[{"data":1,"prerenderedAt":1656},["ShallowReactive",2],{"site-nav-content":3,"hiring-banner-content":177,"blog:\u002Fblog\u002Fwhat-is-causal-ai-for-operations":189,"blog-index-copy":693,"blog:\u002Fblog\u002Fwhat-is-causal-ai-for-operations:surround":714,"site-cta-content":1637},{"header":4,"productNav":9,"nav":42,"footer":61,"askAI":132,"id":161,"title":162,"archived":163,"authors":164,"badge":164,"body":165,"date":164,"definedTerm":164,"department":164,"description":169,"extension":172,"eyebrow":164,"faqHeader":164,"faqs":164,"footerBand":164,"headline":164,"image":164,"industry":164,"jobType":164,"listed":131,"location":164,"navigation":131,"openRoles":164,"pageLayout":164,"path":173,"relatedHeading":164,"seo":174,"series":164,"sitemap":163,"status":164,"stem":175,"subhead":164,"tags":164,"video":164,"whyJoin":164,"workplaceType":164,"__hash__":176},{"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","\u002Foverview","Seven layers. One closed loop.",{"label":15,"to":16,"description":17},"Conflux","\u002Fproduct\u002Fconflux","Where your team, workstreams, and agents meet.",{"label":19,"to":20,"description":21},"Agent Teams","\u002Fproduct\u002Fagent-teams","Specialist teams - governed from day one.",{"label":23,"to":24,"description":25},"Lifecycle Graph","\u002Fproduct\u002Flifecycle-graph","Intelligence that compounds across every interaction.",{"label":27,"to":28,"description":29},"Company Wiki","\u002Fproduct\u002Fwiki","Playbooks and policies where expertise stays.",{"label":31,"to":32,"description":33},"Workstreams","\u002Fproduct\u002Fworkstreams","From brief to signed-off deliverable on one canvas.",{"label":35,"to":36,"description":37},"Perception Console","\u002Fproduct\u002Fperception","Ask your whole business in plain English.",{"label":39,"to":40,"description":41},"Governance","\u002Fproduct\u002Fgovernance","Frontier AI you can actually sign off on.",[43,46,49,52,55,58],{"label":44,"to":45},"Models","\u002Fmodels",{"label":47,"to":48},"Pricing","\u002Fpricing",{"label":50,"to":51},"Integrations","\u002Fintegrations",{"label":53,"to":54},"Security","\u002Fsecurity",{"label":56,"to":57},"Partners","\u002Fpartners",{"label":59,"to":60},"Insights","\u002Fblog",{"productHeading":5,"companyHeading":62,"resourcesHeading":63,"legalHeading":64,"docsLabel":65,"docsUrl":66,"statementLines":67,"copyright":70,"companyLinks":71,"resourcesLinks":86,"legalLinks":102,"socialLinks":109,"bottomLinks":119},"Company","Resources","Legal","Docs","https:\u002F\u002Fdocs.gonimbus.ai",[68,69],"Stop training someone else's model.","Control your AI.","© 2026 Nimbus Intelligence, Inc. All rights reserved.",[72,73,74,75,76,78,81,84],{"label":47,"to":48},{"label":50,"to":51},{"label":53,"to":54},{"label":59,"to":60},{"label":77,"to":57},"Partner Program",{"label":79,"to":80},"Careers","\u002Fcareers",{"label":82,"to":83},"System status","\u002Fstatus",{"label":7,"to":85},"\u002Fcontact",[87,90,93,96,99],{"label":88,"to":89},"Glossary","\u002Fglossary",{"label":91,"to":92},"Compare","\u002Fcompare",{"label":94,"to":95},"Evaluate","\u002Fevaluate",{"label":97,"to":98},"Problems","\u002Fproblems",{"label":100,"to":101},"Use cases","\u002Fuse-cases",[103,106],{"label":104,"to":105},"Terms of Service","\u002Fterms",{"label":107,"to":108},"Privacy Policy","\u002Fprivacy",[110,113,116],{"label":111,"href":112},"LinkedIn","https:\u002F\u002Fwww.linkedin.com\u002Fcompany\u002Fgonimbusai\u002F",{"label":114,"href":115},"X","https:\u002F\u002Fx.com\u002Fgonimbusai",{"label":117,"href":118},"Instagram","https:\u002F\u002Fwww.instagram.com\u002Fgonimbus_ai\u002F",[120,122,124,127,128],{"label":121,"to":105},"Terms",{"label":123,"to":108},"Privacy",{"label":125,"to":126},"Compliance","\u002Fcompliance",{"label":82,"to":83},{"label":129,"to":130,"external":131},"LLMs.txt","\u002Fllms.txt",true,{"text":133,"prompt":134,"platforms":135},"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. Summarize the highlights from Nimbus's website: https:\u002F\u002Fgonimbus.ai",[136,141,146,151,156],{"name":137,"label":138,"icon":139,"hrefPrefix":140},"chatgpt","ChatGPT","simple-icons:openai","https:\u002F\u002Fchatgpt.com\u002F?prompt=",{"name":142,"label":143,"icon":144,"hrefPrefix":145},"perplexity","Perplexity","mdi:magnify","https:\u002F\u002Fwww.perplexity.ai\u002Fsearch\u002Fnew?q=",{"name":147,"label":148,"icon":149,"hrefPrefix":150},"grok","Grok","simple-icons:x","https:\u002F\u002Fx.com\u002Fi\u002Fgrok?text=",{"name":152,"label":153,"icon":154,"hrefPrefix":155},"claude","Claude","simple-icons:anthropic","https:\u002F\u002Fclaude.ai\u002Fnew?q=",{"name":157,"label":158,"icon":159,"hrefPrefix":160},"google-ai","Google AI","simple-icons:google","https:\u002F\u002Fwww.google.com\u002Fsearch?udm=50&aep=11&q=","content\u002Fshared\u002Fnav.md","Site navigation",false,null,{"type":166,"value":167,"toc":168},"minimark",[],{"title":169,"searchDepth":170,"depth":170,"links":171},"",2,[],"md","\u002Fshared\u002Fnav",{"title":162,"description":169},"shared\u002Fnav","Q7sDe7TuGEwhwUMamyeOxjvGlsUaF3Iay9VTW0KaE_U",{"enabled":163,"message":178,"linkLabel":79,"linkHref":80,"id":179,"title":180,"archived":163,"authors":164,"badge":164,"body":181,"date":164,"definedTerm":164,"department":164,"description":169,"extension":172,"eyebrow":164,"faqHeader":164,"faqs":164,"footerBand":164,"headline":164,"image":164,"industry":164,"jobType":164,"listed":131,"location":164,"navigation":131,"openRoles":164,"pageLayout":164,"path":185,"relatedHeading":164,"seo":186,"series":164,"sitemap":163,"status":164,"stem":187,"subhead":164,"tags":164,"video":164,"whyJoin":164,"workplaceType":164,"__hash__":188},"We're hiring! Join the team building the Sentient Enterprise.","content\u002Fshared\u002Fhiring.md","Hiring banner",{"type":166,"value":182,"toc":183},[],{"title":169,"searchDepth":170,"depth":170,"links":184},[],"\u002Fshared\u002Fhiring",{"title":180,"description":169},"shared\u002Fhiring","1zs3boivKda1e-b-hAyuNcmZSKjZUAXmecnwHVgcHzk",{"id":190,"title":191,"archived":163,"authors":192,"badge":196,"body":198,"date":669,"definedTerm":164,"department":164,"description":670,"extension":172,"eyebrow":164,"faqHeader":671,"faqs":674,"footerBand":164,"headline":164,"image":164,"industry":164,"jobType":164,"listed":163,"location":164,"navigation":131,"openRoles":164,"pageLayout":164,"path":684,"relatedHeading":164,"seo":685,"series":686,"sitemap":131,"status":164,"stem":687,"subhead":164,"tags":688,"video":164,"whyJoin":164,"workplaceType":164,"__hash__":692},"content\u002Fblog\u002Fwhat-is-causal-ai-for-operations.md","What is Causal AI for Operations",[193],{"name":194,"to":195},"Nimbus Research","https:\u002F\u002Fgonimbus.ai",{"label":197},"Explainer",{"type":166,"value":199,"toc":644},[200,208,222,229,232,235,240,300,303,318,322,325,328,353,360,374,385,390,400,406,412,418,424,428,434,440,446,452,458,461,464,481,485,494,497,511,515,519,522,526,529,533,536,540,543,547,550,554,560,564,567,571,574,578,581,585,588,592,599,603,606,610,623,627],[201,202,203,204],"p",{},"“Causal AI” is a phrase people type into ChatGPT when they mean: ",[205,206,207],"strong",{},"can we tell why something happened, or are we guessing?",[201,209,210,211,216,217,221],{},"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 ",[212,213,215],"a",{"href":214},"https:\u002F\u002Fplato.stanford.edu\u002Fentries\u002Fcausal-models\u002F","Stanford Encyclopedia of Philosophy’s entry on causal models"," is a fair orientation to that science. Judea Pearl’s overview, ",[212,218,220],{"href":219},"https:\u002F\u002Fftp.cs.ucla.edu\u002Fpub\u002Fstat_ser\u002Fr350.pdf","Causal inference in statistics",", is the technical companion: identification is a design problem, not a paragraph problem.",[201,223,224,225,228],{},"In operations, the question is more everyday and more urgent: ",[205,226,227],{},"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.",[201,230,231],{},"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.",[201,233,234],{},"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.",[236,237,239],"h2",{"id":238},"words-youll-hear","Words you’ll hear",[241,242,243,250,256,262,273,284,294],"ul",{},[244,245,246,249],"li",{},[205,247,248],{},"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.",[244,251,252,255],{},[205,253,254],{},"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.",[244,257,258,261],{},[205,259,260],{},"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.",[244,263,264,267,268,272],{},[205,265,266],{},"Lifecycle graph."," The company’s chain of AI work: what was asked, who signed, what changed. See ",[212,269,271],{"href":270},"\u002Fblog\u002Fwhat-is-a-lifecycle-graph\u002F","What is a lifecycle graph",".",[244,274,275,278,279,283],{},[205,276,277],{},"Provenance."," Who, what, when, derived from what. ",[212,280,282],{"href":281},"https:\u002F\u002Fwww.w3.org\u002FTR\u002Fprov-overview\u002F","W3C PROV"," is the open vocabulary for that idea.",[244,285,286,289,290,272],{},[205,287,288],{},"Confounder."," In science, a hidden third factor. In operations, the hidden factor is often “a human pasted a consumer-model answer into CRM.” See ",[212,291,293],{"href":292},"\u002Fblog\u002Fwhat-is-shadow-ai\u002F","What is shadow AI",[244,295,296,299],{},[205,297,298],{},"Rationale."," The model’s English explanation. Often written after the fact. Not a recorded structure.",[201,301,302],{},"Keep two layers apart:",[304,305,306,312],"ol",{},[244,307,308,311],{},[205,309,310],{},"Causal science."," Did the discount cause the win? Needs a design, not a fluent paragraph.",[244,313,314,317],{},[205,315,316],{},"Causal operations."," Brief → sources → proposal → approval → write → system response. Needs a record.",[236,319,321],{"id":320},"why-you-should-care","Why you should care",[201,323,324],{},"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.",[201,326,327],{},"Causal operations affects you if you:",[241,329,330,336,347],{},[244,331,332,335],{},[205,333,334],{},"Have to explain a change."," “Who caused this field to move, against which rule?”",[244,337,338,346],{},[205,339,340,341,345],{},"Need to know what ",[342,343,344],"em",{},"would"," have happened without approval."," In a real gate, the answer is nothing.",[244,348,349,352],{},[205,350,351],{},"Are tempted to file a model’s “because” as truth."," Natural-language rationales are often written after the fact.",[201,354,355,356,359],{},"You do ",[205,357,358],{},"not"," need a data-science sprint to ask:",[241,361,362,365,368,371],{},[244,363,364],{},"Why was this record changed?",[244,366,367],{},"Which policy version caused this refusal?",[244,369,370],{},"Did a spend cap stop the run?",[244,372,373],{},"Did analysis change the CRM, or only produce a draft?",[201,375,376,377,380,381,384],{},"You ",[205,378,379],{},"should"," need a statistician if you want to know whether signed next-step updates ",[342,382,383],{},"caused"," wins. The work record can attach “this account was treated.” Estimation is extra.",[386,387,389],"h3",{"id":388},"what-changes-by-role","What changes by role",[201,391,392,395,396,399],{},[205,393,394],{},"Finance."," Sampling a journal requires the chain, not a story. Spend caps that fire are causes of ",[342,397,398],{},"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.”",[201,401,402,405],{},[205,403,404],{},"Legal."," 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.",[201,407,408,411],{},[205,409,410],{},"Operations."," 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.",[201,413,414,417],{},[205,415,416],{},"Go-to-market."," 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.",[201,419,420,423],{},[205,421,422],{},"Security."," 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.",[386,425,427],{"id":426},"what-people-get-wrong","What people get wrong",[201,429,430,433],{},[205,431,432],{},"The model’s “because.”"," Fluency is not identification and not a recorded structure.",[201,435,436,439],{},[205,437,438],{},"Two rising lines."," Correlation. File it as a hypothesis.",[201,441,442,445],{},[205,443,444],{},"Using the same AI that proposed the treatment to declare success."," That is marking your own homework.",[201,447,448,451],{},[205,449,450],{},"Skipping the operations layer to buy a science platform."," Without reconstructable interventions, the science team inherits Slack folklore.",[201,453,454,457],{},[205,455,456],{},"Using the graph as a BI tool."," Canonical commercial metrics stay in the warehouse. The graph answers mixed policy \u002F approval \u002F live-system questions.",[201,459,460],{},"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.",[201,462,463],{},"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.",[201,465,466,467,470,471,475,476,480],{},"Adjacent: ",[212,468,469],{"href":270},"lifecycle graph"," is the product shape of the operations layer. ",[212,472,474],{"href":473},"\u002Fblog\u002Fwhat-is-institutional-memory-in-enterprise-ai\u002F","Institutional memory"," is what remains after people leave. ",[212,477,479],{"href":478},"\u002Fblog\u002Fwhat-is-write-back-governance\u002F","Write-back governance"," makes “nothing happened” a possible true answer.",[236,482,484],{"id":483},"how-this-shows-up-in-nimbus","How this shows up in Nimbus",[201,486,487,488,490,491,272],{},"Nimbus implements causal operations as the ",[205,489,23],{}," plus ",[205,492,493],{},"Perception",[201,495,496],{},"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.",[201,498,499,500,503,504,507,508,272],{},"See ",[212,501,23],{"href":502},"\u002Fproduct\u002Flifecycle-graph\u002F"," and ",[212,505,493],{"href":506},"\u002Fproduct\u002Fperception\u002F",". Writes that cannot happen without a signer are ",[212,509,510],{"href":478},"write-back governance",[236,512,514],{"id":513},"questions-people-actually-ask","Questions people actually ask",[386,516,518],{"id":517},"if-the-model-explains-why-is-that-causal-ai","If the model explains “why,” is that causal AI?",[201,520,521],{},"No. A fluent paragraph is not a recorded structure, and it is not a statistical identification.",[386,523,525],{"id":524},"do-we-need-advanced-causal-statistics-to-buy-an-operating-layer","Do we need advanced causal statistics to buy an operating layer?",[201,527,528],{},"No. You need reconstructable interventions. If you later staff a science team, they will thank you for not storing decisions as Slack folklore.",[386,530,532],{"id":531},"can-the-graph-estimate-lift","Can the graph estimate lift?",[201,534,535],{},"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.",[386,537,539],{"id":538},"where-does-this-end-and-a-bi-tool-start","Where does this end and a BI tool start?",[201,541,542],{},"The graph is for AI-work lineage and mixed policy \u002F approval \u002F live-system questions. BI remains for canonical commercial metrics. Dumping bookings into the graph as a fake causal model is a mess.",[386,544,546],{"id":545},"what-is-an-intervention-in-this-sense","What is an intervention in this sense?",[201,548,549],{},"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.",[386,551,553],{"id":552},"why-does-a-fail-closed-gate-matter-for-causality","Why does a fail-closed gate matter for causality?",[201,555,556,557,559],{},"Because the counterfactual is clean: without the named signer, the live system does not change. Fail-open systems cannot say what ",[342,558,344],{}," have happened; they can only hope someone noticed.",[386,561,563],{"id":562},"is-w3c-prov-the-same-as-a-lifecycle-graph","Is W3C PROV the same as a lifecycle graph?",[201,565,566],{},"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.",[386,568,570],{"id":569},"can-we-reconstruct-causes-from-crm-field-history-plus-slack","Can we reconstruct causes from CRM field history plus Slack?",[201,572,573],{},"Field history says the value changed. Slack may contain a rumour. Neither joins playbook version, quoted payload, and signer identity as a single chain.",[386,575,577],{"id":576},"does-causal-ai-mean-the-model-uses-causal-graphs-internally","Does “causal AI” mean the model uses causal graphs internally?",[201,579,580],{},"Sometimes, in research marketing. In this article it means operations can answer why a change happened. Ask vendors which meaning they are selling.",[386,582,584],{"id":583},"how-should-we-talk-to-the-board","How should we talk to the board?",[201,586,587],{},"Separate “we can reconstruct what we did” from “we can estimate market lift.” The first is a control. The second is a study.",[386,589,591],{"id":590},"where-does-the-wiki-fit","Where does the wiki fit?",[201,593,594,595,272],{},"Asserted policy is often the parent of a refusal or a draft. “Which playbook version caused this flag?” is a causal-operations question. See ",[212,596,598],{"href":597},"\u002Fblog\u002Fwhat-is-a-company-wiki-for-ai-agents\u002F","What is a company wiki for AI agents",[386,600,602],{"id":601},"how-is-this-different-from-audit-logging","How is this different from audit logging?",[201,604,605],{},"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.",[236,607,609],{"id":608},"related-reading","Related reading",[201,611,612,614,615,619,620,272],{},[212,613,271],{"href":270},", ",[212,616,618],{"href":617},"\u002Fblog\u002Fwhat-is-an-ai-workstream\u002F","What is an AI workstream",", and ",[212,621,622],{"href":473},"What is institutional memory in enterprise AI",[236,624,626],{"id":625},"sources","Sources",[241,628,629,634,639],{},[244,630,631],{},[212,632,633],{"href":214},"Stanford Encyclopedia of Philosophy, Causal Models",[244,635,636],{},[212,637,638],{"href":219},"Pearl, Causal inference in statistics: An overview",[244,640,641],{},[212,642,643],{"href":281},"W3C PROV overview",{"title":169,"searchDepth":170,"depth":170,"links":645},[646,647,652,653,667,668],{"id":238,"depth":170,"text":239},{"id":320,"depth":170,"text":321,"children":648},[649,651],{"id":388,"depth":650,"text":389},3,{"id":426,"depth":650,"text":427},{"id":483,"depth":170,"text":484},{"id":513,"depth":170,"text":514,"children":654},[655,656,657,658,659,660,661,662,663,664,665,666],{"id":517,"depth":650,"text":518},{"id":524,"depth":650,"text":525},{"id":531,"depth":650,"text":532},{"id":538,"depth":650,"text":539},{"id":545,"depth":650,"text":546},{"id":552,"depth":650,"text":553},{"id":562,"depth":650,"text":563},{"id":569,"depth":650,"text":570},{"id":576,"depth":650,"text":577},{"id":583,"depth":650,"text":584},{"id":590,"depth":650,"text":591},{"id":601,"depth":650,"text":602},{"id":608,"depth":170,"text":609},{"id":625,"depth":170,"text":626},"2026-08-17","Causal AI for operations answers why a change happened, using the actual steps and the approval, not a guess about the chatbot.",{"eyebrow":672,"title":673},"Short answers","The actual steps, not a guessed cause",[675,678,681],{"question":676,"answer":677},"What is causal AI for operations?","The ability to answer why a change happened with the actual steps and the approval, not a guess that the chatbot caused a lift.",{"question":679,"answer":680},"Why is correlation not enough?","A metric moved. That does not tell you which action, which person, or which system produced it.",{"question":682,"answer":683},"What do you need on the record?","The trigger, the steps, the signer, and the system write. Without those, the cause is a story.","\u002Fblog\u002Fwhat-is-causal-ai-for-operations",{"title":191,"description":670},"explainer","blog\u002Fwhat-is-causal-ai-for-operations",[686,689,690,691],"causal-ai","lifecycle-graph","operations","YEdYX44CGV1ueBh0x0l5KK3vsh9X51db48M0nGl9Ang",{"hero":694,"id":696,"title":697,"archived":163,"authors":164,"badge":164,"body":698,"date":164,"definedTerm":164,"department":164,"description":702,"extension":172,"eyebrow":703,"faqHeader":164,"faqs":164,"footerBand":704,"headline":164,"image":164,"industry":164,"jobType":164,"listed":131,"location":164,"navigation":131,"openRoles":164,"pageLayout":164,"path":60,"relatedHeading":710,"seo":711,"series":164,"sitemap":131,"status":164,"stem":712,"subhead":164,"tags":164,"video":164,"whyJoin":164,"workplaceType":164,"__hash__":713},{"filename":695},"u2221455217_Flat_design_of_a_futuristic_minimalist_landscape__5d589295-cdea-4ea9-a262-be766881accf_1.png","content\u002Fblog\u002Findex.md","Exploring the future of intelligence.",{"type":166,"value":699,"toc":700},[],{"title":169,"searchDepth":170,"depth":170,"links":701},[],"Deep dives into pre-cognitive intelligence, sentient enterprises, and the evolving landscape of AI-driven business transformation.","Latest Research",{"headline":705,"description":706,"primaryLabel":707,"primaryTo":708,"secondaryLabel":709,"secondaryTo":12},"Stay at the frontier.","Subscribe for product updates and new insights.","Subscribe","\u002Fnewsletter","Explore the platform","More research",{"title":697,"description":702},"blog\u002Findex","BFSWGYO9bcTlaulivKYWyg08_DJHsdGg3OC6g_CG1Hw",[715,1195],{"id":716,"title":717,"archived":163,"authors":718,"badge":720,"body":721,"date":669,"definedTerm":164,"department":164,"description":1174,"extension":172,"eyebrow":164,"faqHeader":1175,"faqs":1177,"footerBand":164,"headline":164,"image":164,"industry":164,"jobType":164,"listed":163,"location":164,"navigation":131,"openRoles":164,"pageLayout":164,"path":1187,"relatedHeading":164,"seo":1188,"series":686,"sitemap":131,"status":164,"stem":1189,"subhead":164,"tags":1190,"video":164,"whyJoin":164,"workplaceType":164,"__hash__":1194},"content\u002Fblog\u002Fwhat-is-enterprise-rag.md","What is Enterprise RAG",[719],{"name":194,"to":195},{"label":197},{"type":166,"value":722,"toc":1150},[723,733,741,751,759,762,764,829,836,838,841,844,869,876,878,883,888,893,898,908,910,916,926,932,938,944,950,957,964,966,977,984,996,998,1002,1005,1009,1012,1016,1027,1031,1034,1038,1043,1047,1055,1059,1062,1066,1073,1077,1087,1091,1094,1098,1106,1110,1116,1118,1124,1126],[201,724,725,726,729,730,272],{},"RAG stands for ",[205,727,728],{},"retrieval-augmented generation",". In plain language: ",[205,731,732],{},"look up, then answer",[201,734,735,736,740],{},"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 ",[212,737,739],{"href":738},"https:\u002F\u002Farxiv.org\u002Fabs\u002F2005.11401","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.",[201,742,743,746,747,750],{},[205,744,745],{},"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 ",[342,748,749],{},"company"," facts.",[201,752,753,754,758],{},"It is not, by itself, an operating system, a write gate, or a memory of decisions. ",[212,755,757],{"href":756},"https:\u002F\u002Feur-lex.europa.eu\u002Feli\u002Freg\u002F2016\u002F679\u002Foj","GDPR"," does not pause because the “user” of the files is an AI. Retrieval is still processing personal data.",[201,760,761],{},"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.”",[236,763,239],{"id":238},[241,765,766,772,778,784,790,800,806,812,818],{},[244,767,768,771],{},[205,769,770],{},"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.",[244,773,774,777],{},[205,775,776],{},"Embedding \u002F 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.",[244,779,780,783],{},[205,781,782],{},"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.",[244,785,786,789],{},[205,787,788],{},"Hallucination."," Fluent invention. RAG reduces it on company facts. It does not eliminate it, and it does not stop an ungoverned write.",[244,791,792,795,796,799],{},[205,793,794],{},"Asserted policy."," What the company currently wants. That belongs in a ",[212,797,798],{"href":597},"company wiki",", not in whichever PDF sounded closest.",[244,801,802,805],{},[205,803,804],{},"Chunking."," Splitting files so search can retrieve a passage. Bad chunking is how a table’s header parts company from its numbers.",[244,807,808,811],{},[205,809,810],{},"Freshness."," When the index sees a change. At work, “we changed the vendor template yesterday” is an SLA question.",[244,813,814,817],{},[205,815,816],{},"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.",[244,819,820,823,824,828],{},[205,821,822],{},"Prompt injection via documents."," Retrieved text that instructs the model to ignore policy. The ",[212,825,827],{"href":826},"https:\u002F\u002Fgenai.owasp.org\u002Fllm-top-10\u002F","OWASP Top 10 for LLM applications"," treats that as a security surface.",[201,830,831,832,272],{},"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 ",[212,833,835],{"href":834},"\u002Fblog\u002Fnimbus-vs-glean\u002F","Nimbus vs Glean",[236,837,321],{"id":320},[201,839,840],{},"It affects you if answers about policy, customers, or finance will be trusted — and if those answers later need a source you can click.",[201,842,843],{},"Enterprise lookup adds:",[241,845,846,852,857,863],{},[244,847,848,851],{},[205,849,850],{},"Permissions."," SharePoint, Salesforce, and Drive access lists still apply.",[244,853,854,856],{},[205,855,810],{}," If the index updates on Sundays, your SLA is weekly.",[244,858,859,862],{},[205,860,861],{},"Poisoned documents."," Retrieved text can instruct the model to ignore policy.",[244,864,865,868],{},[205,866,867],{},"Purpose."," Indexing everything “just in case” is a privacy and quality problem.",[201,870,871,872,875],{},"Treat RAG as ",[205,873,874],{},"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.",[386,877,389],{"id":388},[201,879,880,882],{},[205,881,394],{}," 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.",[201,884,885,887],{},[205,886,404],{}," 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.",[201,889,890,892],{},[205,891,410],{}," 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.”",[201,894,895,897],{},[205,896,416],{}," 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.",[201,899,900,902,903,907],{},[205,901,422],{}," Superuser crawlers, poisoned documents, and a second store of sensitive text. Security should ask who the retriever authenticates as, whether ",[212,904,906],{"href":905},"\u002Fblog\u002Fwhat-is-model-context-protocol\u002F","MCP"," helpers search as a superuser, and whether prompt injection in a PDF can change tool behaviour. Network search products are not write gates.",[386,909,427],{"id":426},[201,911,912,915],{},[205,913,914],{},"Indexing everything."," Quality falls. Privacy rises. Purpose disappears.",[201,917,918,921,922,272],{},[205,919,920],{},"RAG as an OS."," Lookup does not isolate jobs, quote writes, or store decisions. See ",[212,923,925],{"href":924},"\u002Fblog\u002Fwhat-is-an-enterprise-ai-operating-system\u002F","What is an enterprise AI operating system",[201,927,928,931],{},[205,929,930],{},"RAG as the wiki."," Retrieved files are what exists. The wiki is what is in force.",[201,933,934,937],{},[205,935,936],{},"Citations without versions."," Legal cannot check “the wiki” or “our Drive.”",[201,939,940,943],{},[205,941,942],{},"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.",[201,945,946,949],{},[205,947,948],{},"Assuming hallucination is solved."," Missing files still produce fluent guesses. Ungoverned writes still land.",[201,951,952,953,956],{},"Good looks like: permission-aware retrieval scoped to the ",[212,954,955],{"href":617},"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.”",[201,958,959,960,963],{},"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 ",[212,961,962],{"href":597},"wiki",". The paper does not answer those questions. Your runtime must.",[236,965,484],{"id":483},[201,967,968,969,972,973,976],{},"Nimbus uses RAG-like retrieval ",[205,970,971],{},"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 ",[212,974,975],{"href":473},"institutional memory",", not a forgotten context window.",[201,978,979,980,983],{},"A common plug so AI apps can use the same tools — ",[212,981,982],{"href":905},"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.",[201,985,499,986,614,990,619,993,272],{},[212,987,989],{"href":988},"\u002Fproduct\u002Fwiki\u002F","Wiki",[212,991,31],{"href":992},"\u002Fproduct\u002Fworkstreams\u002F",[212,994,39],{"href":995},"\u002Fproduct\u002Fgovernance\u002F",[236,997,514],{"id":513},[386,999,1001],{"id":1000},"will-rag-stop-the-model-making-things-up","Will RAG stop the model making things up?",[201,1003,1004],{},"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.",[386,1006,1008],{"id":1007},"is-indexing-everything-just-in-case-a-good-idea","Is indexing everything “just in case” a good idea?",[201,1010,1011],{},"No. Indexing without a purpose is a privacy and quality problem. Scope the corpus to the job.",[386,1013,1015],{"id":1014},"how-is-this-different-from-a-company-wiki","How is this different from a company wiki?",[201,1017,1018,1019,1022,1023,1026],{},"The wiki is what the company ",[342,1020,1021],{},"wants"," to be true. RAG is what ",[342,1024,1025],{},"exists"," in files. If they conflict, the wiki should win unless a human promotes a change.",[386,1028,1030],{"id":1029},"can-warehouse-sql-replace-rag","Can warehouse SQL replace RAG?",[201,1032,1033],{},"They answer different questions. Policy prose is retrieval. Regional revenue is a query. Many real jobs need both.",[386,1035,1037],{"id":1036},"is-glean-or-similar-an-enterprise-ai-os","Is Glean (or similar) an enterprise AI OS?",[201,1039,1040,1041,272],{},"Permission-aware search is still search. It does not, by itself, quote a CRM write or bind a named signer. See ",[212,1042,835],{"href":834},[386,1044,1046],{"id":1045},"why-do-invoice-numbers-fail-in-vector-search","Why do invoice numbers fail in vector search?",[201,1048,1049,1050,1054],{},"Embeddings capture similarity of meaning, not identity of tokens. Hybrid search — keywords plus vectors — is how you find ",[1051,1052,1053],"code",{},"INV-88421"," instead of a semantically nearby invoice.",[386,1056,1058],{"id":1057},"how-fast-should-the-index-update","How fast should the index update?",[201,1060,1061],{},"As fasthe decision you are supporting. If a template changed yesterday and the agent still cites last month, your SLA is wrong. Publish the lag.",[386,1063,1065],{"id":1064},"what-is-document-based-prompt-injection","What is document-based prompt injection?",[201,1067,1068,1069,1072],{},"A retrieved file that says, in effect, “ignore previous instructions.” Treat retrieved text as untrusted input. The ",[212,1070,1071],{"href":826},"OWASP LLM list"," is the starting point. A wiki conflict rule and a write gate still matter.",[386,1074,1076],{"id":1075},"does-gdpr-apply-to-the-vector-index","Does GDPR apply to the vector index?",[201,1078,1079,1080,503,1083,272],{},"Yes, if it holds personal data. The index is another copy. Erasure, purpose, and access control apply. See the ",[212,1081,1082],{"href":756},"GDPR text",[212,1084,1086],{"href":1085},"https:\u002F\u002Fico.org.uk\u002Ffor-organisations\u002Fuk-gdpr-guidance-and-resources\u002Fartificial-intelligence\u002F","ICO AI guidance",[386,1088,1090],{"id":1089},"can-mcp-make-retrieval-respect-permissions","Can MCP make retrieval respect permissions?",[201,1092,1093],{},"Only if the helper is built that way. The protocol will happily pass superuser results.",[386,1095,1097],{"id":1096},"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?",[201,1099,1100,1101,1105],{},"Internal policy in a customer bot is how invented fares happen unless a human still owns the commitment. Air Canada’s case — ",[212,1102,1104],{"href":1103},"https:\u002F\u002Fwww.cbc.ca\u002Fnews\u002Fcanada\u002Fbritish-columbia\u002Fair-canada-chatbot-lawsuit-1.7116416","CBC"," — is retrieval-plus-generation without a working gate.",[386,1107,1109],{"id":1108},"how-do-we-know-which-sources-a-decision-used","How do we know which sources a decision used?",[201,1111,1112,1113,1115],{},"Record them on the ",[212,1114,469],{"href":270},". A context window that evaporates is not memory.",[236,1117,609],{"id":608},[201,1119,1120,503,1122,272],{},[212,1121,598],{"href":597},[212,1123,622],{"href":473},[236,1125,626],{"id":625},[241,1127,1128,1132,1136,1140,1145],{},[244,1129,1130],{},[212,1131,739],{"href":738},[244,1133,1134],{},[212,1135,757],{"href":756},[244,1137,1138],{},[212,1139,827],{"href":826},[244,1141,1142],{},[212,1143,1144],{"href":1085},"ICO, AI and data protection",[244,1146,1147],{},[212,1148,1149],{"href":1103},"CBC, Air Canada chatbot lawsuit",{"title":169,"searchDepth":170,"depth":170,"links":1151},[1152,1153,1157,1158,1172,1173],{"id":238,"depth":170,"text":239},{"id":320,"depth":170,"text":321,"children":1154},[1155,1156],{"id":388,"depth":650,"text":389},{"id":426,"depth":650,"text":427},{"id":483,"depth":170,"text":484},{"id":513,"depth":170,"text":514,"children":1159},[1160,1161,1162,1163,1164,1165,1166,1167,1168,1169,1170,1171],{"id":1000,"depth":650,"text":1001},{"id":1007,"depth":650,"text":1008},{"id":1014,"depth":650,"text":1015},{"id":1029,"depth":650,"text":1030},{"id":1036,"depth":650,"text":1037},{"id":1045,"depth":650,"text":1046},{"id":1057,"depth":650,"text":1058},{"id":1064,"depth":650,"text":1065},{"id":1075,"depth":650,"text":1076},{"id":1089,"depth":650,"text":1090},{"id":1096,"depth":650,"text":1097},{"id":1108,"depth":650,"text":1109},{"id":608,"depth":170,"text":609},{"id":625,"depth":170,"text":626},"Enterprise RAG is looking up authorised company files before the AI answers, with permissions respected — lookup-then-answer, not an operating system.",{"eyebrow":672,"title":1176},"Lookup before the answer",[1178,1181,1184],{"question":1179,"answer":1180},"What is enterprise RAG?","Looking up authorised company files before the AI answers, with permissions respected.",{"question":1182,"answer":1183},"Is RAG an operating system?","No. It is lookup-then-answer. It does not run the job or sign the change.",{"question":1185,"answer":1186},"What fails if permissions are ignored?","The answer includes files the person was not allowed to see. Retrieval without access control is a leak.","\u002Fblog\u002Fwhat-is-enterprise-rag",{"title":717,"description":1174},"blog\u002Fwhat-is-enterprise-rag",[686,1191,1192,1193],"rag","retrieval","knowledge","Wc0J02dNHRsHru4MY4eW8f4XWDNYrJZsy0xt5uKxalQ",{"id":1196,"title":1197,"archived":163,"authors":1198,"badge":1200,"body":1201,"date":669,"definedTerm":164,"department":164,"description":1616,"extension":172,"eyebrow":164,"faqHeader":1617,"faqs":1619,"footerBand":164,"headline":164,"image":164,"industry":164,"jobType":164,"listed":163,"location":164,"navigation":131,"openRoles":164,"pageLayout":164,"path":1629,"relatedHeading":164,"seo":1630,"series":686,"sitemap":131,"status":164,"stem":1631,"subhead":164,"tags":1632,"video":164,"whyJoin":164,"workplaceType":164,"__hash__":1636},"content\u002Fblog\u002Fwhat-is-an-agentic-workflow.md","What is an Agentic Workflow",[1199],{"name":194,"to":195},{"label":197},{"type":166,"value":1202,"toc":1592},[1203,1206,1217,1224,1227,1229,1232,1255,1258,1306,1315,1317,1320,1327,1330,1366,1369,1371,1376,1385,1390,1395,1407,1409,1415,1421,1427,1433,1443,1446,1448,1455,1462,1472,1474,1478,1481,1485,1488,1492,1495,1499,1502,1506,1511,1515,1518,1522,1525,1529,1535,1539,1542,1546,1551,1555,1560,1564,1567,1569,1575,1577],[201,1204,1205],{},"“We have an agent” often means a chat that never knows when to stop. Someone types a goal. The model keeps calling tools until the budget dies, or until a human closes the tab. There is no finish line. There is a conversation that looked busy.",[201,1207,1208,1209,1212,1213,1216],{},"A ",[205,1210,1211],{},"workflow"," has steps and a stop. An ",[205,1214,1215],{},"agentic workflow"," is a sequence of steps an AI can run toward a goal, with rules for when to stop — including a person who must approve before a live system changes.",[201,1218,1219,1223],{},[212,1220,1222],{"href":1221},"https:\u002F\u002Fwww.anthropic.com\u002Fengineering\u002Fbuilding-effective-agents","Anthropic’s note on building effective agents"," makes the same cut: workflows with tools and stop conditions, not endless chat. The note is worth reading because it is honest about the boring parts — encoding the job, bounding the tools, and deciding what “done” means — rather than treating fluency as a process.",[201,1225,1226],{},"Older automation without models is brittle but auditable. Models without a workflow are flexible but unaccountable. An agentic workflow is the attempt to get both: language where the input is messy, and a finish line where the company needs one.",[236,1228,239],{"id":238},[201,1230,1231],{},"Vendors collapse three different layers into the word “agentic”:",[241,1233,1234,1240,1246],{},[244,1235,1236,1239],{},[205,1237,1238],{},"Agentic capability."," The model can use tools, plan, and reflect. At work, this is “it can search Drive and draft a note.” It is not yet a job.",[244,1241,1242,1245],{},[205,1243,1244],{},"Agentic workflow."," A designed sequence of those capabilities, with business stop conditions. This article is about this layer. At work, this is “extract, compare to the playbook, quote the CRM fields, wait for the named signer, write or refuse.”",[244,1247,1248,1251,1252,272],{},[205,1249,1250],{},"Agent platform."," Identity, connectors, tests, and governance around many workflows. At work, this is closer to an ",[212,1253,1254],{"href":924},"enterprise AI operating system",[201,1256,1257],{},"Other terms:",[241,1259,1260,1266,1272,1281,1291,1297],{},[244,1261,1262,1265],{},[205,1263,1264],{},"Tool."," An action the AI can take: search files, query CRM, post a message. At work, a tool is a hand. Hands are not roles, and they are not stop conditions.",[244,1267,1268,1271],{},[205,1269,1270],{},"Stop condition."," Budget hit, waiting on approval, error, empty result, human cancel. “The model says it is done” is a weak stop by itself.",[244,1273,1274,1277,1278,272],{},[205,1275,1276],{},"Write-back."," The AI is allowed to change a live system, not just draft. See ",[212,1279,1280],{"href":478},"What is write-back governance",[244,1282,1283,1286,1287,272],{},[205,1284,1285],{},"Human wait."," A step in the sequence, not an interruption. See ",[212,1288,1290],{"href":1289},"\u002Fblog\u002Fwhat-is-human-in-the-loop-ai\u002F","What is human-in-the-loop AI",[244,1292,1293,1296],{},[205,1294,1295],{},"Version."," Which workflow definition ran. When policy changes, retrieval changes. Operators need to know which version ran last Tuesday.",[244,1298,1299,1302,1303,272],{},[205,1300,1301],{},"MCP."," A common plug so AI apps can use the same tools. Plumbing. It does not define your stops. See ",[212,1304,1305],{"href":905},"What is Model Context Protocol",[201,1307,1208,1308,1310,1311,1314],{},[212,1309,955],{"href":617}," is the company object that ",[342,1312,1313],{},"hosts"," the workflow: brief, connectors, people, budget, finish line. The workflow is the sequence. The workstream is the job folder. Mixing those two words is how demos skip isolation.",[236,1316,321],{"id":320},[201,1318,1319],{},"Capability demos look like workflows. They are not. A fluent plan is not a paused run waiting on approval, a failed run that did not retry a write, or a replay of which step ran.",[201,1321,1322,1323,1326],{},"It affects you if the job is ",[205,1324,1325],{},"multi-step, tool-using, and repeated"," — the opposite of one-off chat. Close checklists, renewal playbooks, and incident runbooks already have steps. Encode those. If the job is not written down, you will encode folklore and then fight the folklore.",[201,1328,1329],{},"Practical rules:",[241,1331,1332,1338,1344,1350,1356],{},[244,1333,1334,1337],{},[205,1335,1336],{},"Read-heavy workflows"," can be long. They should still finish in an artefact with sources.",[244,1339,1340,1343],{},[205,1341,1342],{},"Write-heavy workflows"," should be short after the quote: one payload, one gate, one execution, one record. Do not hide ten writes in a “cleanup agent.”",[244,1345,1346,1349],{},[205,1347,1348],{},"Human wait is a step",", not an interruption.",[244,1351,1352,1355],{},[205,1353,1354],{},"Version the workflow."," Policy and retrieval drift. Last Tuesday’s run needs a definition you can still open.",[244,1357,1358,1361,1362,272],{},[205,1359,1360],{},"Budget is a stop."," See ",[212,1363,1365],{"href":1364},"\u002Fblog\u002Fwhat-is-ai-token-economics\u002F","What is AI token economics",[201,1367,1368],{},"A mega-agent with “figure it out” as the spec is not a workflow. It is a hope.",[386,1370,389],{"id":388},[201,1372,1373,1375],{},[205,1374,394],{}," Close and forecast jobs already have checklists. An agentic workflow that posts a journal without a stop at the named signer is not “agentic.” It is unattended posting. Finance also needs spend stops so a retry loop cannot become the month’s inference bill.",[201,1377,1378,1380,1381,1384],{},[205,1379,404],{}," Customer-facing steps and anything that asserts a term need a gate before send. Air Canada’s chatbot invented a bereavement fare and the company was held to it — ",[212,1382,1383],{"href":1103},"CBC’s report"," is the cautionary case for “the workflow ended at the message.” Legal also cares that the workflow version is reconstructable.",[201,1386,1387,1389],{},[205,1388,410],{}," This is the native language: runbooks, queues, retries, and “do not proceed.” Ops should refuse workflows that cannot pause cleanly, cannot show which step failed, and cannot distinguish “waiting on a person” from “waiting on a tool error.”",[201,1391,1392,1394],{},[205,1393,416],{}," Renewal and hygiene jobs are repeated and tool-using. GTM should demand a short write path after the quote, not a weekend “cleanup” that touches hundreds of records behind one click. Time-to-approved-write is the metric, not time-to-first-plan.",[201,1396,1397,1399,1400,1403,1404,1406],{},[205,1398,422],{}," Tool belts are attack surface. Prompt injection that tricks a model into ",[342,1401,1402],{},"requesting"," a write should still die at a fail-closed gate. Importing every MCP helper into one workflow is how a demo becomes one actor with every production login. The ",[212,1405,827],{"href":826}," treats tool use as a security topic for this reason.",[386,1408,427],{"id":426},[201,1410,1411,1414],{},[205,1412,1413],{},"Chat as workflow."," A conversation that looks busy has no durable instance, no version, and no gate.",[201,1416,1417,1420],{},[205,1418,1419],{},"A checklist in a prompt."," A start. Without tools, a durable job, and a stop, it is still a prompt.",[201,1422,1423,1426],{},[205,1424,1425],{},"Replacing a stable bot."," If the job is a scheduled export, older automation is the right tool. Agentic workflows help on messy documents. They are not a prestige upgrade for a cron job.",[201,1428,1429,1432],{},[205,1430,1431],{},"Fully autonomous production."," Only for actions you would already automate without a model, plus logging. If you would not let a scheduled job do it, do not let an agent do it unattended.",[201,1434,1435,1438,1439,272],{},[205,1436,1437],{},"Multi-agent as a requirement."," A single tool-using agent can execute a workflow. Multiple agents help when duties already split. See ",[212,1440,1442],{"href":1441},"\u002Fblog\u002Fwhat-is-multi-agent-ai\u002F","What is multi-agent AI",[201,1444,1445],{},"Good looks like: named steps, bounded tools, explicit stops (including human wait and budget), versioned definitions, read-only by default, fail-closed writes. Failure looks like a flagship model with every connector and a spec that says “be helpful.”",[236,1447,484],{"id":483},[201,1449,1450,1451,272],{},"Nimbus’s delivery unit for operators is the ",[205,1452,1453],{},[212,1454,955],{"href":617},[201,1456,1457,1458,1461],{},"The mapping in everyday terms: the brief is the goal; ",[212,1459,1460],{"href":1441},"agent teams"," run the steps; connectors are the tools (default read-only); wiki is the playbook the steps must respect; governance is the wait\u002Fwrite stop; the Lifecycle Graph is the executed run. Model routing chooses the brain per step; it does not choose the stop.",[201,1463,499,1464,614,1466,619,1470,272],{},[212,1465,31],{"href":992},[212,1467,1469],{"href":1468},"\u002Fproduct\u002Fagent-teams\u002F","Agent teams",[212,1471,39],{"href":995},[236,1473,514],{"id":513},[386,1475,1477],{"id":1476},"is-a-checklist-in-a-prompt-an-agentic-workflow","Is a checklist in a prompt an agentic workflow?",[201,1479,1480],{},"It is a start. If there are no tools, no durable instance, and no gate, it is a prompt.",[386,1482,1484],{"id":1483},"how-is-this-different-from-older-robotic-automation","How is this different from older robotic automation?",[201,1486,1487],{},"Older automation executes deterministic steps. Agentic workflows add language and planning. That helps on messy documents. It also means you need tests and human gates. Do not replace a stable bot with an agent if the job is still a scheduled export.",[386,1489,1491],{"id":1490},"do-agentic-workflows-require-multiple-agents","Do agentic workflows require multiple agents?",[201,1493,1494],{},"No. A single tool-using agent can execute a workflow. Multiple agents help when duties already split in the organisation.",[386,1496,1498],{"id":1497},"can-a-workflow-be-fully-autonomous-in-production","Can a workflow be fully autonomous in production?",[201,1500,1501],{},"Only for actions you would already automate without a model, plus logging.",[386,1503,1505],{"id":1504},"where-do-tool-connection-standards-fit","Where do tool-connection standards fit?",[201,1507,1508,1509,272],{},"A common plug so AI apps can use the same tools is plumbing. It does not define your stops or approvals. See ",[212,1510,1305],{"href":905},[386,1512,1514],{"id":1513},"what-is-a-good-stop-condition-besides-the-model-is-done","What is a good stop condition besides “the model is done”?",[201,1516,1517],{},"Budget ceiling, empty retrieval, tool error, human cancel, and wait-for-named-signer. “Done” from the model is a suggestion. Encode the others.",[386,1519,1521],{"id":1520},"how-long-should-a-write-heavy-workflow-be","How long should a write-heavy workflow be?",[201,1523,1524],{},"Short after the quote. One payload, one gate, one execution, one record. Length belongs in the read and compare steps, not in a bundle of hidden mutations.",[386,1526,1528],{"id":1527},"how-do-we-version-a-workflow-when-the-wiki-changes","How do we version a workflow when the wiki changes?",[201,1530,1531,1532,1534],{},"Treat the playbook version as an input to the run. The ",[212,1533,469],{"href":270}," should cite which wiki version the steps respected. Changing policy without recording which definition ran is how Tuesday becomes unexplained.",[386,1536,1538],{"id":1537},"is-agentic-the-same-as-autonomous","Is “agentic” the same as “autonomous”?",[201,1540,1541],{},"No. Agentic means the model can plan and use tools. Autonomy is a policy about whether a person must still sign. Most production writes should not be autonomous.",[386,1543,1545],{"id":1544},"can-we-import-every-available-tool-and-let-the-model-choose","Can we import every available tool and let the model choose?",[201,1547,1548,1549,272],{},"That is a confused workflow. Least privilege applies to tools as much as to data. See ",[212,1550,618],{"href":617},[386,1552,1554],{"id":1553},"how-does-this-relate-to-human-in-the-loop","How does this relate to human-in-the-loop?",[201,1556,1557,1558,272],{},"Human wait is a first-class step. If the person is only “on the loop” with a kill switch, you have a different design. See ",[212,1559,1290],{"href":1289},[386,1561,1563],{"id":1562},"will-a-better-model-remove-the-need-for-a-workflow","Will a better model remove the need for a workflow?",[201,1565,1566],{},"Stronger models plan more fluently. They still do not know your finish line, your signer, or your budget. Fluency without stops is a more expensive loop.",[236,1568,609],{"id":608},[201,1570,1571,503,1573,272],{},[212,1572,925],{"href":924},[212,1574,1442],{"href":1441},[236,1576,626],{"id":625},[241,1578,1579,1584,1588],{},[244,1580,1581],{},[212,1582,1583],{"href":1221},"Anthropic, Building effective agents",[244,1585,1586],{},[212,1587,1149],{"href":1103},[244,1589,1590],{},[212,1591,827],{"href":826},{"title":169,"searchDepth":170,"depth":170,"links":1593},[1594,1595,1599,1600,1614,1615],{"id":238,"depth":170,"text":239},{"id":320,"depth":170,"text":321,"children":1596},[1597,1598],{"id":388,"depth":650,"text":389},{"id":426,"depth":650,"text":427},{"id":483,"depth":170,"text":484},{"id":513,"depth":170,"text":514,"children":1601},[1602,1603,1604,1605,1606,1607,1608,1609,1610,1611,1612,1613],{"id":1476,"depth":650,"text":1477},{"id":1483,"depth":650,"text":1484},{"id":1490,"depth":650,"text":1491},{"id":1497,"depth":650,"text":1498},{"id":1504,"depth":650,"text":1505},{"id":1513,"depth":650,"text":1514},{"id":1520,"depth":650,"text":1521},{"id":1527,"depth":650,"text":1528},{"id":1537,"depth":650,"text":1538},{"id":1544,"depth":650,"text":1545},{"id":1553,"depth":650,"text":1554},{"id":1562,"depth":650,"text":1563},{"id":608,"depth":170,"text":609},{"id":625,"depth":170,"text":626},"An agentic workflow is a sequence of steps an AI can run toward a goal, with a person who must approve before a live system changes.",{"eyebrow":672,"title":1618},"Steps, with a stop",[1620,1623,1626],{"question":1621,"answer":1622},"What is an agentic workflow?","A sequence of steps an AI can run toward a goal, with rules for when to stop.",{"question":1624,"answer":1625},"When must a person be in it?","Before a live system changes. The workflow can prepare the change. It cannot finish it alone.",{"question":1627,"answer":1628},"Is a chatbot flow the same thing?","A chatbot answers. An agentic workflow is allowed to take steps, which is why the stop has to be explicit.","\u002Fblog\u002Fwhat-is-an-agentic-workflow",{"title":1197,"description":1616},"blog\u002Fwhat-is-an-agentic-workflow",[686,1633,1634,1635],"agentic-workflow","agents","workstreams","jqFWGede9d2XN3pJ1cSmdtlf_WWT8s5XhX9vIrpWXKY",{"fold":1638,"id":1642,"title":1643,"archived":163,"authors":164,"badge":164,"body":1644,"date":164,"definedTerm":164,"department":164,"description":169,"extension":172,"eyebrow":164,"faqHeader":164,"faqs":164,"footerBand":1648,"headline":164,"image":164,"industry":164,"jobType":164,"listed":131,"location":164,"navigation":131,"openRoles":164,"pageLayout":164,"path":1652,"relatedHeading":164,"seo":1653,"series":164,"sitemap":163,"status":164,"stem":1654,"subhead":164,"tags":164,"video":164,"whyJoin":164,"workplaceType":164,"__hash__":1655},{"headline":1639,"description":1640,"primaryLabel":8,"primaryTo":1641,"secondaryLabel":709,"secondaryTo":12},"Run frontier AI your business actually owns.","Governed workstreams, 3,000+ integrations, and a proprietary knowledge graph. Start on Free.","\u002Fsignup?plan=free","content\u002Fshared\u002Fcta.md","Site CTAs",{"type":166,"value":1645,"toc":1646},[],{"title":169,"searchDepth":170,"depth":170,"links":1647},[],{"headline":1649,"description":1650,"primaryLabel":8,"primaryTo":1641,"secondaryLabel":1651,"secondaryTo":85},"See what governed AI looks like on your stack.","Connect your tools, run a workstream, and keep every decision on your ledger. Start on Free.","Talk to our team","\u002Fshared\u002Fcta",{"title":1643,"description":169},"shared\u002Fcta","eYqahyaPnbp8GKrWpoORbZdtmkWmgHr5F61ZHOnb8sY",1792999481329]