[{"data":1,"prerenderedAt":1720},["ShallowReactive",2],{"site-nav-content":3,"blog:/blog/what-is-ai-token-economics":163,"blog-index-copy":671,"blog:/blog/what-is-ai-token-economics:surround":692,"hiring-banner-content":1689,"site-cta-content":1701},{"header":4,"productNav":9,"nav":42,"footer":61,"askAI":116,"id":147,"title":148,"archived":149,"authors":150,"badge":150,"body":151,"date":150,"department":150,"description":155,"extension":158,"eyebrow":150,"faqHeader":150,"faqs":150,"footerBand":150,"headline":150,"image":150,"industry":150,"jobType":150,"listed":149,"location":150,"navigation":115,"openRoles":150,"pageLayout":150,"path":159,"relatedHeading":150,"seo":160,"series":150,"sitemap":115,"status":150,"stem":161,"subhead":150,"tags":150,"video":150,"whyJoin":150,"workplaceType":150,"__hash__":162},{"productLabel":5,"loginLabel":6,"contactLabel":7,"contactSalesLabel":8},"Product","Log in","Contact","Get started for free",[10,14,18,22,26,30,34,38],{"label":11,"to":12,"description":13},"Overview","/overview","Seven layers. One closed loop.",{"label":15,"to":16,"description":17},"Conflux","/product/conflux","Where your team, workstreams, and agents meet.",{"label":19,"to":20,"description":21},"Agent Teams","/product/agent-teams","Specialist teams - governed from day one.",{"label":23,"to":24,"description":25},"Lifecycle Graph","/product/lifecycle-graph","Intelligence that compounds across every interaction.",{"label":27,"to":28,"description":29},"Company Wiki","/product/wiki","Playbooks and policies where expertise stays.",{"label":31,"to":32,"description":33},"Workstreams","/product/workstreams","From brief to signed-off deliverable on one canvas.",{"label":35,"to":36,"description":37},"Perception Console","/product/perception","Ask your whole business in plain English.",{"label":39,"to":40,"description":41},"Governance","/product/governance","Frontier AI you can actually sign off on.",[43,46,49,52,55,58],{"label":44,"to":45},"Models","/models",{"label":47,"to":48},"Pricing","/pricing",{"label":50,"to":51},"Integrations","/integrations",{"label":53,"to":54},"Security","/security",{"label":56,"to":57},"Partners","/partners",{"label":59,"to":60},"Insights","/blog",{"productHeading":5,"companyHeading":62,"legalHeading":63,"docsLabel":64,"docsUrl":65,"statementLines":66,"copyright":69,"companyLinks":70,"legalLinks":85,"socialLinks":95,"bottomLinks":105},"Company","Legal","Docs","https://docs.gonimbus.ai",[67,68],"Stop training someone else's model.","Control your AI.","© 2026 Nimbus Intelligence, Inc. All rights reserved.",[71,72,73,74,75,77,80,83],{"label":47,"to":48},{"label":50,"to":51},{"label":53,"to":54},{"label":59,"to":60},{"label":76,"to":57},"Partner Program",{"label":78,"to":79},"Careers","/careers",{"label":81,"to":82},"System status","/status",{"label":7,"to":84},"/contact",[86,89,92],{"label":87,"to":88},"Terms of Service","/terms",{"label":90,"to":91},"Privacy Policy","/privacy",{"label":93,"to":94},"Compliance","/compliance",[96,99,102],{"label":97,"href":98},"LinkedIn","https://www.linkedin.com/company/gonimbusai/",{"label":100,"href":101},"X","https://x.com/gonimbusai",{"label":103,"href":104},"Instagram","https://www.instagram.com/gonimbus_ai/",[106,108,110,111,112],{"label":107,"to":88},"Terms",{"label":109,"to":91},"Privacy",{"label":93,"to":94},{"label":81,"to":82},{"label":113,"to":114,"external":115},"LLMs.txt","/llms.txt",true,{"text":117,"prompt":118},"Ask AI about Nimbus",{"I'm researching enterprise intelligence platforms and want to know how Nimbus combines perception, collaboration, and autonomous agents to drive strategic decision-making":119,"platforms":121},{" Summarize the highlights from Nimbus's website":120},"https://gonimbus.ai",[122,127,132,137,142],{"name":123,"label":124,"icon":125,"hrefPrefix":126},"chatgpt","ChatGPT","simple-icons:openai","https://chatgpt.com/?prompt=",{"name":128,"label":129,"icon":130,"hrefPrefix":131},"perplexity","Perplexity","mdi:magnify","https://www.perplexity.ai/search/new?q=",{"name":133,"label":134,"icon":135,"hrefPrefix":136},"grok","Grok","simple-icons:x","https://x.com/i/grok?text=",{"name":138,"label":139,"icon":140,"hrefPrefix":141},"claude","Claude","simple-icons:anthropic","https://claude.ai/new?q=",{"name":143,"label":144,"icon":145,"hrefPrefix":146},"google-ai","Google AI","simple-icons:google","https://www.google.com/search?udm=50&aep=11&q=","content/shared/nav.md","Site navigation",false,null,{"type":152,"value":153,"toc":154},"minimark",[],{"title":155,"searchDepth":156,"depth":156,"links":157},"",2,[],"md","/shared/nav",{"title":148,"description":155},"shared/nav","p6IsjEfjcrwkjwvGsPcuEkpoJSeUn2vJXPTxLWQfw9M",{"id":164,"title":165,"archived":149,"authors":166,"badge":169,"body":171,"date":660,"department":150,"description":661,"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":662,"relatedHeading":150,"seo":663,"series":664,"sitemap":115,"status":150,"stem":665,"subhead":150,"tags":666,"video":150,"whyJoin":150,"workplaceType":150,"__hash__":670},"content/blog/what-is-ai-token-economics.md","What is AI Token Economics",[167],{"name":168,"to":120},"Nimbus Research",{"label":170},"Explainer",{"type":152,"value":172,"toc":635},[173,196,202,205,213,218,312,323,327,330,350,353,376,379,384,394,400,406,412,421,425,431,441,447,455,467,479,483,486,492,502,506,510,513,517,520,524,527,531,536,540,543,547,550,554,559,563,566,570,573,577,584,588,593,597,600,604,611,615],[174,175,176,177,181,182,189,190,195],"p",{},"A ",[178,179,180],"strong",{},"token"," is a chunk of text the model reads or writes. You pay per chunk. Different models cost different amounts. Input, output, and sometimes tools all meter differently. ",[183,184,188],"a",{"href":185,"rel":186},"https://openai.com/api/pricing/",[187],"nofollow","OpenAI"," and ",[183,191,194],{"href":192,"rel":193},"https://www.anthropic.com/pricing",[187],"Anthropic"," publish those ladders. Finance still cannot run the business on “12 million tokens of vendor A’s flagship.”",[174,197,198,201],{},[178,199,200],{},"AI token economics"," is treating that usage like a real budget: measuring, allocating, controlling, and attributing spend so operators can quote before a run, cap during it, and attribute after it — instead of a slide that says “unlimited AI.”",[174,203,204],{},"Without it, organisations either freeze (no production AI) or send every small task to the most expensive model until the bill becomes a board slide.",[174,206,207,208,212],{},"The unit problem is the same one cloud had in its first decade: a metered resource sold with a headcount story. Seat licences predict people. Inference predicts work. When those two are collapsed into “unlimited,” the next chunk ",[209,210,211],"em",{},"feels"," free, so people pick the flagship every time. The ladder did not disappear. It hid.",[214,215,217],"h2",{"id":216},"words-youll-hear","Words you’ll hear",[219,220,221,228,234,245,255,261,267,278,284,295,301],"ul",{},[222,223,224,227],"li",{},[178,225,226],{},"Token."," A piece of text the model processes. Not a business unit. At work, a long wiki dump and a short field extract are wildly different token counts for the same “question.”",[222,229,230,233],{},[178,231,232],{},"Seat licence."," Predictable cost per person. Often marketed as “unlimited.” The underlying work is still metered.",[222,235,236,239,240,244],{},[178,237,238],{},"Pass-through API bill."," Each team has keys. Simple. Invites key sprawl and ",[183,241,243],{"href":242},"what-is-shadow-ai","shadow AI"," on personal keys. At work, the invoice lands in engineering while go-to-market did the looping.",[222,246,247,250,251,254],{},[178,248,249],{},"Quote."," A number ",[209,252,253],{},"before"," they run. At work, this is what makes a brief a decision rather than a surprise.",[222,256,257,260],{},[178,258,259],{},"Cap / ceiling."," A hard stop. The loop cannot spend past it. At work, weekend agent loops die here instead of in next month’s cloud bill.",[222,262,263,266],{},[178,264,265],{},"Pool."," Organisation-level allowance. At work, one department should not be able to burn the company pool on a vanity run.",[222,268,269,272,273,277],{},[178,270,271],{},"Attribution."," Chargeback by job, not “the AI bill.” At work, finance can ask which ",[183,274,276],{"href":275},"what-is-an-ai-workstream","workstream"," consumed the units.",[222,279,280,283],{},[178,281,282],{},"NTU (Nimbus Token Unit)."," Nimbus’s normalised work credit for completed AI activity — analysis, tools, runs, writes — sitting above raw provider tokens. Everyday questions can be included; heavier work consumes pool credits. Finance gets one tape measure across vendors and steps.",[222,285,286,289,290,294],{},[178,287,288],{},"Model routing."," Cheaper model for simple steps, stronger only when needed. See ",[183,291,293],{"href":292},"what-is-model-routing","What is model routing",". At work, classify-this-ticket should not pay flagship rates.",[222,296,297,300],{},[178,298,299],{},"Context window."," How much text the model can see at once. Dumping the whole Drive into context is an economic choice, not a quality strategy.",[222,302,303,306,307,311],{},[178,304,305],{},"Stop condition."," Budget hit, empty result, human cancel. Agent loops can dominate the bill without improving the artefact. See ",[183,308,310],{"href":309},"what-is-an-agentic-workflow","What is an agentic workflow",".",[174,313,314,315,318,319,322],{},"The point is ",[178,316,317],{},"value per unit",", not minimum units regardless of outcome. Caching, wiki citations, and memory should make the ",[209,320,321],{},"same"," outcome cheaper over time. If unit cost of an approved update never falls, you are re-deriving folklore every run.",[214,324,326],{"id":325},"why-you-should-care","Why you should care",[174,328,329],{},"It affects you if you:",[219,331,332,338,344],{},[222,333,334,337],{},[178,335,336],{},"Own the budget."," Surprise invoices arrive after agents looped all weekend.",[222,339,340,343],{},[178,341,342],{},"Run the work."," You should see a number before you commit, not a lecture after.",[222,345,346,349],{},[178,347,348],{},"Are tempted to shame people for using AI."," Shame drives personal keys. Cap the official path so it is safe to use.",[174,351,352],{},"Practical rhythm:",[219,354,355,364,370],{},[222,356,357,360,361,363],{},[178,358,359],{},"Name the run."," Unnamed chats cannot be attributed. That is what a ",[183,362,276],{"href":275}," is for.",[222,365,366,369],{},[178,367,368],{},"Separate exploration from production."," Sandboxes can have tighter caps and cheaper default routes.",[222,371,372,375],{},[178,373,374],{},"Review unit cost of outcomes"," — approved updates per unit — not tokens in the abstract.",[174,377,378],{},"Anti-pattern: a single corporate API key in a wiki, no per-job cap, monthly surprise. That is an unmetered utility.",[380,381,383],"h3",{"id":382},"what-changes-by-role","What changes by role",[174,385,386,389,390,393],{},[178,387,388],{},"Finance."," You need a quote, a ceiling, and a chargeback dimension that matches how the business already thinks — by job, department, or cost centre — not by vendor token type. Multi-vendor ladders are incomparable until you normalise. NTU is that normalisation in Nimbus. Finance should also see ",[209,391,392],{},"stops",": a cap that fired is a successful control, not a failed project.",[174,395,396,399],{},[178,397,398],{},"Legal."," Spend logs are not only money. They are a map of which data classes went to which provider. Uncapped personal keys are a processing-agreement gap. Legal will also ask whether you can stop a run, not only whether you can pay for it.",[174,401,402,405],{},[178,403,404],{},"Operations."," Caps are operational stops, like a queue limit. Ops needs to know whether a paused run is waiting on a person or waiting on budget. Mixing those two in one “it failed” status is how you get the wrong pager.",[174,407,408,411],{},[178,409,410],{},"Go-to-market."," GTM feels the quality-versus-cost trade first. A compact model that extracts fields is usually enough. A flagship model that argues a clause may be worth it. Without routing and quotes, GTM either hoards “the best model” or gets blamed for the bill. Neither produces better pipeline hygiene.",[174,413,414,417,418,420],{},[178,415,416],{},"Security."," API keys are credentials. Personal keys in browser plugins are ",[183,419,243],{"href":242},". A pooled official path with per-workstream ceilings reduces key sprawl. Spend spikes can also be an anomaly signal — a loop that never stops is sometimes a bug, sometimes a prompt-injection success.",[380,422,424],{"id":423},"what-people-get-wrong","What people get wrong",[174,426,427,430],{},[178,428,429],{},"“Unlimited” as a strategy."," Seats hide the ladder. They do not delete it. Heavy agentic work will still surface as a true-up, a throttle, or a degraded model.",[174,432,433,436,437,440],{},[178,434,435],{},"Punishing usage."," Chargeback without a sanctioned path recreates personal keys. Celebrate lower units ",[209,438,439],{},"per artefact"," as playbooks and memory compound.",[174,442,443,446],{},[178,444,445],{},"Tokens as the KPI."," Tokens measure consumption. Outcomes measure value. A cheap run that produces a rejected write is still waste. A dearer run that produces one approved journal may be fine.",[174,448,449,452,453,311],{},[178,450,451],{},"One model for everything."," That is a routing failure dressed as quality culture. See ",[183,454,293],{"href":292},[174,456,457,460,461,466],{},[178,458,459],{},"No stop on loops."," ",[183,462,465],{"href":463,"rel":464},"https://www.anthropic.com/engineering/building-effective-agents",[187],"Anthropic’s note on building effective agents"," treats workflows with stop conditions as the grown-up shape. Economics is one of those stops.",[174,468,469,470,189,474,478],{},"Good looks like: named jobs, quotes before commit, hard ceilings, routing policy, attribution, and falling unit cost as the ",[183,471,473],{"href":472},"what-is-a-company-wiki-for-ai-agents","wiki",[183,475,477],{"href":476},"what-is-a-lifecycle-graph","lifecycle graph"," reduce re-derivation. Failure looks like a shared key, a flagship default, and a board slide titled “AI spend.”",[214,480,482],{"id":481},"how-this-shows-up-in-nimbus","How this shows up in Nimbus",[174,484,485],{},"Workstreams show quotes and ceilings before runs. Orgs draw from a pooled NTU allowance. Routing is a policy, not a dropdown labelled “best.” Memory and wiki reduce re-derivation, which is how unit cost of an outcome should fall over time.",[174,487,488,489,491],{},"Everyday questions can sit inside the allowance; heavier analysis, tools, and writes consume pool credits. The ",[183,490,23],{"href":476}," can record spend as part of the chain, so “the run stopped because the ceiling was hit” is a causal fact.",[174,493,494,495,189,497,499,500,311],{},"See ",[183,496,44],{"href":45},[183,498,293],{"href":292},". Product context: ",[183,501,31],{"href":32},[214,503,505],{"id":504},"questions-people-actually-ask","Questions people actually ask",[380,507,509],{"id":508},"why-cant-we-just-pay-seats-and-call-it-unlimited","Why can’t we just pay seats and call it unlimited?",[174,511,512],{},"Seats predict headcount. Production AI spend is inference, tools, and writes. “Unlimited” hides the ladder; it does not delete it.",[380,514,516],{"id":515},"what-should-finance-actually-see","What should finance actually see?",[174,518,519],{},"A quote before commit, a cap during the run, and attribution by job afterwards — in one unit they can compare across vendors and steps.",[380,521,523],{"id":522},"wont-cheaper-models-get-worse-answers","Won’t cheaper models get worse answers?",[174,525,526],{},"For extract and classify, often no. For hard judgment, often yes. That is a routing policy, not a religion. Measure reject rates on the job, not vibes.",[380,528,530],{"id":529},"do-we-punish-teams-for-using-ai","Do we punish teams for using AI?",[174,532,533,534,440],{},"No. Punishing usage revives shadow AI. Celebrate lower units ",[209,535,439],{},[380,537,539],{"id":538},"what-is-an-ntu-in-plain-language","What is an NTU in plain language?",[174,541,542],{},"A normalised work credit above raw provider tokens, so a finance partner is not asked to compare “vendor A input tokens” with “vendor B output tokens” plus tool calls. In Nimbus, completed activity — analysis, tools, runs, writes — is what consumes the unit.",[380,544,546],{"id":545},"should-every-chat-be-billed-to-a-cost-centre","Should every chat be billed to a cost centre?",[174,548,549],{},"Named production jobs, yes. Tiny sanctioned copilots for personal drafting can live on a lighter path. The failure is mixing them so neither can be capped.",[380,551,553],{"id":552},"how-do-agent-loops-blow-the-budget","How do agent loops blow the budget?",[174,555,556,557,311],{},"They call tools, re-read context, and retry without a finish line. Without a ceiling and a stop condition, “being thorough” is an unbounded loop. See ",[183,558,310],{"href":309},[380,560,562],{"id":561},"is-caching-the-same-as-token-economics","Is caching the same as token economics?",[174,564,565],{},"Caching is a tactic. Economics is the management system: quote, cap, attribute, route. Caching without attribution still leaves you unable to explain the bill.",[380,567,569],{"id":568},"do-we-need-a-data-warehouse-to-do-this","Do we need a data warehouse to do this?",[174,571,572],{},"You need events at run time. A warehouse can hold copies for reporting. It cannot quote a run that has not emitted a number yet.",[380,574,576],{"id":575},"how-does-this-relate-to-write-back","How does this relate to write-back?",[174,578,579,580,311],{},"Writes are usually a small number of tokens and a large operational risk. Do not use spend as a substitute for a named signer. Do use spend as a stop so a looping agent cannot keep proposing writes all weekend. See ",[183,581,583],{"href":582},"what-is-write-back-governance","What is write-back governance",[380,585,587],{"id":586},"can-we-lock-one-vendor-to-simplify-pricing","Can we lock one vendor to simplify pricing?",[174,589,590,591,311],{},"You can. You will pay for it in price, outages, and lock-in. A normalised unit plus routing is how finance keeps a second tape measure. See ",[183,592,293],{"href":292},[380,594,596],{"id":595},"why-not-just-set-a-monthly-company-cap","Why not just set a monthly company cap?",[174,598,599],{},"A company cap without per-job attribution is a shared kitchen. The loudest workflow starves the others, and nobody can say which job did it.",[214,601,603],{"id":602},"related-reading","Related reading",[174,605,606,189,608,311],{},[183,607,293],{"href":292},[183,609,610],{"href":275},"What is an AI workstream",[214,612,614],{"id":613},"sources","Sources",[219,616,617,623,629],{},[222,618,619],{},[183,620,622],{"href":185,"rel":621},[187],"OpenAI API pricing",[222,624,625],{},[183,626,628],{"href":192,"rel":627},[187],"Anthropic pricing",[222,630,631],{},[183,632,634],{"href":463,"rel":633},[187],"Anthropic, Building effective agents",{"title":155,"searchDepth":156,"depth":156,"links":636},[637,638,643,644,658,659],{"id":216,"depth":156,"text":217},{"id":325,"depth":156,"text":326,"children":639},[640,642],{"id":382,"depth":641,"text":383},3,{"id":423,"depth":641,"text":424},{"id":481,"depth":156,"text":482},{"id":504,"depth":156,"text":505,"children":645},[646,647,648,649,650,651,652,653,654,655,656,657],{"id":508,"depth":641,"text":509},{"id":515,"depth":641,"text":516},{"id":522,"depth":641,"text":523},{"id":529,"depth":641,"text":530},{"id":538,"depth":641,"text":539},{"id":545,"depth":641,"text":546},{"id":552,"depth":641,"text":553},{"id":561,"depth":641,"text":562},{"id":568,"depth":641,"text":569},{"id":575,"depth":641,"text":576},{"id":586,"depth":641,"text":587},{"id":595,"depth":641,"text":596},{"id":602,"depth":156,"text":603},{"id":613,"depth":156,"text":614},"2026-08-17","AI token economics is treating AI usage like a real budget: you pay per chunk of text the model reads and writes, so finance can quote, cap, and attribute spend instead of hoping for “unlimited AI.”","/blog/what-is-ai-token-economics",{"title":165,"description":661},"explainer","blog/what-is-ai-token-economics",[664,667,668,669],"token-economics","ntu","model-routing","kgDCkz6CVtDSpJc699QfUH1nMlvaVezkE2PrGzu2Puc",{"hero":672,"id":674,"title":675,"archived":149,"authors":150,"badge":150,"body":676,"date":150,"department":150,"description":680,"extension":158,"eyebrow":681,"faqHeader":150,"faqs":150,"footerBand":682,"headline":150,"image":150,"industry":150,"jobType":150,"listed":149,"location":150,"navigation":115,"openRoles":150,"pageLayout":150,"path":60,"relatedHeading":688,"seo":689,"series":150,"sitemap":115,"status":150,"stem":690,"subhead":150,"tags":150,"video":150,"whyJoin":150,"workplaceType":150,"__hash__":691},{"filename":673},"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":677,"toc":678},[],{"title":155,"searchDepth":156,"depth":156,"links":679},[],"Deep dives into pre-cognitive intelligence, sentient enterprises, and the evolving landscape of AI-driven business transformation.","Latest Research",{"headline":683,"description":684,"primaryLabel":685,"primaryTo":686,"secondaryLabel":687,"secondaryTo":12},"Stay at the frontier.","Subscribe for product updates and new insights.","Subscribe","/newsletter","Explore the platform","More research",{"title":675,"description":680},"blog/index","eK1RCXdDW8nfLSyKRXGB1mJm9FAmhAO6GWXwNKOMNVE",[693,1131],{"id":694,"title":695,"archived":149,"authors":696,"badge":698,"body":699,"date":660,"department":150,"description":1122,"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":1123,"relatedHeading":150,"seo":1124,"series":664,"sitemap":115,"status":150,"stem":1125,"subhead":150,"tags":1126,"video":150,"whyJoin":150,"workplaceType":150,"__hash__":1130},"content/blog/what-is-an-agentic-workflow.md","What is an Agentic Workflow",[697],{"name":168,"to":120},{"label":170},{"type":152,"value":700,"toc":1098},[701,704,714,720,723,725,728,752,755,802,811,813,816,823,826,862,865,867,872,883,888,893,908,910,916,922,928,934,944,947,949,956,963,974,976,980,983,987,990,994,997,1001,1004,1008,1013,1017,1020,1024,1027,1031,1037,1041,1044,1048,1053,1057,1062,1066,1069,1071,1078,1080],[174,702,703],{},"“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.",[174,705,176,706,709,710,713],{},[178,707,708],{},"workflow"," has steps and a stop. An ",[178,711,712],{},"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.",[174,715,716,719],{},[183,717,465],{"href":463,"rel":718},[187]," 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.",[174,721,722],{},"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.",[214,724,217],{"id":216},[174,726,727],{},"Vendors collapse three different layers into the word “agentic”:",[219,729,730,736,742],{},[222,731,732,735],{},[178,733,734],{},"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.",[222,737,738,741],{},[178,739,740],{},"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.”",[222,743,744,747,748,311],{},[178,745,746],{},"Agent platform."," Identity, connectors, tests, and governance around many workflows. At work, this is closer to an ",[183,749,751],{"href":750},"what-is-an-enterprise-ai-operating-system","enterprise AI operating system",[174,753,754],{},"Other terms:",[219,756,757,763,768,776,786,792],{},[222,758,759,762],{},[178,760,761],{},"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.",[222,764,765,767],{},[178,766,305],{}," Budget hit, waiting on approval, error, empty result, human cancel. “The model says it is done” is a weak stop by itself.",[222,769,770,773,774,311],{},[178,771,772],{},"Write-back."," The AI is allowed to change a live system, not just draft. See ",[183,775,583],{"href":582},[222,777,778,781,782,311],{},[178,779,780],{},"Human wait."," A step in the sequence, not an interruption. See ",[183,783,785],{"href":784},"what-is-human-in-the-loop-ai","What is human-in-the-loop AI",[222,787,788,791],{},[178,789,790],{},"Version."," Which workflow definition ran. When policy changes, retrieval changes. Operators need to know which version ran last Tuesday.",[222,793,794,797,798,311],{},[178,795,796],{},"MCP."," A common plug so AI apps can use the same tools. Plumbing. It does not define your stops. See ",[183,799,801],{"href":800},"what-is-model-context-protocol","What is Model Context Protocol",[174,803,176,804,806,807,810],{},[183,805,276],{"href":275}," is the company object that ",[209,808,809],{},"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.",[214,812,326],{"id":325},[174,814,815],{},"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.",[174,817,818,819,822],{},"It affects you if the job is ",[178,820,821],{},"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.",[174,824,825],{},"Practical rules:",[219,827,828,834,840,846,852],{},[222,829,830,833],{},[178,831,832],{},"Read-heavy workflows"," can be long. They should still finish in an artefact with sources.",[222,835,836,839],{},[178,837,838],{},"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.”",[222,841,842,845],{},[178,843,844],{},"Human wait is a step",", not an interruption.",[222,847,848,851],{},[178,849,850],{},"Version the workflow."," Policy and retrieval drift. Last Tuesday’s run needs a definition you can still open.",[222,853,854,857,858,311],{},[178,855,856],{},"Budget is a stop."," See ",[183,859,861],{"href":860},"what-is-ai-token-economics","What is AI token economics",[174,863,864],{},"A mega-agent with “figure it out” as the spec is not a workflow. It is a hope.",[380,866,383],{"id":382},[174,868,869,871],{},[178,870,388],{}," 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.",[174,873,874,876,877,882],{},[178,875,398],{}," 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 — ",[183,878,881],{"href":879,"rel":880},"https://www.cbc.ca/news/canada/british-columbia/air-canada-chatbot-lawsuit-1.7116416",[187],"CBC’s report"," is the cautionary case for “the workflow ended at the message.” Legal also cares that the workflow version is reconstructable.",[174,884,885,887],{},[178,886,404],{}," 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.”",[174,889,890,892],{},[178,891,410],{}," 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.",[174,894,895,897,898,901,902,907],{},[178,896,416],{}," Tool belts are attack surface. Prompt injection that tricks a model into ",[209,899,900],{},"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 ",[183,903,906],{"href":904,"rel":905},"https://genai.owasp.org/llm-top-10/",[187],"OWASP Top 10 for LLM applications"," treats tool use as a security topic for this reason.",[380,909,424],{"id":423},[174,911,912,915],{},[178,913,914],{},"Chat as workflow."," A conversation that looks busy has no durable instance, no version, and no gate.",[174,917,918,921],{},[178,919,920],{},"A checklist in a prompt."," A start. Without tools, a durable job, and a stop, it is still a prompt.",[174,923,924,927],{},[178,925,926],{},"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.",[174,929,930,933],{},[178,931,932],{},"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.",[174,935,936,939,940,311],{},[178,937,938],{},"Multi-agent as a requirement."," A single tool-using agent can execute a workflow. Multiple agents help when duties already split. See ",[183,941,943],{"href":942},"what-is-multi-agent-ai","What is multi-agent AI",[174,945,946],{},"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.”",[214,948,482],{"id":481},[174,950,951,952,311],{},"Nimbus’s delivery unit for operators is the ",[178,953,954],{},[183,955,276],{"href":275},[174,957,958,959,962],{},"The mapping in everyday terms: the brief is the goal; ",[183,960,961],{"href":942},"agent teams"," run the steps; connectors are the tools (default read-only); wiki is the playbook the steps must respect; governance is the wait/write stop; the Lifecycle Graph is the executed run. Model routing chooses the brain per step; it does not choose the stop.",[174,964,494,965,967,968,971,972,311],{},[183,966,31],{"href":32},", ",[183,969,970],{"href":20},"Agent teams",", and ",[183,973,39],{"href":40},[214,975,505],{"id":504},[380,977,979],{"id":978},"is-a-checklist-in-a-prompt-an-agentic-workflow","Is a checklist in a prompt an agentic workflow?",[174,981,982],{},"It is a start. If there are no tools, no durable instance, and no gate, it is a prompt.",[380,984,986],{"id":985},"how-is-this-different-from-older-robotic-automation","How is this different from older robotic automation?",[174,988,989],{},"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.",[380,991,993],{"id":992},"do-agentic-workflows-require-multiple-agents","Do agentic workflows require multiple agents?",[174,995,996],{},"No. A single tool-using agent can execute a workflow. Multiple agents help when duties already split in the organisation.",[380,998,1000],{"id":999},"can-a-workflow-be-fully-autonomous-in-production","Can a workflow be fully autonomous in production?",[174,1002,1003],{},"Only for actions you would already automate without a model, plus logging.",[380,1005,1007],{"id":1006},"where-do-tool-connection-standards-fit","Where do tool-connection standards fit?",[174,1009,1010,1011,311],{},"A common plug so AI apps can use the same tools is plumbing. It does not define your stops or approvals. See ",[183,1012,801],{"href":800},[380,1014,1016],{"id":1015},"what-is-a-good-stop-condition-besides-the-model-is-done","What is a good stop condition besides “the model is done”?",[174,1018,1019],{},"Budget ceiling, empty retrieval, tool error, human cancel, and wait-for-named-signer. “Done” from the model is a suggestion. Encode the others.",[380,1021,1023],{"id":1022},"how-long-should-a-write-heavy-workflow-be","How long should a write-heavy workflow be?",[174,1025,1026],{},"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.",[380,1028,1030],{"id":1029},"how-do-we-version-a-workflow-when-the-wiki-changes","How do we version a workflow when the wiki changes?",[174,1032,1033,1034,1036],{},"Treat the playbook version as an input to the run. The ",[183,1035,477],{"href":476}," should cite which wiki version the steps respected. Changing policy without recording which definition ran is how Tuesday becomes unexplained.",[380,1038,1040],{"id":1039},"is-agentic-the-same-as-autonomous","Is “agentic” the same as “autonomous”?",[174,1042,1043],{},"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.",[380,1045,1047],{"id":1046},"can-we-import-every-available-tool-and-let-the-model-choose","Can we import every available tool and let the model choose?",[174,1049,1050,1051,311],{},"That is a confused workflow. Least privilege applies to tools as much as to data. See ",[183,1052,610],{"href":275},[380,1054,1056],{"id":1055},"how-does-this-relate-to-human-in-the-loop","How does this relate to human-in-the-loop?",[174,1058,1059,1060,311],{},"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 ",[183,1061,785],{"href":784},[380,1063,1065],{"id":1064},"will-a-better-model-remove-the-need-for-a-workflow","Will a better model remove the need for a workflow?",[174,1067,1068],{},"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.",[214,1070,603],{"id":602},[174,1072,1073,189,1076,311],{},[183,1074,1075],{"href":750},"What is an enterprise AI operating system",[183,1077,943],{"href":942},[214,1079,614],{"id":613},[219,1081,1082,1087,1093],{},[222,1083,1084],{},[183,1085,634],{"href":463,"rel":1086},[187],[222,1088,1089],{},[183,1090,1092],{"href":879,"rel":1091},[187],"CBC, Air Canada chatbot lawsuit",[222,1094,1095],{},[183,1096,906],{"href":904,"rel":1097},[187],{"title":155,"searchDepth":156,"depth":156,"links":1099},[1100,1101,1105,1106,1120,1121],{"id":216,"depth":156,"text":217},{"id":325,"depth":156,"text":326,"children":1102},[1103,1104],{"id":382,"depth":641,"text":383},{"id":423,"depth":641,"text":424},{"id":481,"depth":156,"text":482},{"id":504,"depth":156,"text":505,"children":1107},[1108,1109,1110,1111,1112,1113,1114,1115,1116,1117,1118,1119],{"id":978,"depth":641,"text":979},{"id":985,"depth":641,"text":986},{"id":992,"depth":641,"text":993},{"id":999,"depth":641,"text":1000},{"id":1006,"depth":641,"text":1007},{"id":1015,"depth":641,"text":1016},{"id":1022,"depth":641,"text":1023},{"id":1029,"depth":641,"text":1030},{"id":1039,"depth":641,"text":1040},{"id":1046,"depth":641,"text":1047},{"id":1055,"depth":641,"text":1056},{"id":1064,"depth":641,"text":1065},{"id":602,"depth":156,"text":603},{"id":613,"depth":156,"text":614},"An 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.","/blog/what-is-an-agentic-workflow",{"title":695,"description":1122},"blog/what-is-an-agentic-workflow",[664,1127,1128,1129],"agentic-workflow","agents","workstreams","OD070bDpNh56Kq9pajGR0tbzXSES_KnS9029g0lELm8",{"id":1132,"title":1133,"archived":149,"authors":1134,"badge":1136,"body":1137,"date":660,"department":150,"description":1680,"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":1681,"relatedHeading":150,"seo":1682,"series":664,"sitemap":115,"status":150,"stem":1683,"subhead":150,"tags":1684,"video":150,"whyJoin":150,"workplaceType":150,"__hash__":1688},"content/blog/what-is-ai-governance.md","What is AI Governance",[1135],{"name":168,"to":120},{"label":170},{"type":152,"value":1138,"toc":1656},[1139,1146,1149,1152,1169,1172,1186,1188,1279,1289,1291,1299,1301,1333,1336,1364,1370,1372,1377,1388,1393,1398,1407,1409,1415,1421,1427,1435,1441,1447,1464,1466,1469,1476,1489,1498,1500,1504,1512,1516,1519,1523,1530,1534,1537,1541,1544,1548,1555,1559,1562,1566,1569,1573,1576,1580,1583,1587,1592,1596,1599,1601,1609,1611],[174,1140,1141,1142,1145],{},"AI governance is the set of rules, ",[178,1143,1144],{},"enforced in the software people actually use",", that decide who may use which AI, on which company data, and whether that AI is allowed to change a live business system — plus a record of what happened afterwards.",[174,1147,1148],{},"A training video is not that. An acceptable-use PDF is not that. An admin toggle the model can ignore is not that. If an unapproved change can still succeed, you have guidance, not governance.",[174,1150,1151],{},"People use the phrase for three different things, and they get mixed up:",[1153,1154,1155,1163,1166],"ol",{},[222,1156,1157,1158,311],{},"A public commitment — for example the ",[183,1159,1162],{"href":1160,"rel":1161},"https://oecd.ai/en/ai-principles",[187],"OECD AI Principles",[222,1164,1165],{},"A company committee with a risk register.",[222,1167,1168],{},"The runtime that actually stops a change to CRM, ERP, or a customer message.",[174,1170,1171],{},"All three are real. Only the third one would have blocked an unlogged field change that later showed up in a forecast.",[174,1173,1174,1179,1180,1185],{},[183,1175,1178],{"href":1176,"rel":1177},"https://www.gartner.com/en/articles/ai-governance-trism",[187],"Gartner’s TRiSM"," language is about that third layer: trust, risk, and security around the systems that run — not a quarterly slide about principles. The ",[183,1181,1184],{"href":1182,"rel":1183},"https://www.nist.gov/itl/ai-risk-management-framework",[187],"NIST AI Risk Management Framework"," says the same thing in public-sector language: Govern, Map, Measure, Manage. Mapping systems and measuring incidents still fail if the product people click can write to Salesforce without a named signer.",[214,1187,217],{"id":216},[219,1189,1190,1196,1209,1218,1224,1230,1245,1254,1266],{},[222,1191,1192,1195],{},[178,1193,1194],{},"Live business system."," CRM, ERP, HR, billing — the tools that hold official numbers and customer records. At work, this is where a fluent sentence becomes a fact other teams will inherit.",[222,1197,1198,1201,1202,1205,1206,1208],{},[178,1199,1200],{},"Write / write-back."," The AI is allowed to ",[209,1203,1204],{},"change"," that system, not only draft a suggestion. See ",[183,1207,583],{"href":582},". At work, a next-step note and an Amount field are not the same risk class.",[222,1210,1211,1214,1215,1217],{},[178,1212,1213],{},"Human-in-the-loop."," A person must approve before the job can finish. See ",[183,1216,785],{"href":784},". At work, the gate shows the payload in the language of the live system, not a wall of prompt text.",[222,1219,1220,1223],{},[178,1221,1222],{},"Named signer."," The identity that authorised the change. At work, “someone in the channel clicked yes” is not a signer.",[222,1225,1226,1229],{},[178,1227,1228],{},"Fail-closed."," Missing approval means nothing happens. Fail-open means the change goes through unless someone happens to stop it.",[222,1231,1232,1235,1236,1238,1239,1244],{},[178,1233,1234],{},"DPIA."," A data-protection impact assessment — thinking through purpose, risk, and personal data ",[209,1237,253],{}," you turn a tool loose. ",[183,1240,1243],{"href":1241,"rel":1242},"https://ico.org.uk/for-organisations/uk-gdpr-guidance-and-resources/artificial-intelligence/",[187],"UK ICO guidance on AI and data protection"," still wants a lawful basis and purpose when the “user” is an AI.",[222,1246,1247,1250,1251,311],{},[178,1248,1249],{},"Shadow AI."," Personal ChatGPT for work because the official path is missing. See ",[183,1252,1253],{"href":242},"What is shadow AI",[222,1255,1256,1259,1260,1265],{},[178,1257,1258],{},"Inventory."," A list of where AI actually runs. The ",[183,1261,1264],{"href":1262,"rel":1263},"https://www.justice.gov/media/1373026/dl",[187],"US plan described in OMB M-24-10"," puts a named owner and an inventory first, not a PDF.",[222,1267,1268,1271,1272,1275,1276,1278],{},[178,1269,1270],{},"Least privilege."," Only the data and tools required for ",[209,1273,1274],{},"this"," job. A ",[183,1277,276],{"href":275}," is how that instinct becomes a company object.",[174,1280,1281,1282,1285,1286,1288],{},"Model safety is adjacent and different. Safety is about what a model will say in the abstract. Enterprise governance is about what ",[209,1283,1284],{},"your"," people and tools may do with ",[209,1287,1284],{}," systems and data. You can have a carefully aligned model and still have ungoverned CRM writes.",[214,1290,326],{"id":325},[174,1292,1293,1294,1298],{},"Without working rules, AI becomes a side effect. A field moves. A journal posts. A customer is told a policy the company does not hold. Nobody can say who allowed it. In February 2024, a British Columbia tribunal held Air Canada responsible for a chatbot that invented a bereavement-fare policy. ",[183,1295,1297],{"href":879,"rel":1296},[187],"CBC reported"," that the airline’s argument — the chatbot is a separate legal entity — failed. A customer-facing commitment without a working gate is still the company’s commitment.",[174,1300,329],{},[219,1302,1303,1309,1315,1327],{},[222,1304,1305,1308],{},[178,1306,1307],{},"Own a number."," Forecasts and close packs inherit whatever changed.",[222,1310,1311,1314],{},[178,1312,1313],{},"Own a customer relationship."," Model output that becomes a commitment is still the company’s commitment.",[222,1316,1317,1320,1321,1326],{},[178,1318,1319],{},"Own risk or legal."," Privacy law does not pause for a chatbot. ",[183,1322,1325],{"href":1323,"rel":1324},"https://eur-lex.europa.eu/eli/reg/2016/679/oj",[187],"GDPR"," still applies to purpose, minimisation, and erasure.",[222,1328,1329,1332],{},[178,1330,1331],{},"Are asked “who is in charge of AI here?”"," An inventory and a named owner beat a principles slide.",[174,1334,1335],{},"Good governance in practice is four working rules:",[219,1337,1338,1344,1350,1356],{},[222,1339,1340,1343],{},[178,1341,1342],{},"People and rights."," Humans and AI tools are both actors. Roles decide what they may start, see, and sign.",[222,1345,1346,1349],{},[178,1347,1348],{},"Data at question time."," Purpose and minimisation still apply when an AI is the one looking.",[222,1351,1352,1355],{},[178,1353,1354],{},"Action rights."," Read-only is a control. Unrestricted tools are an incident waiting for a bad prompt.",[222,1357,1358,1361,1362,311],{},[178,1359,1360],{},"Evidence and spend."," Chat scrollback is not a management system. Uncapped spend is a budget failure and often a security failure. See ",[183,1363,861],{"href":860},[174,1365,1366,1367,1369],{},"Blocking consumer ChatGPT at the office network, while people use personal phones, is not governance. It is a ",[183,1368,243],{"href":242}," problem with extra steps.",[380,1371,383],{"id":382},[174,1373,1374,1376],{},[178,1375,388],{}," Governance is whether an AI-proposed journal can post, against which checklist, with which signer, and whether the spend of the run was capped. “Unlimited AI” is not a control. Surprise inference bills are a governance failure that looks like a cloud invoice.",[174,1378,1379,1381,1382,1387],{},[178,1380,398],{}," Lawful basis, purpose limitation, customer-facing language, and reconstructable authorisation. Legal also has to separate the OECD-style public commitment from the runtime. A principles page does not implement Article-style oversight. For higher-risk systems, ",[183,1383,1386],{"href":1384,"rel":1385},"https://eur-lex.europa.eu/eli/reg/2024/1689/oj",[187],"EU AI law"," Article 14 talks about effective oversight: people must be able to interpret outputs and interrupt the system. A footer that says “generated by AI” is not that.",[174,1389,1390,1392],{},[178,1391,404],{}," Isolation of jobs, connector scope, and a place to put a paused run. Ops already runs change control. Governance is change control that includes a model as a proposer.",[174,1394,1395,1397],{},[178,1396,410],{}," The difference between a draft email and a sent commitment; between a suggested next step and a changed Amount. GTM feels friction first. The honest metric is time-to-approved-write, not time-to-first-answer.",[174,1399,1400,1402,1403,1406],{},[178,1401,416],{}," Identity of the connected user, read versus write, prompt injection as a path to a tool call, and the new store created by logs and indexes. The ",[183,1404,906],{"href":904,"rel":1405},[187]," treats retrieval and tool use as a security surface, not only a quality issue. Network DLP helps with paste-out. It does not quote a CRM change.",[380,1408,424],{"id":423},[174,1410,1411,1414],{},[178,1412,1413],{},"Governance as a committee."," Useful for risk registers. Useless if the product can still write.",[174,1416,1417,1420],{},[178,1418,1419],{},"Governance as model safety."," Refusals on public-web questions do not bind Salesforce.",[174,1422,1423,1426],{},[178,1424,1425],{},"Governance as a secure web gateway."," Necessary for some paste-out paths. Insufficient for writes, approvals, and causal history.",[174,1428,1429,1432,1433,311],{},[178,1430,1431],{},"Governance as blocking."," Blocks without a sanctioned path train people onto phones. See ",[183,1434,1253],{"href":242},[174,1436,1437,1440],{},[178,1438,1439],{},"Theatre."," A checkbox, a prompt that says “ask first,” or an admin toggle the model can ignore.",[174,1442,1443,1444,1446],{},"Good looks like: connectors default to read-only; writes are quoted; a named signer cannot be waived by the model; evidence lives on a ",[183,1445,477],{"href":476},"; spend has a ceiling; scope follows the job. Failure looks like a PDF, a blocked URL, and a personal API key in a wiki.",[174,1448,1449,1450,1453,1454,1457,1458,1460,1461,1463],{},"Adjacent concepts: ",[183,1451,1452],{"href":582},"write-back governance"," is the write subset. ",[183,1455,1456],{"href":784},"Human-in-the-loop"," is the gate. ",[183,1459,31],{"href":275}," are the isolation unit. An ",[183,1462,751],{"href":750}," is the product shape that makes those rules the default path.",[214,1465,482],{"id":481},[174,1467,1468],{},"Nimbus treats governance as how work is released, not as a sidecar policy engine.",[174,1470,1471,1472,1475],{},"Connectors — secure links to live systems — default to ",[178,1473,1474],{},"read-only",". When a change is proposed, the product shows the intended action and waits. A named person must sign. The model cannot waive the gate. Missing approval is fail-closed: nothing happens.",[174,1477,1478,1479,1481,1482,1484,1485,1488],{},"Scope is the ",[183,1480,276],{"href":275},": one job, with the playbooks, systems, teams, and budget that belong to that job. Evidence is the ",[183,1483,23],{"href":476},". The ",[183,1486,1487],{"href":472},"company wiki"," is the asserted policy the run must cite. Model routing does not bypass the gate.",[174,1490,494,1491,1493,1494,311],{},[183,1492,39],{"href":40},". For scoring vendors: ",[183,1495,1497],{"href":1496},"how-to-evaluate-ai-governance-platforms","How to evaluate AI governance platforms",[214,1499,505],{"id":504},[380,1501,1503],{"id":1502},"is-ai-governance-the-same-as-making-the-model-safe","Is AI governance the same as making the model “safe”?",[174,1505,1506,1507,1285,1509,1511],{},"No. Model safety is about what the model will say in the abstract. Enterprise governance is about what ",[209,1508,1284],{},[209,1510,1284],{}," systems and data.",[380,1513,1515],{"id":1514},"can-we-rely-on-the-secure-web-gateway","Can we rely on the secure web gateway?",[174,1517,1518],{},"Network controls help with paste-out. They do not quote a CRM change, bind an approver, or store a causal history. Use both.",[380,1520,1522],{"id":1521},"must-a-person-always-approve","Must a person always approve?",[174,1524,1525,1526,1529],{},"For many operational writes, yes. For read-only analysis, maybe not. The mistake is calling a system “human-approved” because a human ",[209,1527,1528],{},"could"," look, while changes proceed on model initiative.",[380,1531,1533],{"id":1532},"do-the-oecd-ai-principles-require-a-specific-product","Do the OECD AI Principles require a specific product?",[174,1535,1536],{},"No. They are a public commitment. A product can make evidence cheaper to produce. The commitment does not implement a gate.",[380,1538,1540],{"id":1539},"is-a-dpia-enough-to-go-live","Is a DPIA enough to go live?",[174,1542,1543],{},"It is necessary thinking, not a runtime. You still need identity, scope, fail-closed writes, and a record. The DPIA should describe those controls, not replace them.",[380,1545,1547],{"id":1546},"does-blocking-chatgpt-count-as-governance","Does blocking ChatGPT count as governance?",[174,1549,1550,1551,1554],{},"It is a network control. Without a sanctioned path that can see the right files, people use personal phones. Blocking can tighten ",[209,1552,1553],{},"after"," substitution exists.",[380,1556,1558],{"id":1557},"how-is-this-different-from-it-change-management","How is this different from IT change management?",[174,1560,1561],{},"It is the same instinct — who may change production, with what evidence — applied to a proposer that speaks English. Existing CAB processes rarely see model-initiated payloads unless the product emits them.",[380,1563,1565],{"id":1564},"who-should-be-the-named-owner-of-ai","Who should be the named owner of AI?",[174,1567,1568],{},"Someone who can inventory systems and stop a write path, not a volunteer “champion” with no authority over CRM. Federal-style guidance starts with inventory and ownership for a reason.",[380,1570,1572],{"id":1571},"can-we-govern-only-customer-facing-chatbots-and-ignore-internal-copilots","Can we govern only customer-facing chatbots and ignore internal copilots?",[174,1574,1575],{},"Internal tools still process personal data and still write to live systems. Air Canada was customer-facing. Ungoverned CRM hygiene is an internal path to the same class of invented fact.",[380,1577,1579],{"id":1578},"do-we-need-the-eu-ai-act-if-we-are-not-a-high-risk-provider","Do we need the EU AI Act if we are not a high-risk provider?",[174,1581,1582],{},"You may still have GDPR duties, sector rules, and customer contracts. Oversight and records are useful even when a specific Act title does not apply. Do not claim “Act compliant” because you have a button.",[380,1584,1586],{"id":1585},"where-does-spend-fit","Where does spend fit?",[174,1588,1589,1590,311],{},"Uncapped inference is a control failure. Quotes, ceilings, and attribution by job are governance of a scarce, abusable resource. See ",[183,1591,861],{"href":860},[380,1593,1595],{"id":1594},"is-an-acceptable-use-policy-still-worth-writing","Is an acceptable-use policy still worth writing?",[174,1597,1598],{},"Yes, as communication. No, as enforcement. Write the PDF. Then put the same rules in the product people actually use.",[214,1600,603],{"id":602},[174,1602,1603,967,1605,971,1607,311],{},[183,1604,583],{"href":582},[183,1606,1253],{"href":242},[183,1608,1075],{"href":750},[214,1610,614],{"id":613},[219,1612,1613,1619,1624,1630,1635,1640,1645,1651],{},[222,1614,1615],{},[183,1616,1618],{"href":1176,"rel":1617},[187],"Gartner, AI governance and TRiSM",[222,1620,1621],{},[183,1622,1162],{"href":1160,"rel":1623},[187],[222,1625,1626],{},[183,1627,1629],{"href":1241,"rel":1628},[187],"ICO, AI and data protection",[222,1631,1632],{},[183,1633,1184],{"href":1182,"rel":1634},[187],[222,1636,1637],{},[183,1638,1092],{"href":879,"rel":1639},[187],[222,1641,1642],{},[183,1643,1325],{"href":1323,"rel":1644},[187],[222,1646,1647],{},[183,1648,1650],{"href":1384,"rel":1649},[187],"EU AI Act (Regulation 2024/1689)",[222,1652,1653],{},[183,1654,906],{"href":904,"rel":1655},[187],{"title":155,"searchDepth":156,"depth":156,"links":1657},[1658,1659,1663,1664,1678,1679],{"id":216,"depth":156,"text":217},{"id":325,"depth":156,"text":326,"children":1660},[1661,1662],{"id":382,"depth":641,"text":383},{"id":423,"depth":641,"text":424},{"id":481,"depth":156,"text":482},{"id":504,"depth":156,"text":505,"children":1665},[1666,1667,1668,1669,1670,1671,1672,1673,1674,1675,1676,1677],{"id":1502,"depth":641,"text":1503},{"id":1514,"depth":641,"text":1515},{"id":1521,"depth":641,"text":1522},{"id":1532,"depth":641,"text":1533},{"id":1539,"depth":641,"text":1540},{"id":1546,"depth":641,"text":1547},{"id":1557,"depth":641,"text":1558},{"id":1564,"depth":641,"text":1565},{"id":1571,"depth":641,"text":1572},{"id":1578,"depth":641,"text":1579},{"id":1585,"depth":641,"text":1586},{"id":1594,"depth":641,"text":1595},{"id":602,"depth":156,"text":603},{"id":613,"depth":156,"text":614},"AI governance is the working rules for who may use which AI, on which data, and whether it may change a live business system — plus a record of what happened.","/blog/what-is-ai-governance",{"title":1133,"description":1680},"blog/what-is-ai-governance",[664,1685,1686,1687],"governance","compliance","audit","8Gc6YA0kBilCFVsDbCgM_sj0gjR1GSQazeF0iy3Fot8",{"enabled":149,"message":1690,"linkLabel":78,"linkHref":79,"id":1691,"title":1692,"archived":149,"authors":150,"badge":150,"body":1693,"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":1697,"relatedHeading":150,"seo":1698,"series":150,"sitemap":115,"status":150,"stem":1699,"subhead":150,"tags":150,"video":150,"whyJoin":150,"workplaceType":150,"__hash__":1700},"We're hiring! Join the team building the Sentient Enterprise.","content/shared/hiring.md","Hiring banner",{"type":152,"value":1694,"toc":1695},[],{"title":155,"searchDepth":156,"depth":156,"links":1696},[],"/shared/hiring",{"title":1692,"description":155},"shared/hiring","-6bioYD7lKYokGUVU3ff4hHTvB-sDyOMuCptKHnojfk",{"fold":1702,"id":1706,"title":1707,"archived":149,"authors":150,"badge":150,"body":1708,"date":150,"department":150,"description":155,"extension":158,"eyebrow":150,"faqHeader":150,"faqs":150,"footerBand":1712,"headline":150,"image":150,"industry":150,"jobType":150,"listed":149,"location":150,"navigation":115,"openRoles":150,"pageLayout":150,"path":1716,"relatedHeading":150,"seo":1717,"series":150,"sitemap":115,"status":150,"stem":1718,"subhead":150,"tags":150,"video":150,"whyJoin":150,"workplaceType":150,"__hash__":1719},{"headline":1703,"description":1704,"primaryLabel":8,"primaryTo":1705,"secondaryLabel":687,"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":1709,"toc":1710},[],{"title":155,"searchDepth":156,"depth":156,"links":1711},[],{"headline":1713,"description":1714,"primaryLabel":8,"primaryTo":1705,"secondaryLabel":1715,"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":1707,"description":155},"shared/cta","YHK6Fb8AvCPR1zZq7R_xiXUG0hwhP5UxHA8Ix52JQp4",1787194072680]