named signer, not whether the wiring is MCP. Spend also sits above the plug: tool loops can burn ",[199,878,879],{"href":360},"NTUs"," without a ceiling.",[186,882,883,885],{},[221,884,367],{}," Processing agreements, purpose, and customer data in helpers running on laptops. Legal should not treat “open standard” as “safe.” A standard plug does not create a DPIA.",[186,887,888,890],{},[221,889,379],{}," Bounded tool belts per job. Ops should refuse workflows that attach every helper “for flexibility,” and should keep human wait and budget as stops regardless of how tools are wired.",[186,892,893,895],{},[221,894,385],{}," Faster wiring to CRM and Drive can be good — if the connector is still read-only by default. GTM should not confuse a demo that updates an opportunity via MCP with a governed release.",[186,897,898,900,901,906],{},[221,899,391],{}," This is the sharp edge. Helpers run with some identity. Superuser search is still superuser search. Prompt injection can trick a model into requesting a tool call; the ",[199,902,905],{"href":903,"rel":904},"https://genai.owasp.org/llm-top-10/",[203],"OWASP Top 10 for LLM applications"," is the relevant list. The protocol will not save you. Least privilege, read-only defaults, and fail-closed writes will.",[348,908,400],{"id":399},[186,910,911,914],{},[221,912,913],{},"MCP as governance."," Wiring is not a named signer.",[186,916,917,920],{},[221,918,919],{},"MCP as the Salesforce strategy."," You still need identity, read versus write, an approver, and a record.",[186,922,923,926],{},[221,924,925],{},"Refusing products that do not speak MCP."," Interoperable tools are a plus. Absence of MCP is not absence of a connector. Presence of MCP is not presence of governance.",[186,928,929,932,933,936],{},[221,930,931],{},"Replacing the integration platform."," MCP standardises how an AI ",[226,934,935],{},"talks"," to a helper. Your identity, iPaaS, and change-control stack still have to exist.",[186,938,939,942,943,362],{},[221,940,941],{},"Assuming retrieval will respect permissions."," Only if the helper is built that way. See ",[199,944,556],{"href":272},[186,946,947,950,951,953],{},[221,948,949],{},"Collecting every server."," A large tool belt is a confused ",[199,952,859],{"href":858}," and a larger attack surface.",[186,955,956],{},"Good looks like: MCP where it reduces duplicate wiring; operator-facing connectors that stay scoped, encrypted, and read-only by default; writes only after sign-off; no belief that the spec implemented your control framework. Failure looks like a laptop running a superuser helper pointed at production.",[186,958,959,960,963,964,966,967,970,971,973,974,977],{},"Think of the stack in layers, or you will buy the wrong layer. MCP is how a host talks to a helper. A ",[199,961,962],{"href":51},"connector"," is how operators attach a live system to a ",[199,965,512],{"href":511}," with OAuth and a read-only default. ",[199,968,969],{"href":816},"Write-back governance"," is whether a tool call that mutates production is allowed to execute. The ",[199,972,258],{"href":257}," is whether you can still explain the call next quarter. ",[199,975,976],{"href":848},"Model routing"," is which brain issued the call. None of those jobs is in the spec, and that is fine — specs should stay thin. Trouble starts when a thin spec is sold as the thick programme.",[186,979,855,980,984],{},[199,981,983],{"href":982},"what-is-an-enterprise-ai-operating-system","enterprise AI operating system"," sits above plumbing the way an OS sits above USB: isolation, permissions, I/O policy, and state. USB made accessories interchangeable. It did not decide who may format the finance drive.",[207,986,484],{"id":483},[186,988,989,990,993,994,362],{},"Nimbus’s operator-facing integrations are ",[221,991,992],{},"connectors",": scoped per workstream, encrypted per tenant, read-only by default. Action connectors write only after human sign-off. See ",[199,995,50],{"href":51},[186,997,998],{},"MCP can be useful at developer edges. It is not the product’s answer to “who may change CRM.” Governance, wiki, and the Lifecycle Graph still sit above any plug.",[186,1000,1001,1002,1004,1005,362],{},"See ",[199,1003,39],{"href":40}," and ",[199,1006,31],{"href":32},[207,1008,516],{"id":515},[348,1010,1012],{"id":1011},"is-mcp-how-we-should-connect-salesforce","Is MCP how we should connect Salesforce?",[186,1014,1015],{},"Not by itself. You still need identity, read vs write rights, an approver, and a record. A standard plug does not provide those.",[348,1017,1019],{"id":1018},"should-we-refuse-products-that-dont-speak-mcp","Should we refuse products that don’t speak MCP?",[186,1021,1022],{},"No. Interoperable tools are a plus. Absence of MCP is not absence of a connector. Presence of MCP is not presence of governance.",[348,1024,1026],{"id":1025},"does-mcp-replace-our-integration-platform","Does MCP replace our integration platform?",[186,1028,1029,1030,1032],{},"No. It standardises how an AI ",[226,1031,935],{}," to a helper. Your integration, identity, and change-control stack still has to exist.",[348,1034,1036],{"id":1035},"can-mcp-make-retrieval-respect-permissions","Can MCP make retrieval respect permissions?",[186,1038,1039],{},"Only if the helper is built that way. A tool that searches Drive as a superuser is still a superuser.",[348,1041,1043],{"id":1042},"is-mcp-the-same-as-a-nimbus-connector","Is MCP the same as a Nimbus connector?",[186,1045,1046],{},"No. A connector is the operator-facing, company-controlled integration: OAuth, workstream scope, read-only default. MCP is a developer wiring standard that might sit at an edge.",[348,1048,1050],{"id":1049},"does-the-spec-require-fail-closed-writes","Does the spec require fail-closed writes?",[186,1052,1053],{},"No. The spec does not require a quoted Salesforce payload, a named approver, or a fail-closed write. Those are product and policy choices.",[348,1055,1057],{"id":1056},"how-does-this-relate-to-usb-and-lsp","How does this relate to USB and LSP?",[186,1059,1060],{},"USB and LSP are the right analogies: interoperability of accessories and language servers. Neither is an access-control programme. Do not buy MCP as if it were.",[348,1062,1064],{"id":1063},"can-we-let-every-agent-team-install-their-own-mcp-servers","Can we let every agent team install their own MCP servers?",[186,1066,1067],{},"That is how you get overlapping write rights and no inventory. Treat helpers like production integrations: owners, scope, and a default of read-only.",[348,1069,1071],{"id":1070},"does-mcp-choose-the-model","Does MCP choose the model?",[186,1073,1074,1075,362],{},"No. Routing is which brain you pay for. MCP is which hands that brain can use. Decide them separately. See ",[199,1076,849],{"href":848},[348,1078,1080],{"id":1079},"is-we-support-mcp-a-good-rfp-answer-for-governance","Is “we support MCP” a good RFP answer for governance?",[186,1082,1083,1084,362],{},"It is a good answer for tool interoperability. For governance, ask about quotes, named signers, workstream scope, and the ",[199,1085,258],{"href":257},[348,1087,1089],{"id":1088},"what-is-the-security-failure-mode","What is the security failure mode?",[186,1091,1092,1093,1097],{},"A helper with broad credentials, a host with no gate, and a model tricked into calling ",[1094,1095,1096],"code",{},"update_record",". The protocol did its job. Your company did not.",[348,1099,1101],{"id":1100},"should-customer-facing-bots-get-mcp-tools-to-internal-crm","Should customer-facing bots get MCP tools to internal CRM?",[186,1103,1104],{},"That is how a public conversation inherits production hands. Scope tools as tightly as you would scope a workstream — usually, do not.",[207,1106,609],{"id":608},[186,1108,1109,1004,1112,362],{},[199,1110,1111],{"href":858},"What is an agentic workflow",[199,1113,817],{"href":816},[207,1115,622],{"id":621},[215,1117,1118,1124,1130,1135,1140],{},[218,1119,1120],{},[199,1121,1123],{"href":719,"rel":1122},[203],"Anthropic, Introducing the Model Context Protocol",[218,1125,1126],{},[199,1127,1129],{"href":731,"rel":1128},[203],"Model Context Protocol specification",[218,1131,1132],{},[199,1133,749],{"href":747,"rel":1134},[203],[218,1136,1137],{},[199,1138,905],{"href":903,"rel":1139},[203],[218,1141,1142],{},[199,1143,1145],{"href":863,"rel":1144},[203],"Anthropic, Building effective agents",{"title":155,"searchDepth":156,"depth":156,"links":1147},[1148,1149,1153,1154,1168,1169],{"id":209,"depth":156,"text":210},{"id":316,"depth":156,"text":317,"children":1150},[1151,1152],{"id":350,"depth":648,"text":351},{"id":399,"depth":648,"text":400},{"id":483,"depth":156,"text":484},{"id":515,"depth":156,"text":516,"children":1155},[1156,1157,1158,1159,1160,1161,1162,1163,1164,1165,1166,1167],{"id":1011,"depth":648,"text":1012},{"id":1018,"depth":648,"text":1019},{"id":1025,"depth":648,"text":1026},{"id":1035,"depth":648,"text":1036},{"id":1042,"depth":648,"text":1043},{"id":1049,"depth":648,"text":1050},{"id":1056,"depth":648,"text":1057},{"id":1063,"depth":648,"text":1064},{"id":1070,"depth":648,"text":1071},{"id":1079,"depth":648,"text":1080},{"id":1088,"depth":648,"text":1089},{"id":1100,"depth":648,"text":1101},{"id":608,"depth":156,"text":609},{"id":621,"depth":156,"text":622},"Model Context Protocol is a common plug so AI apps can use the same tools, like USB for accessories. It is plumbing, not a company strategy, and it does not decide who may update Salesforce.","/blog/what-is-model-context-protocol",{"title":702,"description":1170},"blog/what-is-model-context-protocol",[671,1175,992,1176],"mcp","tools","pyCsyLW6Bee4HSZZGalnpk2Ef8Ouo6MpiGQykRPKqaI",{"id":1179,"title":1180,"archived":149,"authors":1181,"badge":1183,"body":1184,"date":667,"department":150,"description":1628,"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":1629,"relatedHeading":150,"seo":1630,"series":671,"sitemap":115,"status":150,"stem":1631,"subhead":150,"tags":1632,"video":150,"whyJoin":150,"workplaceType":150,"__hash__":1635},"content/blog/what-is-human-in-the-loop-ai.md","What is Human-in-the-Loop AI",[1182],{"name":180,"to":120},{"label":182},{"type":152,"value":1185,"toc":1604},[1186,1192,1195,1210,1213,1216,1218,1286,1294,1296,1299,1302,1313,1320,1326,1333,1336,1338,1343,1348,1357,1362,1375,1377,1383,1388,1394,1400,1406,1412,1418,1430,1442,1445,1447,1450,1457,1463,1470,1472,1476,1479,1483,1486,1490,1493,1497,1500,1504,1507,1511,1518,1522,1528,1532,1535,1539,1542,1546,1549,1553,1556,1560,1563,1565,1576,1578],[186,1187,1188,1189,362],{},"Human-in-the-loop AI, in everyday language, means ",[221,1190,1191],{},"a person must approve before the AI can finish the job",[186,1193,1194],{},"Not “a human might read the chat.” Not a footer that says this content was generated. A gate the software cannot skip.",[186,1196,1197,1198,1203,1204,1209],{},"In February 2024, a British Columbia tribunal held Air Canada responsible for a chatbot that invented a bereavement-fare policy. ",[199,1199,1202],{"href":1200,"rel":1201},"https://www.cbc.ca/news/canada/british-columbia/air-canada-chatbot-lawsuit-1.7116416",[203],"CBC reported"," that the airline’s argument — the chatbot is a separate legal entity — failed. The decision is ",[199,1205,1208],{"href":1206,"rel":1207},"https://decisions.civilresolutionbc.ca/crt/crtd/en/525448/1/document.do",[203],"Moffatt v. Air Canada",". A customer relied on the invented fare. A human did not catch the fiction before it became a commitment.",[186,1211,1212],{},"That is the class of failure this article is about.",[186,1214,1215],{},"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.",[207,1217,210],{"id":209},[215,1219,1220,1226,1236,1242,1254,1260,1266,1272,1280],{},[218,1221,1222,1225],{},[221,1223,1224],{},"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.",[218,1227,1228,1231,1232,1235],{},[221,1229,1230],{},"On the loop."," The system runs; a human ",[226,1233,1234],{},"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.",[218,1237,1238,1241],{},[221,1239,1240],{},"Theatre."," A checkbox “I understand this is AI,” or a prompt that says “ask the user first,” while the model may still act.",[218,1243,1244,1247,1248,1253],{},[221,1245,1246],{},"Effective oversight."," ",[199,1249,1252],{"href":1250,"rel":1251},"https://eur-lex.europa.eu/eli/reg/2024/1689/oj",[203],"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.",[218,1255,1256,1259],{},[221,1257,1258],{},"Rubber stamp."," A gate that fires so often people auto-click. That is not oversight. It is fatigue.",[218,1261,1262,1265],{},[221,1263,1264],{},"Named signer."," Identity bound to the decision. Shared inboxes destroy this.",[218,1267,1268,1271],{},[221,1269,1270],{},"Quote / payload."," The exact change in the language of the live system — opportunity fields, journal lines, email body — not a wall of prompt text.",[218,1273,1274,1277,1278,362],{},[221,1275,1276],{},"Fail-closed."," Missing approval means nothing happens. See ",[199,1279,817],{"href":816},[218,1281,1282,1285],{},[221,1283,1284],{},"Maker-checker."," An older control: one person proposes, another authorises. HITL for AI is that instinct when the proposer is a model.",[186,1287,1288,1293],{},[199,1289,1292],{"href":1290,"rel":1291},"https://www.reuters.com/legal/new-york-lawyers-sanctioned-using-fake-chatgpt-cases-legal-brief-2023-06-22/",[203],"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.",[207,1295,317],{"id":316},[186,1297,1298],{},"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.",[186,1300,1301],{},"It affects you if AI can:",[215,1303,1304,1307,1310],{},[218,1305,1306],{},"change records or money",[218,1308,1309],{},"send a customer a message that asserts a policy, price, or term",[218,1311,1312],{},"affect employment, credit, or people’s rights",[186,1314,1315,1316,1319],{},"Place people where ",[221,1317,1318],{},"risk and reversibility"," change: before writes, before external messages, and at exception thresholds (amount, region, data class).",[186,1321,1322,1323,1325],{},"Do ",[221,1324,493],{}," put humans on every sentence. A gate that fires fifty times a day will be auto-clicked.",[186,1327,1328,1329,1332],{},"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 ",[226,1330,1331],{},"and"," legal emails is how you get a rubber stamp.",[186,1334,1335],{},"Show the change in the language of the live system. A person cannot oversee what they cannot parse.",[348,1337,351],{"id":350},[186,1339,1340,1342],{},[221,1341,356],{}," 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.",[186,1344,1345,1347],{},[221,1346,367],{}," 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.",[186,1349,1350,1352,1353,1356],{},[221,1351,379],{}," Place gates at reversibility boundaries. Ops should measure time-to-approved-write and reject rate, and should treat human wait as a first-class ",[199,1354,1355],{"href":858},"workflow"," step, not a Slack nudge.",[186,1358,1359,1361],{},[221,1360,385],{}," 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.",[186,1363,1364,1366,1367,1370,1371,1374],{},[221,1365,391],{}," The gate must be unskippable by the model, including after prompt injection. A jailbreak can trick the model into ",[226,1368,1369],{},"requesting"," a bad write. It should not be able to ",[226,1372,1373],{},"execute"," without a quote and a signer. Identity binding matters: a generic “approve” in a shared inbox is not a control.",[348,1376,400],{"id":399},[186,1378,1379,1382],{},[221,1380,1381],{},"On the loop as in the loop."," A kill switch is not a per-action signer.",[186,1384,1385,1387],{},[221,1386,1240],{}," Footers, checkboxes, and “shall I proceed?” in unbound chat.",[186,1389,1390,1393],{},[221,1391,1392],{},"Too many gates."," Fatigue produces rubber stamps. Fewer gates, better quotes.",[186,1395,1396,1399],{},[221,1397,1398],{},"Wrong human."," Whoever is online, or an AI champion spanning domains.",[186,1401,1402,1405],{},[221,1403,1404],{},"Chat as the quote."," Prompt text is not field-level change.",[186,1407,1408,1411],{},[221,1409,1410],{},"HITL as sufficient for the EU AI Act."," Oversight is necessary, not sufficient, for higher-risk systems.",[186,1413,1414,1415,1417],{},"Good looks like: read-only analysis without a click per sentence; quoted writes; named roles; fail-closed execution; rejects stored on the ",[199,1416,258],{"href":257},"; metrics on reject rates. Failure looks like a prompt, a footer, and a customer who relied on the bot.",[186,1419,1420,1421,1424,1425,1429],{},"The person should sit at ",[221,1422,1423],{},"release",", not at every internal hand-off between ",[199,1426,1428],{"href":1427},"what-is-multi-agent-ai","agent teams",". Internal critics can reduce garbage. They are not the signer.",[186,1431,1432,1433,1435,1436,1438,1439,1441],{},"Adjacent ideas are easy to mix. ",[199,1434,969],{"href":816}," is the fail-closed property of the write. HITL is the human act that satisfies it. A ",[199,1437,258],{"href":257}," is how you prove the act later. A ",[199,1440,512],{"href":511}," is whose job the gate belongs to. None of those is a footer on a chatbot.",[186,1443,1444],{},"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.",[207,1446,484],{"id":483},[186,1448,1449],{},"Read-only connectors mean the loop can analyse without a human per sentence.",[186,1451,1452,1453,1456],{},"When a write is proposed, governance ",[221,1454,1455],{},"quotes"," it and stops. Named roles must sign. Agent teams can draft. They cannot waive the gate. Missing approval is fail-closed.",[186,1458,1459,1460,1462],{},"The ",[199,1461,23],{"href":257}," stores the human act: who signed, what they saw, what happened next — including rejects. Perception can list rejected items.",[186,1464,1001,1465,1467,1468,362],{},[199,1466,39],{"href":40},". Companion: ",[199,1469,817],{"href":816},[207,1471,516],{"id":515},[348,1473,1475],{"id":1474},"isnt-this-just-slower-ai","Isn’t this just slower AI?",[186,1477,1478],{},"It is slower than ungoverned writes and faster than incident response. Invented policy is cheaper to catch in a quote than in a tribunal.",[348,1480,1482],{"id":1481},"who-should-be-the-human","Who should be the human?",[186,1484,1485],{},"The owner of the live-system change or the customer commitment, not “whoever is online.” Shared inboxes destroy accountability.",[348,1487,1489],{"id":1488},"does-a-person-at-the-gate-satisfy-eu-ai-law-by-itself","Does a person-at-the-gate satisfy EU AI law by itself?",[186,1491,1492],{},"No. Higher-risk systems have a bundle of duties. Oversight is necessary, not sufficient. Do not claim “Act compliant” because you have a button.",[348,1494,1496],{"id":1495},"how-do-we-stop-rubber-stamping","How do we stop rubber-stamping?",[186,1498,1499],{},"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.",[348,1501,1503],{"id":1502},"is-a-chat-saying-shall-i-proceed-enough","Is a chat saying “shall I proceed?” enough?",[186,1505,1506],{},"Only if it is bound to identity, shows the payload, and cannot be skipped.",[348,1508,1510],{"id":1509},"what-is-the-difference-between-in-the-loop-and-on-the-loop","What is the difference between in the loop and on the loop?",[186,1512,1513,1514,1517],{},"In the loop: the job cannot finish the risky step without a human act. On the loop: a human ",[226,1515,1516],{},"may"," intervene. Vendors blur them because the second is cheaper to ship.",[348,1519,1521],{"id":1520},"can-agent-teams-approve-each-others-work","Can agent teams approve each other’s work?",[186,1523,1524,1525,362],{},"They can criticise drafts. Release still needs a named human. Multi-agent review is not a signer. See ",[199,1526,1527],{"href":1427},"What is multi-agent AI",[348,1529,1531],{"id":1530},"do-read-only-jobs-need-a-person-every-time","Do read-only jobs need a person every time?",[186,1533,1534],{},"Usually not. That is the point of connectors defaulting to read-only. Put people where reversibility changes.",[348,1536,1538],{"id":1537},"how-does-this-relate-to-air-canada","How does this relate to Air Canada?",[186,1540,1541],{},"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.",[348,1543,1545],{"id":1544},"what-about-mata-v-avianca","What about Mata v. Avianca?",[186,1547,1548],{},"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.",[348,1550,1552],{"id":1551},"can-we-batch-approve-200-records","Can we batch-approve 200 records?",[186,1554,1555],{},"Not as one click with no visible set. Bulk without inspection is a rubber stamp with worse radius. Show the set.",[348,1557,1559],{"id":1558},"does-logging-approvals-in-slack-count","Does logging approvals in Slack count?",[186,1561,1562],{},"Only if identity, payload, and outcome are bound and retained as a control record. A thumbs-up emoji is theatre.",[207,1564,609],{"id":608},[186,1566,1567,1004,1570,1572,1573,1575],{},[199,1568,1569],{"href":796},"What is AI governance",[199,1571,1527],{"href":1427}," — the person should sit at ",[221,1574,1423],{},", not at every internal hand-off.",[207,1577,622],{"id":621},[215,1579,1580,1586,1592,1598],{},[218,1581,1582],{},[199,1583,1585],{"href":1200,"rel":1584},[203],"CBC, Air Canada chatbot lawsuit",[218,1587,1588],{},[199,1589,1591],{"href":1206,"rel":1590},[203],"Civil Resolution Tribunal, Moffatt v. Air Canada",[218,1593,1594],{},[199,1595,1597],{"href":1250,"rel":1596},[203],"EU AI Act (Regulation 2024/1689), including Article 14",[218,1599,1600],{},[199,1601,1603],{"href":1290,"rel":1602},[203],"Reuters, New York lawyers sanctioned for ChatGPT fake cases",{"title":155,"searchDepth":156,"depth":156,"links":1605},[1606,1607,1611,1612,1626,1627],{"id":209,"depth":156,"text":210},{"id":316,"depth":156,"text":317,"children":1608},[1609,1610],{"id":350,"depth":648,"text":351},{"id":399,"depth":648,"text":400},{"id":483,"depth":156,"text":484},{"id":515,"depth":156,"text":516,"children":1613},[1614,1615,1616,1617,1618,1619,1620,1621,1622,1623,1624,1625],{"id":1474,"depth":648,"text":1475},{"id":1481,"depth":648,"text":1482},{"id":1488,"depth":648,"text":1489},{"id":1495,"depth":648,"text":1496},{"id":1502,"depth":648,"text":1503},{"id":1509,"depth":648,"text":1510},{"id":1520,"depth":648,"text":1521},{"id":1530,"depth":648,"text":1531},{"id":1537,"depth":648,"text":1538},{"id":1544,"depth":648,"text":1545},{"id":1551,"depth":648,"text":1552},{"id":1558,"depth":648,"text":1559},{"id":608,"depth":156,"text":609},{"id":621,"depth":156,"text":622},"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":1180,"description":1628},"blog/what-is-human-in-the-loop-ai",[671,1633,797,1634],"human-in-the-loop","approvals","8xJuZYtNYOnCYCsxxbZqk340T6yMdyoPDYPvvIta_v8",{"fold":1637,"id":1641,"title":1642,"archived":149,"authors":150,"badge":150,"body":1643,"date":150,"department":150,"description":155,"extension":158,"eyebrow":150,"faqHeader":150,"faqs":150,"footerBand":1647,"headline":150,"image":150,"industry":150,"jobType":150,"listed":149,"location":150,"navigation":115,"openRoles":150,"pageLayout":150,"path":1651,"relatedHeading":150,"seo":1652,"series":150,"sitemap":115,"status":150,"stem":1653,"subhead":150,"tags":150,"video":150,"whyJoin":150,"workplaceType":150,"__hash__":1654},{"headline":1638,"description":1639,"primaryLabel":8,"primaryTo":1640,"secondaryLabel":694,"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":1644,"toc":1645},[],{"title":155,"searchDepth":156,"depth":156,"links":1646},[],{"headline":1648,"description":1649,"primaryLabel":8,"primaryTo":1640,"secondaryLabel":1650,"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":1642,"description":155},"shared/cta","YHK6Fb8AvCPR1zZq7R_xiXUG0hwhP5UxHA8Ix52JQp4",1787194066611]