[{"data":1,"prerenderedAt":1649},["ShallowReactive",2],{"site-nav-content":3,"hiring-banner-content":177,"blog:\u002Fblog\u002Fwhat-is-an-agentic-workflow":189,"blog-index-copy":687,"blog:\u002Fblog\u002Fwhat-is-an-agentic-workflow:surround":708,"site-cta-content":1630},{"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":663,"definedTerm":164,"department":164,"description":664,"extension":172,"eyebrow":164,"faqHeader":665,"faqs":668,"footerBand":164,"headline":164,"image":164,"industry":164,"jobType":164,"listed":163,"location":164,"navigation":131,"openRoles":164,"pageLayout":164,"path":678,"relatedHeading":164,"seo":679,"series":680,"sitemap":131,"status":164,"stem":681,"subhead":164,"tags":682,"video":164,"whyJoin":164,"workplaceType":164,"__hash__":686},"content\u002Fblog\u002Fwhat-is-an-agentic-workflow.md","What is an Agentic Workflow",[193],{"name":194,"to":195},"Nimbus Research","https:\u002F\u002Fgonimbus.ai",{"label":197},"Explainer",{"type":166,"value":199,"toc":638},[200,204,216,224,227,232,235,262,265,315,327,331,334,341,344,380,383,388,394,405,411,417,432,436,442,448,454,460,470,473,477,484,491,506,510,514,517,521,524,528,531,535,538,542,547,551,554,558,561,565,573,577,580,584,590,594,599,603,606,610,618,622],[201,202,203],"p",{},"“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.",[201,205,206,207,211,212,215],{},"A ",[208,209,210],"strong",{},"workflow"," has steps and a stop. An ",[208,213,214],{},"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.",[201,217,218,223],{},[219,220,222],"a",{"href":221},"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.",[201,225,226],{},"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.",[228,229,231],"h2",{"id":230},"words-youll-hear","Words you’ll hear",[201,233,234],{},"Vendors collapse three different layers into the word “agentic”:",[236,237,238,245,251],"ul",{},[239,240,241,244],"li",{},[208,242,243],{},"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.",[239,246,247,250],{},[208,248,249],{},"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.”",[239,252,253,256,257,261],{},[208,254,255],{},"Agent platform."," Identity, connectors, tests, and governance around many workflows. At work, this is closer to an ",[219,258,260],{"href":259},"\u002Fblog\u002Fwhat-is-an-enterprise-ai-operating-system\u002F","enterprise AI operating system",".",[201,263,264],{},"Other terms:",[236,266,267,273,279,289,299,305],{},[239,268,269,272],{},[208,270,271],{},"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.",[239,274,275,278],{},[208,276,277],{},"Stop condition."," Budget hit, waiting on approval, error, empty result, human cancel. “The model says it is done” is a weak stop by itself.",[239,280,281,284,285,261],{},[208,282,283],{},"Write-back."," The AI is allowed to change a live system, not just draft. See ",[219,286,288],{"href":287},"\u002Fblog\u002Fwhat-is-write-back-governance\u002F","What is write-back governance",[239,290,291,294,295,261],{},[208,292,293],{},"Human wait."," A step in the sequence, not an interruption. See ",[219,296,298],{"href":297},"\u002Fblog\u002Fwhat-is-human-in-the-loop-ai\u002F","What is human-in-the-loop AI",[239,300,301,304],{},[208,302,303],{},"Version."," Which workflow definition ran. When policy changes, retrieval changes. Operators need to know which version ran last Tuesday.",[239,306,307,310,311,261],{},[208,308,309],{},"MCP."," A common plug so AI apps can use the same tools. Plumbing. It does not define your stops. See ",[219,312,314],{"href":313},"\u002Fblog\u002Fwhat-is-model-context-protocol\u002F","What is Model Context Protocol",[201,316,206,317,321,322,326],{},[219,318,320],{"href":319},"\u002Fblog\u002Fwhat-is-an-ai-workstream\u002F","workstream"," is the company object that ",[323,324,325],"em",{},"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.",[228,328,330],{"id":329},"why-you-should-care","Why you should care",[201,332,333],{},"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.",[201,335,336,337,340],{},"It affects you if the job is ",[208,338,339],{},"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.",[201,342,343],{},"Practical rules:",[236,345,346,352,358,364,370],{},[239,347,348,351],{},[208,349,350],{},"Read-heavy workflows"," can be long. They should still finish in an artefact with sources.",[239,353,354,357],{},[208,355,356],{},"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.”",[239,359,360,363],{},[208,361,362],{},"Human wait is a step",", not an interruption.",[239,365,366,369],{},[208,367,368],{},"Version the workflow."," Policy and retrieval drift. Last Tuesday’s run needs a definition you can still open.",[239,371,372,375,376,261],{},[208,373,374],{},"Budget is a stop."," See ",[219,377,379],{"href":378},"\u002Fblog\u002Fwhat-is-ai-token-economics\u002F","What is AI token economics",[201,381,382],{},"A mega-agent with “figure it out” as the spec is not a workflow. It is a hope.",[384,385,387],"h3",{"id":386},"what-changes-by-role","What changes by role",[201,389,390,393],{},[208,391,392],{},"Finance."," 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.",[201,395,396,399,400,404],{},[208,397,398],{},"Legal."," 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 — ",[219,401,403],{"href":402},"https:\u002F\u002Fwww.cbc.ca\u002Fnews\u002Fcanada\u002Fbritish-columbia\u002Fair-canada-chatbot-lawsuit-1.7116416","CBC’s report"," is the cautionary case for “the workflow ended at the message.” Legal also cares that the workflow version is reconstructable.",[201,406,407,410],{},[208,408,409],{},"Operations."," 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.”",[201,412,413,416],{},[208,414,415],{},"Go-to-market."," 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.",[201,418,419,422,423,426,427,431],{},[208,420,421],{},"Security."," Tool belts are attack surface. Prompt injection that tricks a model into ",[323,424,425],{},"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 ",[219,428,430],{"href":429},"https:\u002F\u002Fgenai.owasp.org\u002Fllm-top-10\u002F","OWASP Top 10 for LLM applications"," treats tool use as a security topic for this reason.",[384,433,435],{"id":434},"what-people-get-wrong","What people get wrong",[201,437,438,441],{},[208,439,440],{},"Chat as workflow."," A conversation that looks busy has no durable instance, no version, and no gate.",[201,443,444,447],{},[208,445,446],{},"A checklist in a prompt."," A start. Without tools, a durable job, and a stop, it is still a prompt.",[201,449,450,453],{},[208,451,452],{},"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.",[201,455,456,459],{},[208,457,458],{},"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.",[201,461,462,465,466,261],{},[208,463,464],{},"Multi-agent as a requirement."," A single tool-using agent can execute a workflow. Multiple agents help when duties already split. See ",[219,467,469],{"href":468},"\u002Fblog\u002Fwhat-is-multi-agent-ai\u002F","What is multi-agent AI",[201,471,472],{},"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.”",[228,474,476],{"id":475},"how-this-shows-up-in-nimbus","How this shows up in Nimbus",[201,478,479,480,261],{},"Nimbus’s delivery unit for operators is the ",[208,481,482],{},[219,483,320],{"href":319},[201,485,486,487,490],{},"The mapping in everyday terms: the brief is the goal; ",[219,488,489],{"href":468},"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.",[201,492,493,494,497,498,502,503,261],{},"See ",[219,495,31],{"href":496},"\u002Fproduct\u002Fworkstreams\u002F",", ",[219,499,501],{"href":500},"\u002Fproduct\u002Fagent-teams\u002F","Agent teams",", and ",[219,504,39],{"href":505},"\u002Fproduct\u002Fgovernance\u002F",[228,507,509],{"id":508},"questions-people-actually-ask","Questions people actually ask",[384,511,513],{"id":512},"is-a-checklist-in-a-prompt-an-agentic-workflow","Is a checklist in a prompt an agentic workflow?",[201,515,516],{},"It is a start. If there are no tools, no durable instance, and no gate, it is a prompt.",[384,518,520],{"id":519},"how-is-this-different-from-older-robotic-automation","How is this different from older robotic automation?",[201,522,523],{},"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.",[384,525,527],{"id":526},"do-agentic-workflows-require-multiple-agents","Do agentic workflows require multiple agents?",[201,529,530],{},"No. A single tool-using agent can execute a workflow. Multiple agents help when duties already split in the organisation.",[384,532,534],{"id":533},"can-a-workflow-be-fully-autonomous-in-production","Can a workflow be fully autonomous in production?",[201,536,537],{},"Only for actions you would already automate without a model, plus logging.",[384,539,541],{"id":540},"where-do-tool-connection-standards-fit","Where do tool-connection standards fit?",[201,543,544,545,261],{},"A common plug so AI apps can use the same tools is plumbing. It does not define your stops or approvals. See ",[219,546,314],{"href":313},[384,548,550],{"id":549},"what-is-a-good-stop-condition-besides-the-model-is-done","What is a good stop condition besides “the model is done”?",[201,552,553],{},"Budget ceiling, empty retrieval, tool error, human cancel, and wait-for-named-signer. “Done” from the model is a suggestion. Encode the others.",[384,555,557],{"id":556},"how-long-should-a-write-heavy-workflow-be","How long should a write-heavy workflow be?",[201,559,560],{},"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.",[384,562,564],{"id":563},"how-do-we-version-a-workflow-when-the-wiki-changes","How do we version a workflow when the wiki changes?",[201,566,567,568,572],{},"Treat the playbook version as an input to the run. The ",[219,569,571],{"href":570},"\u002Fblog\u002Fwhat-is-a-lifecycle-graph\u002F","lifecycle graph"," should cite which wiki version the steps respected. Changing policy without recording which definition ran is how Tuesday becomes unexplained.",[384,574,576],{"id":575},"is-agentic-the-same-as-autonomous","Is “agentic” the same as “autonomous”?",[201,578,579],{},"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.",[384,581,583],{"id":582},"can-we-import-every-available-tool-and-let-the-model-choose","Can we import every available tool and let the model choose?",[201,585,586,587,261],{},"That is a confused workflow. Least privilege applies to tools as much as to data. See ",[219,588,589],{"href":319},"What is an AI workstream",[384,591,593],{"id":592},"how-does-this-relate-to-human-in-the-loop","How does this relate to human-in-the-loop?",[201,595,596,597,261],{},"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 ",[219,598,298],{"href":297},[384,600,602],{"id":601},"will-a-better-model-remove-the-need-for-a-workflow","Will a better model remove the need for a workflow?",[201,604,605],{},"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.",[228,607,609],{"id":608},"related-reading","Related reading",[201,611,612,615,616,261],{},[219,613,614],{"href":259},"What is an enterprise AI operating system"," and ",[219,617,469],{"href":468},[228,619,621],{"id":620},"sources","Sources",[236,623,624,629,634],{},[239,625,626],{},[219,627,628],{"href":221},"Anthropic, Building effective agents",[239,630,631],{},[219,632,633],{"href":402},"CBC, Air Canada chatbot lawsuit",[239,635,636],{},[219,637,430],{"href":429},{"title":169,"searchDepth":170,"depth":170,"links":639},[640,641,646,647,661,662],{"id":230,"depth":170,"text":231},{"id":329,"depth":170,"text":330,"children":642},[643,645],{"id":386,"depth":644,"text":387},3,{"id":434,"depth":644,"text":435},{"id":475,"depth":170,"text":476},{"id":508,"depth":170,"text":509,"children":648},[649,650,651,652,653,654,655,656,657,658,659,660],{"id":512,"depth":644,"text":513},{"id":519,"depth":644,"text":520},{"id":526,"depth":644,"text":527},{"id":533,"depth":644,"text":534},{"id":540,"depth":644,"text":541},{"id":549,"depth":644,"text":550},{"id":556,"depth":644,"text":557},{"id":563,"depth":644,"text":564},{"id":575,"depth":644,"text":576},{"id":582,"depth":644,"text":583},{"id":592,"depth":644,"text":593},{"id":601,"depth":644,"text":602},{"id":608,"depth":170,"text":609},{"id":620,"depth":170,"text":621},"2026-08-17","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":666,"title":667},"Short answers","Steps, with a stop",[669,672,675],{"question":670,"answer":671},"What is an agentic workflow?","A sequence of steps an AI can run toward a goal, with rules for when to stop.",{"question":673,"answer":674},"When must a person be in it?","Before a live system changes. The workflow can prepare the change. It cannot finish it alone.",{"question":676,"answer":677},"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":191,"description":664},"explainer","blog\u002Fwhat-is-an-agentic-workflow",[680,683,684,685],"agentic-workflow","agents","workstreams","jqFWGede9d2XN3pJ1cSmdtlf_WWT8s5XhX9vIrpWXKY",{"hero":688,"id":690,"title":691,"archived":163,"authors":164,"badge":164,"body":692,"date":164,"definedTerm":164,"department":164,"description":696,"extension":172,"eyebrow":697,"faqHeader":164,"faqs":164,"footerBand":698,"headline":164,"image":164,"industry":164,"jobType":164,"listed":131,"location":164,"navigation":131,"openRoles":164,"pageLayout":164,"path":60,"relatedHeading":704,"seo":705,"series":164,"sitemap":131,"status":164,"stem":706,"subhead":164,"tags":164,"video":164,"whyJoin":164,"workplaceType":164,"__hash__":707},{"filename":689},"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":693,"toc":694},[],{"title":169,"searchDepth":170,"depth":170,"links":695},[],"Deep dives into pre-cognitive intelligence, sentient enterprises, and the evolving landscape of AI-driven business transformation.","Latest Research",{"headline":699,"description":700,"primaryLabel":701,"primaryTo":702,"secondaryLabel":703,"secondaryTo":12},"Stay at the frontier.","Subscribe for product updates and new insights.","Subscribe","\u002Fnewsletter","Explore the platform","More research",{"title":691,"description":696},"blog\u002Findex","BFSWGYO9bcTlaulivKYWyg08_DJHsdGg3OC6g_CG1Hw",[709,1167],{"id":710,"title":711,"archived":163,"authors":712,"badge":714,"body":715,"date":663,"definedTerm":164,"department":164,"description":1146,"extension":172,"eyebrow":164,"faqHeader":1147,"faqs":1149,"footerBand":164,"headline":164,"image":164,"industry":164,"jobType":164,"listed":163,"location":164,"navigation":131,"openRoles":164,"pageLayout":164,"path":1159,"relatedHeading":164,"seo":1160,"series":680,"sitemap":131,"status":164,"stem":1161,"subhead":164,"tags":1162,"video":164,"whyJoin":164,"workplaceType":164,"__hash__":1166},"content\u002Fblog\u002Fwhat-is-causal-ai-for-operations.md","What is Causal AI for Operations",[713],{"name":194,"to":195},{"label":197},{"type":166,"value":716,"toc":1122},[717,723,736,743,746,749,751,807,810,825,827,830,833,857,864,878,889,891,900,905,910,915,920,922,928,934,940,946,952,955,958,973,975,984,987,999,1001,1005,1008,1012,1015,1019,1022,1026,1029,1033,1036,1040,1046,1050,1053,1057,1060,1064,1067,1071,1074,1078,1085,1089,1092,1094,1103,1105],[201,718,719,720],{},"“Causal AI” is a phrase people type into ChatGPT when they mean: ",[208,721,722],{},"can we tell why something happened, or are we guessing?",[201,724,725,726,730,731,735],{},"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 ",[219,727,729],{"href":728},"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, ",[219,732,734],{"href":733},"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.",[201,737,738,739,742],{},"In operations, the question is more everyday and more urgent: ",[208,740,741],{},"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.",[201,744,745],{},"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.",[201,747,748],{},"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.",[228,750,231],{"id":230},[236,752,753,759,765,771,780,791,801],{},[239,754,755,758],{},[208,756,757],{},"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.",[239,760,761,764],{},[208,762,763],{},"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.",[239,766,767,770],{},[208,768,769],{},"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.",[239,772,773,776,777,261],{},[208,774,775],{},"Lifecycle graph."," The company’s chain of AI work: what was asked, who signed, what changed. See ",[219,778,779],{"href":570},"What is a lifecycle graph",[239,781,782,785,786,790],{},[208,783,784],{},"Provenance."," Who, what, when, derived from what. ",[219,787,789],{"href":788},"https:\u002F\u002Fwww.w3.org\u002FTR\u002Fprov-overview\u002F","W3C PROV"," is the open vocabulary for that idea.",[239,792,793,796,797,261],{},[208,794,795],{},"Confounder."," In science, a hidden third factor. In operations, the hidden factor is often “a human pasted a consumer-model answer into CRM.” See ",[219,798,800],{"href":799},"\u002Fblog\u002Fwhat-is-shadow-ai\u002F","What is shadow AI",[239,802,803,806],{},[208,804,805],{},"Rationale."," The model’s English explanation. Often written after the fact. Not a recorded structure.",[201,808,809],{},"Keep two layers apart:",[811,812,813,819],"ol",{},[239,814,815,818],{},[208,816,817],{},"Causal science."," Did the discount cause the win? Needs a design, not a fluent paragraph.",[239,820,821,824],{},[208,822,823],{},"Causal operations."," Brief → sources → proposal → approval → write → system response. Needs a record.",[228,826,330],{"id":329},[201,828,829],{},"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.",[201,831,832],{},"Causal operations affects you if you:",[236,834,835,841,851],{},[239,836,837,840],{},[208,838,839],{},"Have to explain a change."," “Who caused this field to move, against which rule?”",[239,842,843,850],{},[208,844,845,846,849],{},"Need to know what ",[323,847,848],{},"would"," have happened without approval."," In a real gate, the answer is nothing.",[239,852,853,856],{},[208,854,855],{},"Are tempted to file a model’s “because” as truth."," Natural-language rationales are often written after the fact.",[201,858,859,860,863],{},"You do ",[208,861,862],{},"not"," need a data-science sprint to ask:",[236,865,866,869,872,875],{},[239,867,868],{},"Why was this record changed?",[239,870,871],{},"Which policy version caused this refusal?",[239,873,874],{},"Did a spend cap stop the run?",[239,876,877],{},"Did analysis change the CRM, or only produce a draft?",[201,879,880,881,884,885,888],{},"You ",[208,882,883],{},"should"," need a statistician if you want to know whether signed next-step updates ",[323,886,887],{},"caused"," wins. The work record can attach “this account was treated.” Estimation is extra.",[384,890,387],{"id":386},[201,892,893,895,896,899],{},[208,894,392],{}," Sampling a journal requires the chain, not a story. Spend caps that fire are causes of ",[323,897,898],{},"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.”",[201,901,902,904],{},[208,903,398],{}," 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.",[201,906,907,909],{},[208,908,409],{}," 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.",[201,911,912,914],{},[208,913,415],{}," 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.",[201,916,917,919],{},[208,918,421],{}," 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.",[384,921,435],{"id":434},[201,923,924,927],{},[208,925,926],{},"The model’s “because.”"," Fluency is not identification and not a recorded structure.",[201,929,930,933],{},[208,931,932],{},"Two rising lines."," Correlation. File it as a hypothesis.",[201,935,936,939],{},[208,937,938],{},"Using the same AI that proposed the treatment to declare success."," That is marking your own homework.",[201,941,942,945],{},[208,943,944],{},"Skipping the operations layer to buy a science platform."," Without reconstructable interventions, the science team inherits Slack folklore.",[201,947,948,951],{},[208,949,950],{},"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.",[201,953,954],{},"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.",[201,956,957],{},"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.",[201,959,960,961,963,964,968,969,972],{},"Adjacent: ",[219,962,571],{"href":570}," is the product shape of the operations layer. ",[219,965,967],{"href":966},"\u002Fblog\u002Fwhat-is-institutional-memory-in-enterprise-ai\u002F","Institutional memory"," is what remains after people leave. ",[219,970,971],{"href":287},"Write-back governance"," makes “nothing happened” a possible true answer.",[228,974,476],{"id":475},[201,976,977,978,980,981,261],{},"Nimbus implements causal operations as the ",[208,979,23],{}," plus ",[208,982,983],{},"Perception",[201,985,986],{},"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.",[201,988,493,989,615,992,995,996,261],{},[219,990,23],{"href":991},"\u002Fproduct\u002Flifecycle-graph\u002F",[219,993,983],{"href":994},"\u002Fproduct\u002Fperception\u002F",". Writes that cannot happen without a signer are ",[219,997,998],{"href":287},"write-back governance",[228,1000,509],{"id":508},[384,1002,1004],{"id":1003},"if-the-model-explains-why-is-that-causal-ai","If the model explains “why,” is that causal AI?",[201,1006,1007],{},"No. A fluent paragraph is not a recorded structure, and it is not a statistical identification.",[384,1009,1011],{"id":1010},"do-we-need-advanced-causal-statistics-to-buy-an-operating-layer","Do we need advanced causal statistics to buy an operating layer?",[201,1013,1014],{},"No. You need reconstructable interventions. If you later staff a science team, they will thank you for not storing decisions as Slack folklore.",[384,1016,1018],{"id":1017},"can-the-graph-estimate-lift","Can the graph estimate lift?",[201,1020,1021],{},"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.",[384,1023,1025],{"id":1024},"where-does-this-end-and-a-bi-tool-start","Where does this end and a BI tool start?",[201,1027,1028],{},"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.",[384,1030,1032],{"id":1031},"what-is-an-intervention-in-this-sense","What is an intervention in this sense?",[201,1034,1035],{},"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.",[384,1037,1039],{"id":1038},"why-does-a-fail-closed-gate-matter-for-causality","Why does a fail-closed gate matter for causality?",[201,1041,1042,1043,1045],{},"Because the counterfactual is clean: without the named signer, the live system does not change. Fail-open systems cannot say what ",[323,1044,848],{}," have happened; they can only hope someone noticed.",[384,1047,1049],{"id":1048},"is-w3c-prov-the-same-as-a-lifecycle-graph","Is W3C PROV the same as a lifecycle graph?",[201,1051,1052],{},"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.",[384,1054,1056],{"id":1055},"can-we-reconstruct-causes-from-crm-field-history-plus-slack","Can we reconstruct causes from CRM field history plus Slack?",[201,1058,1059],{},"Field history says the value changed. Slack may contain a rumour. Neither joins playbook version, quoted payload, and signer identity as a single chain.",[384,1061,1063],{"id":1062},"does-causal-ai-mean-the-model-uses-causal-graphs-internally","Does “causal AI” mean the model uses causal graphs internally?",[201,1065,1066],{},"Sometimes, in research marketing. In this article it means operations can answer why a change happened. Ask vendors which meaning they are selling.",[384,1068,1070],{"id":1069},"how-should-we-talk-to-the-board","How should we talk to the board?",[201,1072,1073],{},"Separate “we can reconstruct what we did” from “we can estimate market lift.” The first is a control. The second is a study.",[384,1075,1077],{"id":1076},"where-does-the-wiki-fit","Where does the wiki fit?",[201,1079,1080,1081,261],{},"Asserted policy is often the parent of a refusal or a draft. “Which playbook version caused this flag?” is a causal-operations question. See ",[219,1082,1084],{"href":1083},"\u002Fblog\u002Fwhat-is-a-company-wiki-for-ai-agents\u002F","What is a company wiki for AI agents",[384,1086,1088],{"id":1087},"how-is-this-different-from-audit-logging","How is this different from audit logging?",[201,1090,1091],{},"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.",[228,1093,609],{"id":608},[201,1095,1096,497,1098,502,1100,261],{},[219,1097,779],{"href":570},[219,1099,589],{"href":319},[219,1101,1102],{"href":966},"What is institutional memory in enterprise AI",[228,1104,621],{"id":620},[236,1106,1107,1112,1117],{},[239,1108,1109],{},[219,1110,1111],{"href":728},"Stanford Encyclopedia of Philosophy, Causal Models",[239,1113,1114],{},[219,1115,1116],{"href":733},"Pearl, Causal inference in statistics: An overview",[239,1118,1119],{},[219,1120,1121],{"href":788},"W3C PROV overview",{"title":169,"searchDepth":170,"depth":170,"links":1123},[1124,1125,1129,1130,1144,1145],{"id":230,"depth":170,"text":231},{"id":329,"depth":170,"text":330,"children":1126},[1127,1128],{"id":386,"depth":644,"text":387},{"id":434,"depth":644,"text":435},{"id":475,"depth":170,"text":476},{"id":508,"depth":170,"text":509,"children":1131},[1132,1133,1134,1135,1136,1137,1138,1139,1140,1141,1142,1143],{"id":1003,"depth":644,"text":1004},{"id":1010,"depth":644,"text":1011},{"id":1017,"depth":644,"text":1018},{"id":1024,"depth":644,"text":1025},{"id":1031,"depth":644,"text":1032},{"id":1038,"depth":644,"text":1039},{"id":1048,"depth":644,"text":1049},{"id":1055,"depth":644,"text":1056},{"id":1062,"depth":644,"text":1063},{"id":1069,"depth":644,"text":1070},{"id":1076,"depth":644,"text":1077},{"id":1087,"depth":644,"text":1088},{"id":608,"depth":170,"text":609},{"id":620,"depth":170,"text":621},"Causal AI for operations answers why a change happened, using the actual steps and the approval, not a guess about the chatbot.",{"eyebrow":666,"title":1148},"The actual steps, not a guessed cause",[1150,1153,1156],{"question":1151,"answer":1152},"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":1154,"answer":1155},"Why is correlation not enough?","A metric moved. That does not tell you which action, which person, or which system produced it.",{"question":1157,"answer":1158},"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":711,"description":1146},"blog\u002Fwhat-is-causal-ai-for-operations",[680,1163,1164,1165],"causal-ai","lifecycle-graph","operations","YEdYX44CGV1ueBh0x0l5KK3vsh9X51db48M0nGl9Ang",{"id":1168,"title":1169,"archived":163,"authors":1170,"badge":1172,"body":1173,"date":663,"definedTerm":164,"department":164,"description":1609,"extension":172,"eyebrow":164,"faqHeader":1610,"faqs":1612,"footerBand":164,"headline":164,"image":164,"industry":164,"jobType":164,"listed":163,"location":164,"navigation":131,"openRoles":164,"pageLayout":164,"path":1622,"relatedHeading":164,"seo":1623,"series":680,"sitemap":131,"status":164,"stem":1624,"subhead":164,"tags":1625,"video":164,"whyJoin":164,"workplaceType":164,"__hash__":1629},"content\u002Fblog\u002Fwhat-is-ai-token-economics.md","What is AI Token Economics",[1171],{"name":194,"to":195},{"label":197},{"type":166,"value":1174,"toc":1585},[1175,1190,1196,1199,1206,1208,1295,1306,1308,1311,1331,1334,1357,1360,1362,1371,1376,1381,1386,1394,1396,1402,1412,1418,1426,1435,1444,1446,1449,1455,1465,1467,1471,1474,1478,1481,1485,1488,1492,1497,1501,1504,1508,1511,1515,1520,1524,1527,1531,1534,1538,1543,1547,1552,1556,1559,1561,1567,1569],[201,1176,206,1177,1180,1181,615,1185,1189],{},[208,1178,1179],{},"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. ",[219,1182,1184],{"href":1183},"https:\u002F\u002Fopenai.com\u002Fapi\u002Fpricing\u002F","OpenAI",[219,1186,1188],{"href":1187},"https:\u002F\u002Fwww.anthropic.com\u002Fpricing","Anthropic"," publish those ladders. Finance still cannot run the business on “12 million tokens of vendor A’s flagship.”",[201,1191,1192,1195],{},[208,1193,1194],{},"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.”",[201,1197,1198],{},"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.",[201,1200,1201,1202,1205],{},"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 ",[323,1203,1204],{},"feels"," free, so people pick the flagship every time. The ladder did not disappear. It hid.",[228,1207,231],{"id":230},[236,1209,1210,1216,1222,1232,1242,1248,1254,1263,1269,1280,1286],{},[239,1211,1212,1215],{},[208,1213,1214],{},"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.”",[239,1217,1218,1221],{},[208,1219,1220],{},"Seat licence."," Predictable cost per person. Often marketed as “unlimited.” The underlying work is still metered.",[239,1223,1224,1227,1228,1231],{},[208,1225,1226],{},"Pass-through API bill."," Each team has keys. Simple. Invites key sprawl and ",[219,1229,1230],{"href":799},"shadow AI"," on personal keys. At work, the invoice lands in engineering while go-to-market did the looping.",[239,1233,1234,1237,1238,1241],{},[208,1235,1236],{},"Quote."," A number ",[323,1239,1240],{},"before"," they run. At work, this is what makes a brief a decision rather than a surprise.",[239,1243,1244,1247],{},[208,1245,1246],{},"Cap \u002F 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.",[239,1249,1250,1253],{},[208,1251,1252],{},"Pool."," Organisation-level allowance. At work, one department should not be able to burn the company pool on a vanity run.",[239,1255,1256,1259,1260,1262],{},[208,1257,1258],{},"Attribution."," Chargeback by job, not “the AI bill.” At work, finance can ask which ",[219,1261,320],{"href":319}," consumed the units.",[239,1264,1265,1268],{},[208,1266,1267],{},"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.",[239,1270,1271,1274,1275,1279],{},[208,1272,1273],{},"Model routing."," Cheaper model for simple steps, stronger only when needed. See ",[219,1276,1278],{"href":1277},"\u002Fblog\u002Fwhat-is-model-routing\u002F","What is model routing",". At work, classify-this-ticket should not pay flagship rates.",[239,1281,1282,1285],{},[208,1283,1284],{},"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.",[239,1287,1288,1290,1291,261],{},[208,1289,277],{}," Budget hit, empty result, human cancel. Agent loops can dominate the bill without improving the artefact. See ",[219,1292,1294],{"href":1293},"\u002Fblog\u002Fwhat-is-an-agentic-workflow\u002F","What is an agentic workflow",[201,1296,1297,1298,1301,1302,1305],{},"The point is ",[208,1299,1300],{},"value per unit",", not minimum units regardless of outcome. Caching, wiki citations, and memory should make the ",[323,1303,1304],{},"same"," outcome cheaper over time. If unit cost of an approved update never falls, you are re-deriving folklore every run.",[228,1307,330],{"id":329},[201,1309,1310],{},"It affects you if you:",[236,1312,1313,1319,1325],{},[239,1314,1315,1318],{},[208,1316,1317],{},"Own the budget."," Surprise invoices arrive after agents looped all weekend.",[239,1320,1321,1324],{},[208,1322,1323],{},"Run the work."," You should see a number before you commit, not a lecture after.",[239,1326,1327,1330],{},[208,1328,1329],{},"Are tempted to shame people for using AI."," Shame drives personal keys. Cap the official path so it is safe to use.",[201,1332,1333],{},"Practical rhythm:",[236,1335,1336,1345,1351],{},[239,1337,1338,1341,1342,1344],{},[208,1339,1340],{},"Name the run."," Unnamed chats cannot be attributed. That is what a ",[219,1343,320],{"href":319}," is for.",[239,1346,1347,1350],{},[208,1348,1349],{},"Separate exploration from production."," Sandboxes can have tighter caps and cheaper default routes.",[239,1352,1353,1356],{},[208,1354,1355],{},"Review unit cost of outcomes"," — approved updates per unit — not tokens in the abstract.",[201,1358,1359],{},"Anti-pattern: a single corporate API key in a wiki, no per-job cap, monthly surprise. That is an unmetered utility.",[384,1361,387],{"id":386},[201,1363,1364,1366,1367,1370],{},[208,1365,392],{}," 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 ",[323,1368,1369],{},"stops",": a cap that fired is a successful control, not a failed project.",[201,1372,1373,1375],{},[208,1374,398],{}," 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.",[201,1377,1378,1380],{},[208,1379,409],{}," 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.",[201,1382,1383,1385],{},[208,1384,415],{}," 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.",[201,1387,1388,1390,1391,1393],{},[208,1389,421],{}," API keys are credentials. Personal keys in browser plugins are ",[219,1392,1230],{"href":799},". 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.",[384,1395,435],{"id":434},[201,1397,1398,1401],{},[208,1399,1400],{},"“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.",[201,1403,1404,1407,1408,1411],{},[208,1405,1406],{},"Punishing usage."," Chargeback without a sanctioned path recreates personal keys. Celebrate lower units ",[323,1409,1410],{},"per artefact"," as playbooks and memory compound.",[201,1413,1414,1417],{},[208,1415,1416],{},"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.",[201,1419,1420,1423,1424,261],{},[208,1421,1422],{},"One model for everything."," That is a routing failure dressed as quality culture. See ",[219,1425,1278],{"href":1277},[201,1427,1428,1431,1432,1434],{},[208,1429,1430],{},"No stop on loops."," ",[219,1433,222],{"href":221}," treats workflows with stop conditions as the grown-up shape. Economics is one of those stops.",[201,1436,1437,1438,615,1441,1443],{},"Good looks like: named jobs, quotes before commit, hard ceilings, routing policy, attribution, and falling unit cost as the ",[219,1439,1440],{"href":1083},"wiki",[219,1442,571],{"href":570}," reduce re-derivation. Failure looks like a shared key, a flagship default, and a board slide titled “AI spend.”",[228,1445,476],{"id":475},[201,1447,1448],{},"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.",[201,1450,1451,1452,1454],{},"Everyday questions can sit inside the allowance; heavier analysis, tools, and writes consume pool credits. The ",[219,1453,23],{"href":570}," can record spend as part of the chain, so “the run stopped because the ceiling was hit” is a causal fact.",[201,1456,493,1457,615,1460,1462,1463,261],{},[219,1458,44],{"href":1459},"\u002Fmodels\u002F",[219,1461,1278],{"href":1277},". Product context: ",[219,1464,31],{"href":496},[228,1466,509],{"id":508},[384,1468,1470],{"id":1469},"why-cant-we-just-pay-seats-and-call-it-unlimited","Why can’t we just pay seats and call it unlimited?",[201,1472,1473],{},"Seats predict headcount. Production AI spend is inference, tools, and writes. “Unlimited” hides the ladder; it does not delete it.",[384,1475,1477],{"id":1476},"what-should-finance-actually-see","What should finance actually see?",[201,1479,1480],{},"A quote before commit, a cap during the run, and attribution by job afterwards — in one unit they can compare across vendors and steps.",[384,1482,1484],{"id":1483},"wont-cheaper-models-get-worse-answers","Won’t cheaper models get worse answers?",[201,1486,1487],{},"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.",[384,1489,1491],{"id":1490},"do-we-punish-teams-for-using-ai","Do we punish teams for using AI?",[201,1493,1494,1495,1411],{},"No. Punishing usage revives shadow AI. Celebrate lower units ",[323,1496,1410],{},[384,1498,1500],{"id":1499},"what-is-an-ntu-in-plain-language","What is an NTU in plain language?",[201,1502,1503],{},"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.",[384,1505,1507],{"id":1506},"should-every-chat-be-billed-to-a-cost-centre","Should every chat be billed to a cost centre?",[201,1509,1510],{},"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.",[384,1512,1514],{"id":1513},"how-do-agent-loops-blow-the-budget","How do agent loops blow the budget?",[201,1516,1517,1518,261],{},"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 ",[219,1519,1294],{"href":1293},[384,1521,1523],{"id":1522},"is-caching-the-same-as-token-economics","Is caching the same as token economics?",[201,1525,1526],{},"Caching is a tactic. Economics is the management system: quote, cap, attribute, route. Caching without attribution still leaves you unable to explain the bill.",[384,1528,1530],{"id":1529},"do-we-need-a-data-warehouse-to-do-this","Do we need a data warehouse to do this?",[201,1532,1533],{},"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.",[384,1535,1537],{"id":1536},"how-does-this-relate-to-write-back","How does this relate to write-back?",[201,1539,1540,1541,261],{},"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 ",[219,1542,288],{"href":287},[384,1544,1546],{"id":1545},"can-we-lock-one-vendor-to-simplify-pricing","Can we lock one vendor to simplify pricing?",[201,1548,1549,1550,261],{},"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 ",[219,1551,1278],{"href":1277},[384,1553,1555],{"id":1554},"why-not-just-set-a-monthly-company-cap","Why not just set a monthly company cap?",[201,1557,1558],{},"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.",[228,1560,609],{"id":608},[201,1562,1563,615,1565,261],{},[219,1564,1278],{"href":1277},[219,1566,589],{"href":319},[228,1568,621],{"id":620},[236,1570,1571,1576,1581],{},[239,1572,1573],{},[219,1574,1575],{"href":1183},"OpenAI API pricing",[239,1577,1578],{},[219,1579,1580],{"href":1187},"Anthropic pricing",[239,1582,1583],{},[219,1584,628],{"href":221},{"title":169,"searchDepth":170,"depth":170,"links":1586},[1587,1588,1592,1593,1607,1608],{"id":230,"depth":170,"text":231},{"id":329,"depth":170,"text":330,"children":1589},[1590,1591],{"id":386,"depth":644,"text":387},{"id":434,"depth":644,"text":435},{"id":475,"depth":170,"text":476},{"id":508,"depth":170,"text":509,"children":1594},[1595,1596,1597,1598,1599,1600,1601,1602,1603,1604,1605,1606],{"id":1469,"depth":644,"text":1470},{"id":1476,"depth":644,"text":1477},{"id":1483,"depth":644,"text":1484},{"id":1490,"depth":644,"text":1491},{"id":1499,"depth":644,"text":1500},{"id":1506,"depth":644,"text":1507},{"id":1513,"depth":644,"text":1514},{"id":1522,"depth":644,"text":1523},{"id":1529,"depth":644,"text":1530},{"id":1536,"depth":644,"text":1537},{"id":1545,"depth":644,"text":1546},{"id":1554,"depth":644,"text":1555},{"id":608,"depth":170,"text":609},{"id":620,"depth":170,"text":621},"AI token economics treats model usage as a budget: you pay per chunk of text read and written, so finance can quote, cap, and attribute it.",{"eyebrow":666,"title":1611},"A budget for what the model reads and writes",[1613,1616,1619],{"question":1614,"answer":1615},"What is AI token economics?","Treating AI usage as a real budget. You pay per chunk of text the model reads and writes.",{"question":1617,"answer":1618},"Why not buy unlimited AI?","Unlimited hides which job cost what. Finance cannot quote, cap, or attribute the spend.",{"question":1620,"answer":1621},"What should a team see before a run?","A quote and a ceiling. The run stops when the ceiling is hit.","\u002Fblog\u002Fwhat-is-ai-token-economics",{"title":1169,"description":1609},"blog\u002Fwhat-is-ai-token-economics",[680,1626,1627,1628],"token-economics","ntu","model-routing","RvB7TAIXInHDZunYt1V0yUevXBn0a5hFMuTvtxxuXAA",{"fold":1631,"id":1635,"title":1636,"archived":163,"authors":164,"badge":164,"body":1637,"date":164,"definedTerm":164,"department":164,"description":169,"extension":172,"eyebrow":164,"faqHeader":164,"faqs":164,"footerBand":1641,"headline":164,"image":164,"industry":164,"jobType":164,"listed":131,"location":164,"navigation":131,"openRoles":164,"pageLayout":164,"path":1645,"relatedHeading":164,"seo":1646,"series":164,"sitemap":163,"status":164,"stem":1647,"subhead":164,"tags":164,"video":164,"whyJoin":164,"workplaceType":164,"__hash__":1648},{"headline":1632,"description":1633,"primaryLabel":8,"primaryTo":1634,"secondaryLabel":703,"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":1638,"toc":1639},[],{"title":169,"searchDepth":170,"depth":170,"links":1640},[],{"headline":1642,"description":1643,"primaryLabel":8,"primaryTo":1634,"secondaryLabel":1644,"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":1636,"description":169},"shared\u002Fcta","eYqahyaPnbp8GKrWpoORbZdtmkWmgHr5F61ZHOnb8sY",1792999480958]