[{"data":1,"prerenderedAt":1632},["ShallowReactive",2],{"site-nav-content":3,"blog:\u002Fblog\u002Fwhat-is-an-enterprise-ai-operating-system":177,"blog-index-copy":720,"blog:\u002Fblog\u002Fwhat-is-an-enterprise-ai-operating-system:surround":741,"hiring-banner-content":1601,"site-cta-content":1613},{"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",{"id":178,"title":179,"archived":163,"authors":180,"badge":184,"body":186,"date":697,"definedTerm":164,"department":164,"description":191,"extension":172,"eyebrow":164,"faqHeader":698,"faqs":701,"footerBand":164,"headline":164,"image":164,"industry":164,"jobType":164,"listed":131,"location":164,"navigation":131,"openRoles":164,"pageLayout":164,"path":711,"relatedHeading":164,"seo":712,"series":713,"sitemap":131,"status":164,"stem":714,"subhead":164,"tags":715,"video":164,"whyJoin":164,"workplaceType":164,"__hash__":719},"content\u002Fblog\u002Fwhat-is-an-enterprise-ai-operating-system.md","What is an Enterprise AI Operating System",[181],{"name":182,"to":183},"Nimbus Research","https:\u002F\u002Fgonimbus.ai",{"label":185},"Explainer",{"type":166,"value":187,"toc":672},[188,192,195,198,206,209,214,290,293,326,339,346,350,356,359,362,365,368,373,379,393,399,405,411,415,421,435,441,447,453,456,468,472,475,517,539,543,547,550,554,562,566,569,573,576,580,583,587,590,594,597,601,608,612,615,619,624,628,631,635,646,650,656,660],[189,190,191],"p",{},"An enterprise AI operating system is the layer between the AI model and how departments actually work — like Windows sits between the chip and your apps.",[189,193,194],{},"If your question is “which model should we buy,” you are shopping for a chip. If your question is “how do revenue, legal, and finance run the same loop without a personal-account workaround,” you are shopping for an OS.",[189,196,197],{},"It is not a chatbot with company login. It is not a model API with a prompt library.",[189,199,200,205],{},[201,202,204],"a",{"href":203},"https:\u002F\u002Fwww.mckinsey.com\u002Fcapabilities\u002Fquantumblack\u002Four-insights\u002Fthe-state-of-ai","McKinsey’s 2025 State of AI survey"," found that 88% of companies use AI in at least one function — and that a majority are still piloting. About one in three report that they are scaling. Buying another model does not close that gap. The missing layer is how work actually runs.",[189,207,208],{},"The OS metaphor is useful if you keep it honest. An operating system does not replace your spreadsheet or your CRM. It gives applications isolation, permissions, input and output, and a place to keep state after the window closes. An enterprise AI OS does the same for work that uses models: isolation of jobs, rights over tools and data, reads and writes to live systems, budgets, and a record that survives the session.",[210,211,213],"h2",{"id":212},"words-youll-hear","Words you’ll hear",[215,216,217,225,231,242,248,259,270,280],"ul",{},[218,219,220,224],"li",{},[221,222,223],"strong",{},"Copilot."," A high-quality assistant for a person. Admin controls, company login. Not, by itself, how several departments finish one job under a named signer. At work, this is “help me draft.” It is not “release this CRM change.”",[218,226,227,230],{},[221,228,229],{},"Operating system (in this sense)."," Process isolation, permissions, input\u002Foutput to live systems, budgeting, and durable state — the jobs a kernel does for apps.",[218,232,233,236,237,241],{},[221,234,235],{},"System of record."," CRM, ERP, HR — still authoritative. The OS is the system of ",[238,239,240],"em",{},"work",", not a second CRM.",[218,243,244,247],{},[221,245,246],{},"Forward-deployed engineer."," A vendor consultant who sits with you for months. Some programmes need that. Many companies need governed work this quarter without it.",[218,249,250,253,254,258],{},[221,251,252],{},"NIST AI RMF."," ",[201,255,257],{"href":256},"https:\u002F\u002Fwww.nist.gov\u002Fitl\u002Fai-risk-management-framework","Govern, Map, Measure, Manage"," — public-sector language for the same kernel idea: identity, tool rights, and a record attached to real actions.",[218,260,261,264,265,269],{},[221,262,263],{},"Workstream."," The process-isolation unit: one job, one scope, one finish line. See ",[201,266,268],{"href":267},"\u002Fblog\u002Fwhat-is-an-ai-workstream\u002F","What is an AI workstream",".",[218,271,272,275,276,269],{},[221,273,274],{},"Fail-closed writes."," Missing named signer means nothing happens. See ",[201,277,279],{"href":278},"\u002Fblog\u002Fwhat-is-write-back-governance\u002F","What is write-back governance",[218,281,282,285,286,269],{},[221,283,284],{},"Usage limits and quotes."," Work is quoted before it runs and capped by limits and budgets. See ",[201,287,289],{"href":288},"\u002Fblog\u002Fwhat-is-ai-token-economics\u002F","What is AI token economics",[189,291,292],{},"Five jobs cluster around the term:",[294,295,296,302,308,314,320],"ol",{},[218,297,298,301],{},[221,299,300],{},"Process isolation."," Go-to-market does not silently inherit finance’s ERP login.",[218,303,304,307],{},[221,305,306],{},"Resource management."," Inference and tool calls are budgeted.",[218,309,310,313],{},[221,311,312],{},"I\u002FO control."," Reads and writes to CRM and ERP are first-class — not “chat that sometimes calls an API.”",[218,315,316,319],{},[221,317,318],{},"Permissioning."," Identity and context decide what an agent can see and do. A signed PDF is not enforcement.",[218,321,322,325],{},[221,323,324],{},"Durable state."," Outcomes, approvals, and rationale survive the session.",[189,327,328,329,333,334,338],{},"Copilots generally fail the last three. ",[201,330,332],{"href":331},"\u002Fblog\u002Fnimbus-vs-chatgpt-enterprise\u002F","ChatGPT Enterprise"," and ",[201,335,337],{"href":336},"\u002Fblog\u002Fnimbus-vs-claude\u002F","Claude for Work"," are excellent assistants. They are not this job.",[189,340,341,345],{},[201,342,344],{"href":343},"\u002Fblog\u002Fwhat-is-model-context-protocol\u002F","Model Context Protocol"," is also not this job. A common plug for tools is USB. USB did not create Windows.",[210,347,349],{"id":348},"why-you-should-care","Why you should care",[189,351,352,353,355],{},"Operators do not “open the OS” the way they open a model playground. They open ",[221,354,240],{},": a brief, a scoped live system, a review, a release.",[189,357,358],{},"It affects you if AI is starting to touch revenue, financial close, customer records, or regulated processes. Chat history does not answer “who approved this, against which policy?”",[189,360,361],{},"The OS also matters if you refuse a six-to-twelve-month vendor-engineer programme as the only path to production.",[189,363,364],{},"The anti-pattern is using an OS as a better chatbot: one user, one thread, no write path, no memory beyond the conversation. If nobody except the original operator can reconstruct what happened, you have a log, not an operating system.",[189,366,367],{},"Personal copilots optimise for “the model always answers.” An OS optimises for “the company only acts when the gate says so.” A spend cap or a missing approval is a successful outcome.",[369,370,372],"h3",{"id":371},"what-changes-by-role","What changes by role",[189,374,375,378],{},[221,376,377],{},"Finance."," The OS is how close and forecast jobs get a budget, a read-only ERP connector, a wiki checklist, and a named signer — without a second ledger. Finance should still own NetSuite. The OS should point at it.",[189,380,381,384,385,388,389,269],{},[221,382,383],{},"Legal."," Reconstructable authorisation, purpose-limited scope, and a place that is not a personal chat vendor. Legal should evaluate whether unapproved writes are ",[238,386,387],{},"impossible",", not whether a policy PDF exists. See ",[201,390,392],{"href":391},"\u002Fblog\u002Fwhat-is-ai-governance\u002F","What is AI governance",[189,394,395,398],{},[221,396,397],{},"Operations."," Isolation and durable state are ops problems. Ops should ask whether a paused run is a first-class object, whether connectors default to read-only, and whether Perception (or equivalent) can answer “why did this change?” without a data team reconstructing Slack.",[189,400,401,404],{},[221,402,403],{},"Go-to-market."," Cross-department loops — legal on a renewal, finance on a discount — need a shared job, not a shared inbox. GTM should not have to choose between a copilot that cannot write safely and a spreadsheet export to a consumer model.",[189,406,407,410],{},[221,408,409],{},"Security."," Identity, least privilege, fail-closed I\u002FO, and not turning the OS into a second store of the whole company. Security also cares that self-service configuration does not mean tenant-wide write keys.",[369,412,414],{"id":413},"what-people-get-wrong","What people get wrong",[189,416,417,420],{},[221,418,419],{},"“ChatGPT with integrations.”"," Plugins without scoped work, approval architecture, and durable decision records are plugins. A copilot with automation actions can move data. It cannot, by itself, make unapproved writes impossible.",[189,422,423,426,427,430,431,434],{},[221,424,425],{},"MLOps as a substitute."," MLOps governs ",[238,428,429],{},"model production",". An enterprise AI OS governs ",[238,432,433],{},"operational work that uses models",". They stack.",[189,436,437,440],{},[221,438,439],{},"Replacing the CRM."," Salesforce, NetSuite, Workday, and the warehouse remain authoritative. Duplicating them is a second system of record.",[189,442,443,446],{},[221,444,445],{},"OS as chatbot."," One user, one thread, no write path, no memory. That is a copilot with extra vocabulary.",[189,448,449,452],{},[221,450,451],{},"Forward-deployed as the only path."," Some warehouses need specialists. Most operators need to attach a connector and set a named signer in the UI.",[189,454,455],{},"Good looks like: workstreams, wiki, read-only-default connectors, agent teams, Lifecycle Graph, model routing, run quotes, fail-closed writes, self-service configuration. Failure looks like another model contract plus a six-month SOW.",[189,457,458,459,463,464,269],{},"For the copilot-versus-OS choice, see ",[201,460,462],{"href":461},"\u002Fblog\u002Fcollaborative-ai-and-personal-assistants\u002F","collaborative AI and personal assistants",". For vendor scoring, ",[201,465,467],{"href":466},"\u002Fblog\u002Fhow-to-evaluate-an-agent-harness\u002F","how to evaluate an agent harness",[210,469,471],{"id":470},"how-this-shows-up-in-nimbus","How this shows up in Nimbus",[189,473,474],{},"Nimbus is a self-service enterprise AI OS. Operators configure it in the product.",[215,476,477,484,490,496,502,511],{},[218,478,479,483],{},[221,480,481],{},[201,482,31],{"href":267}," isolate process.",[218,485,486,489],{},[221,487,488],{},"Wiki"," holds asserted policy — approved playbooks, not a dump of PDFs a search might find.",[218,491,492,495],{},[221,493,494],{},"Connectors"," attach live systems. Default is read-only. Write-back is opt-in and gated.",[218,497,498,501],{},[221,499,500],{},"Agent teams"," are department-shaped.",[218,503,504,506,507,510],{},[221,505,23],{}," stores the causal record. ",[221,508,509],{},"Perception"," queries it in ordinary language.",[218,512,513,516],{},[221,514,515],{},"Model routing"," puts routine extract on cheaper models.",[189,518,519,520,333,523,525,526,529,530,529,533,529,536,269],{},"See ",[201,521,11],{"href":522},"\u002Foverview\u002F",[201,524,467],{"href":466},". Product surfaces: ",[201,527,31],{"href":528},"\u002Fproduct\u002Fworkstreams\u002F",", ",[201,531,39],{"href":532},"\u002Fproduct\u002Fgovernance\u002F",[201,534,23],{"href":535},"\u002Fproduct\u002Flifecycle-graph\u002F",[201,537,509],{"href":538},"\u002Fproduct\u002Fperception\u002F",[210,540,542],{"id":541},"questions-people-actually-ask","Questions people actually ask",[369,544,546],{"id":545},"is-an-enterprise-ai-os-just-chatgpt-with-integrations","Is an enterprise AI OS just “ChatGPT with integrations”?",[189,548,549],{},"No. Integrations without scoped work, approval architecture, and durable decision records are plugins. A copilot with automation actions can move data. It cannot, by itself, make unapproved writes impossible.",[369,551,553],{"id":552},"how-is-this-different-from-mlops","How is this different from MLOps?",[189,555,556,557,430,559,561],{},"MLOps governs ",[238,558,429],{},[238,560,433],{},". They stack. They do not substitute.",[369,563,565],{"id":564},"do-we-still-need-a-crm-if-we-buy-an-os","Do we still need a CRM if we buy an OS?",[189,567,568],{},"Yes. Salesforce, NetSuite, Workday, and the warehouse remain authoritative.",[369,570,572],{"id":571},"does-every-company-need-an-os","Does every company need an OS?",[189,574,575],{},"If the job is personal drafting with no writes to live systems, a governed copilot may be enough. The OS becomes the right abstraction when work crosses departments, when writes are material, and when you must reconstruct decisions.",[369,577,579],{"id":578},"is-this-the-same-as-an-integration-platform-ipaas","Is this the same as an integration platform (iPaaS)?",[189,581,582],{},"No. iPaaS moves data on schedules and triggers. An AI OS runs language-using jobs with scope, spend, and a human gate. You may still need iPaaS. It does not quote a named signer on a CRM payload.",[369,584,586],{"id":585},"does-operating-system-mean-we-install-software-on-laptops","Does “operating system” mean we install software on laptops?",[189,588,589],{},"No. It is a layer for work, not a desktop kernel. The metaphor is isolation, permissions, I\u002FO, and state.",[369,591,593],{"id":592},"can-we-build-this-ourselves-on-a-model-api","Can we build this ourselves on a model API?",[189,595,596],{},"You can assemble pieces. You will still need isolation, connectors, gates, spend, and a graph. Most “we built a GPT” programmes stall at the copilot layer. McKinsey’s split between using AI and scaling it is that stall in survey form.",[369,598,600],{"id":599},"where-do-agent-teams-fit","Where do agent teams fit?",[189,602,603,604,269],{},"They are the department-shaped specialists the OS schedules onto workstreams. They are not the OS. See ",[201,605,607],{"href":606},"\u002Fblog\u002Fwhat-is-multi-agent-ai\u002F","What is multi-agent AI",[369,609,611],{"id":610},"how-does-nists-ai-rmf-map","How does NIST’s AI RMF map?",[189,613,614],{},"Govern (owners, policy), Map (inventory of jobs and systems), Measure (evidence, spend, rejects), Manage (fail-closed writes, incident path). A product can make those cheaper. A framework PDF cannot enforce them.",[369,616,618],{"id":617},"what-is-perception-in-this-picture","What is Perception in this picture?",[189,620,621,622,269],{},"Ordinary-language questions over the company’s graph, wiki, and scoped systems — with the next step being a workstream, not another search. See ",[201,623,509],{"href":538},[369,625,627],{"id":626},"do-we-need-a-forward-deployed-engineer-to-go-live","Do we need a forward-deployed engineer to go live?",[189,629,630],{},"Nothe default path. If operators cannot attach a read-only connector and set a named signer in the UI, you do not have a self-service OS. Specialists belong on genuine exceptions, such as a warehouse with no OAuth.",[369,632,634],{"id":633},"is-search-rag-an-os","Is search (RAG) an OS?",[189,636,637,638,333,642,269],{},"No. Lookup-then-answer is infrastructure. It does not isolate jobs or gate writes. See ",[201,639,641],{"href":640},"\u002Fblog\u002Fwhat-is-enterprise-rag\u002F","What is enterprise RAG",[201,643,645],{"href":644},"\u002Fblog\u002Fnimbus-vs-glean\u002F","Nimbus vs Glean",[210,647,649],{"id":648},"related-reading","Related reading",[189,651,652,333,654,269],{},[201,653,268],{"href":267},[201,655,392],{"href":391},[210,657,659],{"id":658},"sources","Sources",[215,661,662,667],{},[218,663,664],{},[201,665,666],{"href":203},"McKinsey, The state of AI (2025)",[218,668,669],{},[201,670,671],{"href":256},"NIST AI Risk Management Framework",{"title":169,"searchDepth":170,"depth":170,"links":673},[674,675,680,681,695,696],{"id":212,"depth":170,"text":213},{"id":348,"depth":170,"text":349,"children":676},[677,679],{"id":371,"depth":678,"text":372},3,{"id":413,"depth":678,"text":414},{"id":470,"depth":170,"text":471},{"id":541,"depth":170,"text":542,"children":682},[683,684,685,686,687,688,689,690,691,692,693,694],{"id":545,"depth":678,"text":546},{"id":552,"depth":678,"text":553},{"id":564,"depth":678,"text":565},{"id":571,"depth":678,"text":572},{"id":578,"depth":678,"text":579},{"id":585,"depth":678,"text":586},{"id":592,"depth":678,"text":593},{"id":599,"depth":678,"text":600},{"id":610,"depth":678,"text":611},{"id":617,"depth":678,"text":618},{"id":626,"depth":678,"text":627},{"id":633,"depth":678,"text":634},{"id":648,"depth":170,"text":649},{"id":658,"depth":170,"text":659},"2026-08-17",{"eyebrow":699,"title":700},"Short answers","The layer between the model and the work",[702,705,708],{"question":703,"answer":704},"What is an enterprise AI operating system?","The layer between the AI model and how departments actually work, the way an operating system sits between the chip and the applications.",{"question":706,"answer":707},"Is the model the operating system?","No. The model is a component. The operating system is how work, permissions, and records are organised around it.",{"question":709,"answer":710},"What should departments find there?","The same job, the same rules, and a record, instead of a different chatbot in every function.","\u002Fblog\u002Fwhat-is-an-enterprise-ai-operating-system",{"title":179,"description":191},"explainer","blog\u002Fwhat-is-an-enterprise-ai-operating-system",[713,716,717,718],"enterprise-ai","operating-system","governance","viSXGpdmqdd2fuaUmfH6OngdOgRbSEtUnbseaPgQvis",{"hero":721,"id":723,"title":724,"archived":163,"authors":164,"badge":164,"body":725,"date":164,"definedTerm":164,"department":164,"description":729,"extension":172,"eyebrow":730,"faqHeader":164,"faqs":164,"footerBand":731,"headline":164,"image":164,"industry":164,"jobType":164,"listed":131,"location":164,"navigation":131,"openRoles":164,"pageLayout":164,"path":60,"relatedHeading":737,"seo":738,"series":164,"sitemap":131,"status":164,"stem":739,"subhead":164,"tags":164,"video":164,"whyJoin":164,"workplaceType":164,"__hash__":740},{"filename":722},"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":726,"toc":727},[],{"title":169,"searchDepth":170,"depth":170,"links":728},[],"Deep dives into pre-cognitive intelligence, sentient enterprises, and the evolving landscape of AI-driven business transformation.","Latest Research",{"headline":732,"description":733,"primaryLabel":734,"primaryTo":735,"secondaryLabel":736,"secondaryTo":12},"Stay at the frontier.","Subscribe for product updates and new insights.","Subscribe","\u002Fnewsletter","Explore the platform","More research",{"title":724,"description":729},"blog\u002Findex","BFSWGYO9bcTlaulivKYWyg08_DJHsdGg3OC6g_CG1Hw",[742,1144],{"id":743,"title":744,"archived":163,"authors":745,"badge":747,"body":749,"date":697,"definedTerm":164,"department":164,"description":1122,"extension":172,"eyebrow":164,"faqHeader":1123,"faqs":1125,"footerBand":164,"headline":164,"image":164,"industry":164,"jobType":164,"listed":131,"location":164,"navigation":131,"openRoles":164,"pageLayout":164,"path":1135,"relatedHeading":164,"seo":1136,"series":1137,"sitemap":131,"status":164,"stem":1138,"subhead":164,"tags":1139,"video":164,"whyJoin":164,"workplaceType":164,"__hash__":1143},"content\u002Fblog\u002Fwhat-to-look-for-in-model-routing.md","What to Look for in Model Routing",[746],{"name":182,"to":183},{"label":748},"Evaluation",{"type":166,"value":750,"toc":1105},[751,756,759,775,777,822,833,835,838,858,863,883,889,892,896,935,945,948,952,955,982,985,987,990,997,1006,1008,1012,1018,1022,1025,1029,1032,1036,1043,1047,1054,1056,1066,1068],[189,752,753,755],{},[221,754,515],{}," is the policy that maps a task class to a model class before inference runs. It is not a brand preference. It is not a dropdown labelled “best.”",[189,757,758],{},"Someone using the flagship model to label a ticket is how you pay frontier prices for work a compact model could have finished in a second. That is not a moral failing. It is a missing policy. The product either chooses before the call, or a person chooses in a menu, or the default is the largest model “for quality.” Only the first is routing.",[189,760,761,333,765,769,770,774],{},[201,762,764],{"href":763},"https:\u002F\u002Fopenai.com\u002Fapi\u002Fpricing\u002F","OpenAI’s API pricing",[201,766,768],{"href":767},"https:\u002F\u002Fwww.anthropic.com\u002Fpricing","Anthropic’s pricing"," make the same point in public: compact and frontier models are not the same invoice line. ",[201,771,773],{"href":772},"https:\u002F\u002Fhai.stanford.edu\u002Fai-index\u002F2025-ai-index-report","Stanford HAI’s 2025 AI Index"," has tracked how fast inference cost and capability moved — which is exactly why “best model” is not a routing policy. Best for a memo is not best for a classify step. Best last quarter is not best this quarter.",[210,776,213],{"id":212},[215,778,779,785,791,797,806,816],{},[218,780,781,784],{},[221,782,783],{},"Frontier \u002F flagship model."," The strongest (and usually most expensive) model a lab currently sells. Reserved for synthesis, hard reasoning, and novel language. Not for labelling.",[218,786,787,790],{},[221,788,789],{},"Compact \u002F small model."," Faster and cheaper. Often enough for extract, classify, and summarise. “Small” is a cost and latency class, not an insult.",[218,792,793,796],{},[221,794,795],{},"Task class."," The kind of step: classify, retrieve, forecast, synthesise. If the platform cannot name the class, it cannot route. It can only default.",[218,798,799,802,803,805],{},[221,800,801],{},"Run quote."," What a loop will take, as a share of a usage limit or as money, so you can quote and cap it. See ",[201,804,289],{"href":288},". Seats hide routing. A quote per run makes it visible.",[218,807,808,811,812,815],{},[221,809,810],{},"Model-agnostic."," The platform can call more than one provider. That is a menu, not a policy, until it ",[238,813,814],{},"chooses"," by task class. Extra logos with a hidden flagship default is lock-in with branding.",[218,817,818,821],{},[221,819,820],{},"Always-flagship."," Marketing for “we use the best model.” A classify job does not need a long-context reasoner. Quality theatre is a cost event.",[189,823,824,825,828,829,832],{},"Routing is also not ",[221,826,827],{},"fine-tuning"," (changing a model’s weights), and it is not ",[221,830,831],{},"orchestration"," (what steps exist). You can orchestrate a brilliant graph and still send every node to the flagship. You can fine-tune a compact model and still need a policy that sends classify there. Evaluate them separately.",[210,834,349],{"id":348},[189,836,837],{},"Teams do not wake up and choose waste. They inherit a default.",[215,839,840,846,852],{},[218,841,842,845],{},[221,843,844],{},"Single-model shop."," Every label, every search, every memo calls the same flagship. Finance sees one invoice and cannot split labelling from reasoning. You cannot cap what you cannot see.",[218,847,848,851],{},[221,849,850],{},"User-picked dropdown."," Power users pick the most expensive option “to be safe.” New hires copy that habit. Routing is now a training problem. Training problems do not survive quarter-end.",[218,853,854,857],{},[221,855,856],{},"Always-flagship as quality theatre."," Best for whom? Best for a forecast interpolation is often a time-series path, not a frontier model inventing a number that looks fine in a short demo.",[189,859,860,862],{},[201,861,204],{"href":203}," keeps showing the operational gap: regular use, then a struggle to scale because cost and workflow were never treated as a system. Routing is that system for inference. Without it, scale is a token bill.",[189,864,865,869,870,873,874,333,878,882],{},[201,866,868],{"href":867},"https:\u002F\u002Fwww.gartner.com\u002Fen\u002Farticles\u002Fai-governance-trism","Gartner’s AI TRiSM framing"," implies you can ",[238,871,872],{},"see"," which model ran, on which data, at what cost. A platform that cannot show that is not ready, regardless of its red-team slides. ",[201,875,877],{"href":876},"https:\u002F\u002Feur-lex.europa.eu\u002Feli\u002Freg\u002F2022\u002F2554\u002Foj","DORA",[201,879,881],{"href":880},"https:\u002F\u002Feur-lex.europa.eu\u002Feli\u002Fdir\u002F2022\u002F2555\u002Foj","NIS2"," change the evidence question: you should understand ICT dependencies. “We are not sure which model ran last Tuesday” is a dependency you cannot explain.",[189,884,885,886,888],{},"Choosing a model is choosing a brain. Choosing tools is choosing hands. Decide them separately. A compact model that extracts a refund still cannot write it without a quoted named signer if that is the workstream policy. Routing does not replace ",[201,887,718],{"href":391},". Governance does not replace routing. You need both.",[189,890,891],{},"Red flags: “we support many providers” with no task-class map; seat pricing that includes unlimited flagship; users pick the model in production; classify and memo share a model id in the demo; no quote before a run expands; fallback is “switch the dropdown”; graph does not record model id per step; forecasting done by an LLM in the demo on a short series.",[210,893,895],{"id":894},"what-to-look-for","What to look for",[215,897,898,904,910,919,925],{},[218,899,900,903],{},[221,901,902],{},"They can refuse the flagship."," Run a classify-only job and show the model id. If classify used the same model as the memo, routing is a slide. Refusal is the proof. Support for many models is not.",[218,905,906,909],{},[221,907,908],{},"Task-class map you can read."," Classify → compact. Retrieve → embeddings, not stuffing hundreds of tickets into a long window. Forecast → a time-series path, not a frontier model interpolating a spreadsheet. Reason → frontier. If they cannot name the classes, they cannot route them.",[218,911,912,915,916,269],{},[221,913,914],{},"Quotes before the run expands."," Operators see an estimate and can set a workstream cap. Seat licences hide routing. Unlimited flagship under a seat is always-flagship with a predictable opex line. See ",[201,917,918],{"href":288},"Total cost of ownership for enterprise AI",[218,920,921,924],{},[221,922,923],{},"Fallback is a logged promotion",", not “users will switch the dropdown.” Compact models fail on novel schemas and policy-edge language. Temporarily raise that class, budget-aware, then revert. A promotion without a log is a silent cost change. A dropdown is a training problem.",[218,926,927,930,931,269],{},[221,928,929],{},"The graph records the model id per step."," Six months later you can answer “which model drafted this?” without grepping provider dashboards. That is audit as well as cost. See ",[201,932,934],{"href":933},"\u002Fblog\u002Fwhat-is-a-lifecycle-graph\u002F","What is a lifecycle graph",[189,936,937,938,941,942,944],{},"Ask for four artefacts from one workstream run: model id per step; cost per task class; a classify job that did ",[238,939,940],{},"not"," use the flagship; a forecast that did ",[238,943,940],{}," use an LLM as the estimator. If the vendor can only show a chat transcript and a blended token total, routing is not in the product.",[189,946,947],{},"Why each artefact matters: model id is Measure in NIST language. Cost per class is how Finance splits labelling from reasoning. Classify-without-flagship is the refusal test. Forecast-without-LLM is whether they know the difference between narration and estimation. Demos are short series. Production is seasonality, holidays, and missing days.",[369,949,951],{"id":950},"what-a-live-demo-should-prove","What a live demo should prove",[189,953,954],{},"Do not accept a provider logo wall.",[294,956,957,964,967,970,973,976,979],{},[218,958,959,960,963],{},"Run one ",[201,961,962],{"href":267},"workstream"," with extract, classify, retrieve, and a memo.",[218,965,966],{},"Show model id per step. Classify and extract are compact. The memo may be frontier.",[218,968,969],{},"Show cost (or tokens) per task class, quoted before the run grows, with a cap on the workstream.",[218,971,972],{},"Force a compact failure on a novel schema. Show a logged promotion, then a revert — not a user switching a dropdown.",[218,974,975],{},"Show a forecast path that is not an LLM interpolating a sheet. Narration can still be frontier.",[218,977,978],{},"Query the graph: which model drafted this payload? Answer from the ledger, not from a provider console.",[218,980,981],{},"Confirm a named-role gate still applies regardless of which model drafted. Routing chooses the brain. Governance still decides the write.",[189,983,984],{},"If they pass by opening a playground and picking “best,” you evaluated a dropdown.",[210,986,471],{"id":470},[189,988,989],{},"Nimbus treats routing as an operating decision tied to workstream steps: task type, sensitivity, and cost — not “best everywhere.” Compact models handle extract. Frontier models are reserved for synthesis. Spend is quoted per workstream and visible per step.",[189,991,992,993,996],{},"Release gates apply regardless of which model drafted the payload. Connectors stay read-only by default. Routing decides ",[238,994,995],{},"which brain"," reads them. Governance still decides whether anything writes.",[189,998,519,999,1002,1003,1005],{},[201,1000,44],{"href":1001},"\u002Fmodels\u002F",". For the unit of account, ",[201,1004,289],{"href":288},". Score the four artefacts above. The product claim is the policy, not the catalogue.",[210,1007,542],{"id":541},[369,1009,1011],{"id":1010},"is-model-agnostic-the-same-as-routing","Is “model-agnostic” the same as routing?",[189,1013,1014,1015,1017],{},"No. Model-agnostic means more than one provider. Routing means it ",[221,1016,814],{}," by task class, with a default that is cheap where cheap is correct. A hidden always-flagship default is lock-in with extra logos. Ask what happens if the operator never touches a dropdown. If the answer is flagship, you have your policy.",[369,1019,1021],{"id":1020},"should-operators-ever-pick-a-model","Should operators ever pick a model?",[189,1023,1024],{},"Rarely, and as an override. Production operators should brief outcomes. If quality depends on each user knowing which model is good at JSON, you have staffed a routing department by accident. Overrides should be logged, budget-aware, and exceptional. A dropdown on every run is how always-flagship returns through the side door.",[369,1026,1028],{"id":1027},"why-not-put-forecasting-in-the-llm-if-the-numbers-look-fine-in-the-demo","Why not put forecasting in the LLM if the numbers look fine in the demo?",[189,1030,1031],{},"Demos are short series. Production is seasonality, holidays, and missing days. Keep narration on the frontier model and estimation on a time-series path. A fluent number is not a control. The warehouse or the statistical path already owns the number. RAG plus a frontier model is for policy language, not for revenue by region.",[369,1033,1035],{"id":1034},"what-if-legal-requires-a-single-approved-model-vendor","What if legal requires a single approved model vendor?",[189,1037,1038,1039,1042],{},"Routing still applies ",[221,1040,1041],{},"inside"," that vendor’s catalogue: compact vs frontier vs embedding. Single-vendor is a contracting constraint, not an excuse to max tokens. Model-agnostic is nice. Task-class mapping inside one catalogue is the control. Do not skip routing because the RFP named one lab.",[369,1044,1046],{"id":1045},"does-nis2-or-dora-change-the-routing-question","Does NIS2 or DORA change the routing question?",[189,1048,1049,1050,1053],{},"They change the ",[221,1051,1052],{},"evidence"," question. DORA and NIS2 expect you to understand ICT dependencies. “We are not sure which model ran last Tuesday” is a dependency you cannot explain. Record model id per step on the graph. That is enough to start. You do not need a new product category. You need Measure.",[210,1055,649],{"id":648},[189,1057,1058,529,1060,1062,1063,269],{},[201,1059,268],{"href":267},[201,1061,467],{"href":466},", and ",[201,1064,1065],{"href":288},"what is AI token economics",[210,1067,659],{"id":658},[215,1069,1070,1075,1080,1085,1090,1095,1100],{},[218,1071,1072],{},[201,1073,1074],{"href":763},"OpenAI API pricing",[218,1076,1077],{},[201,1078,1079],{"href":767},"Anthropic pricing",[218,1081,1082],{},[201,1083,1084],{"href":772},"Stanford HAI, 2025 AI Index Report",[218,1086,1087],{},[201,1088,1089],{"href":203},"McKinsey, The state of AI in 2025",[218,1091,1092],{},[201,1093,1094],{"href":867},"Gartner, AI TRiSM \u002F AI governance",[218,1096,1097],{},[201,1098,1099],{"href":876},"DORA (Regulation 2022\u002F2554)",[218,1101,1102],{},[201,1103,1104],{"href":880},"NIS2 (Directive 2022\u002F2555)",{"title":169,"searchDepth":170,"depth":170,"links":1106},[1107,1108,1109,1112,1113,1120,1121],{"id":212,"depth":170,"text":213},{"id":348,"depth":170,"text":349},{"id":894,"depth":170,"text":895,"children":1110},[1111],{"id":950,"depth":678,"text":951},{"id":470,"depth":170,"text":471},{"id":541,"depth":170,"text":542,"children":1114},[1115,1116,1117,1118,1119],{"id":1010,"depth":678,"text":1011},{"id":1020,"depth":678,"text":1021},{"id":1027,"depth":678,"text":1028},{"id":1034,"depth":678,"text":1035},{"id":1045,"depth":678,"text":1046},{"id":648,"depth":170,"text":649},{"id":658,"depth":170,"text":659},"Model routing is a policy that uses a cheaper model for simple steps and a stronger model only when the task needs it — not a dropdown labelled “best.”",{"eyebrow":699,"title":1124},"What to check before you trust the router",[1126,1129,1132],{"question":1127,"answer":1128},"What should a routing policy specify?","Which steps get a cheaper model, which steps require a stronger one, and the cost ceiling for the job.",{"question":1130,"answer":1131},"What is a warning sign?","A single default labelled best, with no record of which model actually ran.",{"question":1133,"answer":1134},"How do you know routing is working?","Simple steps cost less, hard steps still pass a check, and finance can see the split.","\u002Fblog\u002Fwhat-to-look-for-in-model-routing",{"title":744,"description":1122},"evaluation","blog\u002Fwhat-to-look-for-in-model-routing",[1137,1140,1141,1142],"models","routing","cost","S2koGZz5yT1ldL2TBYZ9zTTxEZHhrGXowtqnJjeGBHw",{"id":1145,"title":1146,"archived":163,"authors":1147,"badge":1149,"body":1150,"date":697,"definedTerm":164,"department":164,"description":1582,"extension":172,"eyebrow":164,"faqHeader":1583,"faqs":1585,"footerBand":164,"headline":164,"image":164,"industry":164,"jobType":164,"listed":131,"location":164,"navigation":131,"openRoles":164,"pageLayout":164,"path":1595,"relatedHeading":164,"seo":1596,"series":713,"sitemap":131,"status":164,"stem":1597,"subhead":164,"tags":1598,"video":164,"whyJoin":164,"workplaceType":164,"__hash__":1600},"content\u002Fblog\u002Fwhat-is-an-ai-workstream.md","What is an AI Workstream",[1148],{"name":182,"to":183},{"label":185},{"type":166,"value":1151,"toc":1558},[1152,1159,1162,1165,1173,1176,1178,1253,1261,1263,1269,1272,1291,1294,1319,1322,1324,1329,1337,1342,1347,1355,1357,1363,1368,1374,1380,1386,1392,1395,1403,1405,1412,1415,1422,1435,1437,1441,1444,1448,1451,1455,1458,1462,1468,1472,1477,1481,1484,1488,1495,1499,1502,1506,1511,1515,1518,1522,1525,1529,1534,1536,1544,1546],[189,1153,1154,1155,1158],{},"An AI workstream is a ",[221,1156,1157],{},"shared workspace for one job",": a brief, the tools allowed, the people and AI on it, a budget, and a finish line.",[189,1160,1161],{},"A Slack channel with a bot is not a job. It is a room. Anyone can paste anything. The bot never knows when the work is done. Next quarter, nobody can say which systems were in play or who was allowed to change them.",[189,1163,1164],{},"If you cannot name the systems in scope and the approval policy on writes, you do not have a workstream. You have a conversation.",[189,1166,1167,1168,1172],{},"Software teams already learned this. Work lives in issues and tickets, not in unbounded chat. ",[201,1169,1171],{"href":1170},"https:\u002F\u002Fwww.atlassian.com\u002Fagile\u002Fproject-management\u002Fepics-stories-themes","Atlassian’s epics and stories"," are named packages with a boundary. AI operations are still catching up. The missing object is often the work package: a place where the job actually lives.",[189,1174,1175],{},"The analogy is not decoration. Tickets have a requester, a scope, an owner, and a closed state. Copilots have a thread. Threads do not archive cleanly, do not attach least-privilege connectors, and do not carry a named signer. When AI started touching live systems, the thread stopped being a sufficient container.",[210,1177,213],{"id":212},[215,1179,1180,1186,1192,1207,1213,1221,1227,1235,1245],{},[218,1181,1182,1185],{},[221,1183,1184],{},"Brief."," What this job is for, and what “done” means. At work, “Q3 regional discount hygiene” is a brief. “My stuff” is not.",[218,1187,1188,1191],{},[221,1189,1190],{},"Connector."," A secure link to a live system (CRM, ERP, Drive). Attach what this job needs — not every system “just in case.” Default is read-only.",[218,1193,1194,1197,1198,1201,1202,1206],{},[221,1195,1196],{},"Scope \u002F least privilege."," Only the data and tools required for ",[238,1199,1200],{},"this"," job. ",[201,1203,1205],{"href":1204},"https:\u002F\u002Fwww.hhs.gov\u002Fhipaa\u002Ffor-professionals\u002Fprivacy\u002Fguidance\u002Fminimum-necessary-requirement\u002Findex.html","HIPAA’s minimum necessary"," is the same instinct: do not attach every system to every task. GDPR purpose limitation is the privacy-law cousin.",[218,1208,1209,1212],{},[221,1210,1211],{},"God workspace."," One org-wide chat that can see every folder and every CRM object because setup was easier. At work, this is how recruiting sees finance forecasts.",[218,1214,1215,1218,1219,269],{},[221,1216,1217],{},"Agent team."," The AI specialists assigned to the job. The workstream is the stage; the team is the cast. See ",[201,1220,607],{"href":606},[218,1222,1223,1226],{},[221,1224,1225],{},"Named signer."," Who must approve a write. At work, this is a role that already owns that class of change.",[218,1228,1229,1232,1233,269],{},[221,1230,1231],{},"Budget."," The spend ceiling for the job. See ",[201,1234,289],{"href":288},[218,1236,1237,1240,1241,269],{},[221,1238,1239],{},"Wiki section."," The asserted playbooks this job may load. See ",[201,1242,1244],{"href":1243},"\u002Fblog\u002Fwhat-is-a-company-wiki-for-ai-agents\u002F","What is a company wiki for AI agents",[218,1246,1247,1250,1251,269],{},[221,1248,1249],{},"Lifecycle Graph."," The chain this job emits as it runs. See ",[201,1252,934],{"href":933},[189,1254,1255,1256,1260],{},"An ",[201,1257,1259],{"href":1258},"\u002Fblog\u002Fwhat-is-an-agentic-workflow\u002F","agentic workflow"," is the sequence of steps. The workstream is the durable instance those steps run inside. A workflow definition without a workstream is a script on someone’s laptop. A workstream without a workflow is a folder with no process.",[210,1262,349],{"id":348},[189,1264,1265,1266,1268],{},"Without a boundary, two departments sharing an AI tool will either over-share (the recruiting job can see finance forecasts) or under-share (people export spreadsheets to personal ChatGPT). The workstream is the compromise: enough context to do ",[238,1267,1200],{}," job, not the whole company.",[189,1270,1271],{},"It affects you if work:",[215,1273,1274,1277,1280,1283,1286],{},[218,1275,1276],{},"touches more than one system",[218,1278,1279],{},"involves more than one role",[218,1281,1282],{},"can change a live record",[218,1284,1285],{},"needs a budget you can attribute",[218,1287,1288,1289],{},"must still be explainable after people leave — see ",[201,1290,934],{"href":933},[189,1292,1293],{},"Open workstreams the way you would open a ticket:",[215,1295,1296,1303,1310,1313,1316],{},[218,1297,1298,1299,1302],{},"One workstream per ",[221,1300,1301],{},"outcome",", not per person. “Q3 regional discount hygiene” can have several humans. “My stuff” cannot be governed or archived.",[218,1304,1305,1306,1309],{},"Attach the ",[221,1307,1308],{},"minimum"," connectors.",[218,1311,1312],{},"Set the write policy on day one, even if you start read-only.",[218,1314,1315],{},"Reuse templates, not last month’s chat thread.",[218,1317,1318],{},"Close or archive when the job ends. A sprint that never ends is not a sprint.",[189,1320,1321],{},"A standing “Ask AI” workstream with org-wide connectors recreates the copilot, including the blast radius.",[369,1323,372],{"id":371},[189,1325,1326,1328],{},[221,1327,377],{}," Chargeback becomes possible because the job is named. Close workstreams can attach ERP read-only, load the close checklist from the wiki, and keep GTM out of the ledger. A company-wide AI pool with no workstream attribution is a shared kitchen.",[189,1330,1331,1333,1334,1336],{},[221,1332,383],{}," Scope is a processing purpose. A workstream for a renewal can include legal and go-to-market on ",[238,1335,1200],{}," goal without merging their entire universes. Legal also gets a closed state: when the job ends, retention follows the type of record instead of an immortal channel.",[189,1338,1339,1341],{},[221,1340,397],{}," This is the ticket analogue they already wanted. Ops should refuse god workspaces, insist on a finish line, and treat human wait as a status, not a side conversation in Slack.",[189,1343,1344,1346],{},[221,1345,403],{}," Cross-functional launches finally have a place that is not a merged Slack. GTM still should not get finance’s ERP “for context.” Templates beat copying last quarter’s thread, which silently copies last quarter’s over-attached connectors.",[189,1348,1349,1351,1352,1354],{},[221,1350,409],{}," Least privilege is now a product object, not a memo. Connectors default to read-only. Adding a write path is a deliberate change to ",[238,1353,1200],{}," job, not a tenant-wide toggle. A workstream that never closes is a standing access grant.",[369,1356,414],{"id":413},[189,1358,1359,1362],{},[221,1360,1361],{},"One workstream per person."," You cannot archive “my stuff.” You cannot attribute it. You cannot apply least privilege.",[189,1364,1365,1367],{},[221,1366,1211],{}," Setup is easier. Blast radius is the company.",[189,1369,1370,1373],{},[221,1371,1372],{},"ChatGPT Projecthe unit."," Some files, some instructions. Typically no connector-level least privilege, quoted writes, spend caps, or lasting record. Fine for personal research. Not an operations unit.",[189,1375,1376,1379],{},[221,1377,1378],{},"Too small."," If setup exceeds the job, use a lighter sanctioned copilot path. Do not open a workstream to rewrite one sentence.",[189,1381,1382,1385],{},[221,1383,1384],{},"Too large."," If you cannot explain the purpose in one sentence, or you keep attaching “one more connector,” split.",[189,1387,1388,1391],{},[221,1389,1390],{},"Never closing."," Standing rooms recreate Slack, including the archaeology problem.",[189,1393,1394],{},"Good looks like: one outcome, minimum connectors, write policy on day one, wiki sections subscribed, budget capped, named signer, archive when done. Failure looks like an org-wide copilot with every OAuth grant and a channel that outlives the campaign.",[189,1396,1397,1398,1402],{},"The ",[201,1399,1401],{"href":1400},"\u002Fblog\u002Fwhat-is-an-enterprise-ai-operating-system\u002F","enterprise AI OS"," metaphor is isolation plus I\u002FO plus state. The workstream is the isolation unit. Without it, connectors, wiki, and agent teams have nowhere to attach that an auditor could name.",[210,1404,471],{"id":470},[189,1406,1407,1408,1411],{},"In Nimbus, workstreams are how ",[201,1409,1410],{"href":1258},"agentic workflows"," become company objects rather than a file only one engineer can run.",[189,1413,1414],{},"Each workstream carries a brief, wiki sections (approved playbooks), connector attachments (read-only by default), agent team assignment, spend budget, release policy on writes, and nodes on the Lifecycle Graph.",[189,1416,1417,1418,1421],{},"Cross-department work is multiple teams on one workstream, not a merged Slack. Operators open this themselves; the point of an ",[201,1419,1420],{"href":1400},"enterprise AI operating system"," is that the job folder is a product, not a forward-deployed spreadsheet.",[189,1423,519,1424,1426,1427,529,1430,529,1433,269],{},[201,1425,31],{"href":528},". Related product: ",[201,1428,500],{"href":1429},"\u002Fproduct\u002Fagent-teams\u002F",[201,1431,488],{"href":1432},"\u002Fproduct\u002Fwiki\u002F",[201,1434,39],{"href":532},[210,1436,542],{"id":541},[369,1438,1440],{"id":1439},"is-a-chatgpt-project-a-workstream","Is a ChatGPT “Project” a workstream?",[189,1442,1443],{},"It is a weak analogue: some files, some custom instructions. It typically lacks connector-level least privilege, quoted writes, spend caps, and a lasting record. Useful for personal research. Not an operations unit.",[369,1445,1447],{"id":1446},"how-small-is-too-small","How small is too small?",[189,1449,1450],{},"If the setup cost exceeds the job, use a lighter sanctioned copilot path. Do not create a workstream to rewrite one sentence.",[369,1452,1454],{"id":1453},"how-large-is-too-large","How large is too large?",[189,1456,1457],{},"If you cannot explain the purpose in one sentence, or you keep attaching “one more connector,” split.",[369,1459,1461],{"id":1460},"can-one-workstream-serve-multiple-departments","Can one workstream serve multiple departments?",[189,1463,1464,1465,1467],{},"Yes — go-to-market and legal on a renewal, for example. They share ",[238,1466,1200],{}," goal’s scope, not each other’s entire universe.",[369,1469,1471],{"id":1470},"how-do-we-budget-them","How do we budget them?",[189,1473,1474,1475,269],{},"Caps per workstream, plus an organisation pool. Chargeback by workstream beats “the AI bill.” See ",[201,1476,289],{"href":288},[369,1478,1480],{"id":1479},"is-a-slack-channel-with-a-bot-enough-if-we-add-a-approve-command","Is a Slack channel with a bot enough if we add a \u002Fapprove command?",[189,1482,1483],{},"No. A command is not connector least privilege, a quoted payload, a durable chain, or an archive policy. It is still a room.",[369,1485,1487],{"id":1486},"who-is-allowed-to-open-a-workstream","Who is allowed to open a workstream?",[189,1489,1490,1491,269],{},"Whoever is allowed to open that class of job in analogue life — with the same instinct as who may open a ticket or a change request. An “AI team” bottleneck recreates the waitlist that causes ",[201,1492,1494],{"href":1493},"\u002Fblog\u002Fwhat-is-shadow-ai\u002F","shadow AI",[369,1496,1498],{"id":1497},"what-happens-when-the-job-ends","What happens when the job ends?",[189,1500,1501],{},"Close or archive. Revoke standing connector usefulness. Keep the reconstructable chain according to retention, not the entire chat.",[369,1503,1505],{"id":1504},"do-we-need-a-workstream-for-read-only-analysis","Do we need a workstream for read-only analysis?",[189,1507,1508,1509,269],{},"When the analysis crosses systems, roles, or must be replayed later, yes. When it is personal drafting with no live-system scope, a sanctioned copilot may be enough. See ",[201,1510,462],{"href":461},[369,1512,1514],{"id":1513},"how-do-wiki-and-connectors-differ-inside-a-workstream","How do wiki and connectors differ inside a workstream?",[189,1516,1517],{},"Wiki is asserted policy the job must follow. Connectors are live systems the job may read (and, if enabled, write). Mixing them into one “knowledge” pile is how Drive folklore overwrites the playbook.",[369,1519,1521],{"id":1520},"can-we-keep-one-standing-workstream-for-ask-anything","Can we keep one standing workstream for “ask anything”?",[189,1523,1524],{},"You can. You will recreate the copilot, including over-share. Standing Q&A belongs on a tightly scoped, read-only path if it exists at all.",[369,1526,1528],{"id":1527},"how-does-this-relate-to-agent-teams","How does this relate to agent teams?",[189,1530,1531,1532,269],{},"The workstream is the job. The agent team is the cast assigned to it. Changing the cast does not change the brief, the connectors, or the signer. See ",[201,1533,607],{"href":606},[210,1535,649],{"id":648},[189,1537,1538,333,1541,269],{},[201,1539,1540],{"href":1400},"What is an enterprise AI operating system",[201,1542,1543],{"href":1258},"What is an agentic workflow",[210,1545,659],{"id":658},[215,1547,1548,1553],{},[218,1549,1550],{},[201,1551,1552],{"href":1170},"Atlassian, epics, stories, and themes",[218,1554,1555],{},[201,1556,1557],{"href":1204},"HHS, HIPAA minimum necessary requirement",{"title":169,"searchDepth":170,"depth":170,"links":1559},[1560,1561,1565,1566,1580,1581],{"id":212,"depth":170,"text":213},{"id":348,"depth":170,"text":349,"children":1562},[1563,1564],{"id":371,"depth":678,"text":372},{"id":413,"depth":678,"text":414},{"id":470,"depth":170,"text":471},{"id":541,"depth":170,"text":542,"children":1567},[1568,1569,1570,1571,1572,1573,1574,1575,1576,1577,1578,1579],{"id":1439,"depth":678,"text":1440},{"id":1446,"depth":678,"text":1447},{"id":1453,"depth":678,"text":1454},{"id":1460,"depth":678,"text":1461},{"id":1470,"depth":678,"text":1471},{"id":1479,"depth":678,"text":1480},{"id":1486,"depth":678,"text":1487},{"id":1497,"depth":678,"text":1498},{"id":1504,"depth":678,"text":1505},{"id":1513,"depth":678,"text":1514},{"id":1520,"depth":678,"text":1521},{"id":1527,"depth":678,"text":1528},{"id":648,"depth":170,"text":649},{"id":658,"depth":170,"text":659},"An AI workstream is one shared job: a brief, the tools allowed, the people and AI on it, a budget, and a finish line.",{"eyebrow":699,"title":1584},"One job, one shared place",[1586,1589,1592],{"question":1587,"answer":1588},"What is an AI workstream?","A shared workspace for one job: a brief, the tools allowed, the people and AI on it, a budget, and a finish line.",{"question":1590,"answer":1591},"Is a Slack channel with a bot a workstream?","No. A channel has no brief, no allowed-tool list, and no finish line a person can refuse.",{"question":1593,"answer":1594},"What makes it finish?","A named outcome and someone who can say the work does not go out.","\u002Fblog\u002Fwhat-is-an-ai-workstream",{"title":1146,"description":1582},"blog\u002Fwhat-is-an-ai-workstream",[713,1599,716,718],"workstreams","GqysvBAVOydIlMQ2bTUJZvFY1Z9D-lp9m7VqL4XVPA0",{"enabled":163,"message":1602,"linkLabel":79,"linkHref":80,"id":1603,"title":1604,"archived":163,"authors":164,"badge":164,"body":1605,"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":1609,"relatedHeading":164,"seo":1610,"series":164,"sitemap":163,"status":164,"stem":1611,"subhead":164,"tags":164,"video":164,"whyJoin":164,"workplaceType":164,"__hash__":1612},"We're hiring! Join the team building the Sentient Enterprise.","content\u002Fshared\u002Fhiring.md","Hiring banner",{"type":166,"value":1606,"toc":1607},[],{"title":169,"searchDepth":170,"depth":170,"links":1608},[],"\u002Fshared\u002Fhiring",{"title":1604,"description":169},"shared\u002Fhiring","1zs3boivKda1e-b-hAyuNcmZSKjZUAXmecnwHVgcHzk",{"fold":1614,"id":1618,"title":1619,"archived":163,"authors":164,"badge":164,"body":1620,"date":164,"definedTerm":164,"department":164,"description":169,"extension":172,"eyebrow":164,"faqHeader":164,"faqs":164,"footerBand":1624,"headline":164,"image":164,"industry":164,"jobType":164,"listed":131,"location":164,"navigation":131,"openRoles":164,"pageLayout":164,"path":1628,"relatedHeading":164,"seo":1629,"series":164,"sitemap":163,"status":164,"stem":1630,"subhead":164,"tags":164,"video":164,"whyJoin":164,"workplaceType":164,"__hash__":1631},{"headline":1615,"description":1616,"primaryLabel":8,"primaryTo":1617,"secondaryLabel":736,"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":1621,"toc":1622},[],{"title":169,"searchDepth":170,"depth":170,"links":1623},[],{"headline":1625,"description":1626,"primaryLabel":8,"primaryTo":1617,"secondaryLabel":1627,"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":1619,"description":169},"shared\u002Fcta","eYqahyaPnbp8GKrWpoORbZdtmkWmgHr5F61ZHOnb8sY",1792999485790]