[{"data":1,"prerenderedAt":1588},["ShallowReactive",2],{"site-nav-content":3,"blog:/blog/what-is-model-routing":163,"blog-index-copy":630,"blog:/blog/what-is-model-routing:surround":651,"hiring-banner-content":1557,"site-cta-content":1569},{"header":4,"productNav":9,"nav":42,"footer":61,"askAI":116,"id":147,"title":148,"archived":149,"authors":150,"badge":150,"body":151,"date":150,"department":150,"description":155,"extension":158,"eyebrow":150,"faqHeader":150,"faqs":150,"footerBand":150,"headline":150,"image":150,"industry":150,"jobType":150,"listed":149,"location":150,"navigation":115,"openRoles":150,"pageLayout":150,"path":159,"relatedHeading":150,"seo":160,"series":150,"sitemap":115,"status":150,"stem":161,"subhead":150,"tags":150,"video":150,"whyJoin":150,"workplaceType":150,"__hash__":162},{"productLabel":5,"loginLabel":6,"contactLabel":7,"contactSalesLabel":8},"Product","Log in","Contact","Get started for free",[10,14,18,22,26,30,34,38],{"label":11,"to":12,"description":13},"Overview","/overview","Seven layers. One closed loop.",{"label":15,"to":16,"description":17},"Conflux","/product/conflux","Where your team, workstreams, and agents meet.",{"label":19,"to":20,"description":21},"Agent Teams","/product/agent-teams","Specialist teams - governed from day one.",{"label":23,"to":24,"description":25},"Lifecycle Graph","/product/lifecycle-graph","Intelligence that compounds across every interaction.",{"label":27,"to":28,"description":29},"Company Wiki","/product/wiki","Playbooks and policies where expertise stays.",{"label":31,"to":32,"description":33},"Workstreams","/product/workstreams","From brief to signed-off deliverable on one canvas.",{"label":35,"to":36,"description":37},"Perception Console","/product/perception","Ask your whole business in plain English.",{"label":39,"to":40,"description":41},"Governance","/product/governance","Frontier AI you can actually sign off on.",[43,46,49,52,55,58],{"label":44,"to":45},"Models","/models",{"label":47,"to":48},"Pricing","/pricing",{"label":50,"to":51},"Integrations","/integrations",{"label":53,"to":54},"Security","/security",{"label":56,"to":57},"Partners","/partners",{"label":59,"to":60},"Insights","/blog",{"productHeading":5,"companyHeading":62,"legalHeading":63,"docsLabel":64,"docsUrl":65,"statementLines":66,"copyright":69,"companyLinks":70,"legalLinks":85,"socialLinks":95,"bottomLinks":105},"Company","Legal","Docs","https://docs.gonimbus.ai",[67,68],"Stop training someone else's model.","Control your AI.","© 2026 Nimbus Intelligence, Inc. All rights reserved.",[71,72,73,74,75,77,80,83],{"label":47,"to":48},{"label":50,"to":51},{"label":53,"to":54},{"label":59,"to":60},{"label":76,"to":57},"Partner Program",{"label":78,"to":79},"Careers","/careers",{"label":81,"to":82},"System status","/status",{"label":7,"to":84},"/contact",[86,89,92],{"label":87,"to":88},"Terms of Service","/terms",{"label":90,"to":91},"Privacy Policy","/privacy",{"label":93,"to":94},"Compliance","/compliance",[96,99,102],{"label":97,"href":98},"LinkedIn","https://www.linkedin.com/company/gonimbusai/",{"label":100,"href":101},"X","https://x.com/gonimbusai",{"label":103,"href":104},"Instagram","https://www.instagram.com/gonimbus_ai/",[106,108,110,111,112],{"label":107,"to":88},"Terms",{"label":109,"to":91},"Privacy",{"label":93,"to":94},{"label":81,"to":82},{"label":113,"to":114,"external":115},"LLMs.txt","/llms.txt",true,{"text":117,"prompt":118},"Ask AI about Nimbus",{"I'm researching enterprise intelligence platforms and want to know how Nimbus combines perception, collaboration, and autonomous agents to drive strategic decision-making":119,"platforms":121},{" Summarize the highlights from Nimbus's website":120},"https://gonimbus.ai",[122,127,132,137,142],{"name":123,"label":124,"icon":125,"hrefPrefix":126},"chatgpt","ChatGPT","simple-icons:openai","https://chatgpt.com/?prompt=",{"name":128,"label":129,"icon":130,"hrefPrefix":131},"perplexity","Perplexity","mdi:magnify","https://www.perplexity.ai/search/new?q=",{"name":133,"label":134,"icon":135,"hrefPrefix":136},"grok","Grok","simple-icons:x","https://x.com/i/grok?text=",{"name":138,"label":139,"icon":140,"hrefPrefix":141},"claude","Claude","simple-icons:anthropic","https://claude.ai/new?q=",{"name":143,"label":144,"icon":145,"hrefPrefix":146},"google-ai","Google AI","simple-icons:google","https://www.google.com/search?udm=50&aep=11&q=","content/shared/nav.md","Site navigation",false,null,{"type":152,"value":153,"toc":154},"minimark",[],{"title":155,"searchDepth":156,"depth":156,"links":157},"",2,[],"md","/shared/nav",{"title":148,"description":155},"shared/nav","p6IsjEfjcrwkjwvGsPcuEkpoJSeUn2vJXPTxLWQfw9M",{"id":164,"title":165,"archived":149,"authors":166,"badge":169,"body":171,"date":619,"department":150,"description":620,"extension":158,"eyebrow":150,"faqHeader":150,"faqs":150,"footerBand":150,"headline":150,"image":150,"industry":150,"jobType":150,"listed":149,"location":150,"navigation":115,"openRoles":150,"pageLayout":150,"path":621,"relatedHeading":150,"seo":622,"series":623,"sitemap":115,"status":150,"stem":624,"subhead":150,"tags":625,"video":150,"whyJoin":150,"workplaceType":150,"__hash__":629},"content/blog/what-is-model-routing.md","What is Model Routing",[167],{"name":168,"to":120},"Nimbus Research",{"label":170},"Explainer",{"type":152,"value":172,"toc":594},[173,182,185,188,204,209,275,282,289,293,296,303,327,339,344,350,359,365,371,377,381,387,397,403,409,415,421,429,444,452,456,459,462,473,477,481,484,488,491,495,498,502,505,509,512,516,522,526,529,533,536,540,543,547,552,556,559,563,566,570,576,580],[174,175,176,177,181],"p",{},"Labs ship a ladder of models: small and cheap, large and expensive. ",[178,179,180],"strong",{},"Model routing"," is the policy above that ladder: use a cheaper, faster model for simple steps, and a stronger model only when the task needs it.",[174,183,184],{},"It is not a dropdown labelled “best.” Someone typing “use the best model” for a classify-this-ticket step is how a flagship invoice gets burned on work a compact model could have finished in a second.",[174,186,187],{},"Done well, extract runs on compact models and hard reasoning runs on frontier models. Done poorly, every step hits the most expensive model, spend becomes a surprise, and “we use the best model” becomes an unexamined religion.",[174,189,190,197,198,203],{},[191,192,196],"a",{"href":193,"rel":194},"https://openai.com/api/pricing/",[195],"nofollow","OpenAI"," and ",[191,199,202],{"href":200,"rel":201},"https://www.anthropic.com/pricing",[195],"Anthropic"," publish those ladders in public. The prices change. The shape does not: input, output, and sometimes tools meter differently, and the top rung is many times the compact rung. Finance cannot treat “always flagship” as a quality culture. It is an unbudgeted preference.",[205,206,208],"h2",{"id":207},"words-youll-hear","Words you’ll hear",[210,211,212,219,225,231,237,248,258,264],"ul",{},[213,214,215,218],"li",{},[178,216,217],{},"Frontier / flagship model."," The strongest (and usually most expensive) model a lab currently sells. At work, this is for judgment: does this clause violate the playbook?",[213,220,221,224],{},[178,222,223],{},"Compact / small model."," Faster and cheaper. Often enough for extract, classify, and summarise. At work, this is “pull the fields from the export.”",[213,226,227,230],{},[178,228,229],{},"Cascade."," Try cheap first; spend the expensive call only when the cheap one is not enough.",[213,232,233,236],{},[178,234,235],{},"Fallback."," If a provider is down or over budget, send the step somewhere else.",[213,238,239,242,243,247],{},[178,240,241],{},"Orchestration."," What steps exist. Different from routing, which is ",[244,245,246],"em",{},"which brain"," each step uses. You can orchestrate a brilliant multi-agent graph and still send every node to the flagship.",[213,249,250,253,254,257],{},[178,251,252],{},"Quality bar."," The reject-rate or rework threshold that decides whether a compact model is good enough on ",[244,255,256],{},"this"," job.",[213,259,260,263],{},[178,261,262],{},"Data residency / data class."," A cheap endpoint may be forbidden for a class of records. Routing is then a compliance table, not only a cost table.",[213,265,266,269,270,274],{},[178,267,268],{},"NTU."," The normalised unit routing is trying to protect. See ",[191,271,273],{"href":272},"what-is-ai-token-economics","What is AI token economics",".",[174,276,277,278,281],{},"Routing is also not ",[178,279,280],{},"fine-tuning"," (changing a model’s weights). Fine-tuning is a research and ops programme. Routing is an operating policy over models you already buy.",[174,283,284,285,274],{},"Constraints that belong in the route table: data residency, evaluation (you cannot route on vibes), and security (a model with web tools is a different actor than a model with none). Choosing a model is choosing a brain. Choosing tools is choosing hands. Decide them separately. See ",[191,286,288],{"href":287},"what-is-model-context-protocol","What is Model Context Protocol",[205,290,292],{"id":291},"why-you-should-care","Why you should care",[174,294,295],{},"It affects you if you pay the bill, or if quality on a step is load-bearing.",[174,297,298,299,302],{},"Talk about it as a ",[178,300,301],{},"budget and quality conversation",", not as an ML research project:",[210,304,305,311,317],{},[213,306,307,310],{},[178,308,309],{},"Tag the steps."," Extracting fields from an export is not the same as arguing whether a clause violates policy. If your platform cannot name steps, it cannot route them.",[213,312,313,316],{},[178,314,315],{},"Set a quality bar per step."," “Compact model until human reject rate exceeds X on this job.” Without a bar, routing becomes “always escalate because someone was once unhappy.”",[213,318,319,322,323,274],{},[178,320,321],{},"Keep the gate regardless of model."," A cheap model with a write tool is still a write tool. See ",[191,324,326],{"href":325},"what-is-write-back-governance","What is write-back governance",[174,328,329,330,333,334,338],{},"A spend ceiling without routing still lets every step hit the flagship until the ceiling kills the run. Routing is how you stay under the ceiling ",[244,331,332],{},"and"," finish the job. See ",[191,335,337],{"href":336},"what-is-an-agentic-workflow","What is an agentic workflow"," for why loops without stops dominate the bill.",[340,341,343],"h3",{"id":342},"what-changes-by-role","What changes by role",[174,345,346,349],{},[178,347,348],{},"Finance."," Routing is the practical lever on unit cost. Quotes should assume the policy, not the flagship. Finance should ask for approved-updates per NTU, and for evidence that extract steps are not on the top rung. Locking one vendor forever is a pricing and outage choice; routing across providers is a second tape measure.",[174,351,352,355,356,358],{},[178,353,354],{},"Legal."," Data class and residency can forbid the cheap endpoint. Legal should sit on the route table for those classes, not discover them on an invoice. Customer-facing language may need a stronger model ",[244,357,332],{}," a named signer; routing does not replace the gate.",[174,360,361,364],{},[178,362,363],{},"Operations."," Steps must be named or you cannot route them. Ops should own fallbacks when a provider is down, and should refuse a single “best” toggle that bypasses the table.",[174,366,367,370],{},[178,368,369],{},"Go-to-market."," Quality anxiety is strongest here. Measure reject rates on the job. A compact model that extracts next steps may be fine; a compact model that invents a concession is not. Routing on one unhappy anecdote will pin every step to flagship.",[174,372,373,376],{},[178,374,375],{},"Security."," A model with browsing or unconstrained tools is a different actor. Routing should not silently add hands. Prompt injection plus a flagship model plus write tools is a worse combination than a compact extract-only step behind a fail-closed gate.",[340,378,380],{"id":379},"what-people-get-wrong","What people get wrong",[174,382,383,386],{},[178,384,385],{},"Always the smartest model."," Use the weakest model that meets the quality bar for that step. Flagship is for judgment, not for labelling.",[174,388,389,392,393,274],{},[178,390,391],{},"Routing as multi-agent."," Several agents is a cast. Routing is which brain each step pays for. See ",[191,394,396],{"href":395},"what-is-multi-agent-ai","What is multi-agent AI",[174,398,399,402],{},[178,400,401],{},"Routing as fine-tuning."," Different programme.",[174,404,405,408],{},[178,406,407],{},"Dropdown labelled “best.”"," That is not a policy. It is a preference that cannot be audited.",[174,410,411,414],{},[178,412,413],{},"Dropping the write gate for a “trusted” model."," Trust the gate. Models change weekly.",[174,416,417,420],{},[178,418,419],{},"Routing on vibes."," One anecdote becomes a permanent escalate. Measure rework.",[174,422,423,424,428],{},"Good looks like: named steps, a route table with cost, quality bar, and data class, cascade where it helps, fallback across providers, gates independent of model, falling unit cost as the ",[191,425,427],{"href":426},"what-is-a-company-wiki-for-ai-agents","wiki"," reduces re-derivation. Failure looks like flagship-everywhere and a board slide about the bill.",[174,430,431,432,435,436,439,440,443],{},"Adjacent ideas worth keeping separate: ",[191,433,434],{"href":272},"token economics"," is quote, cap, and attribute. Routing is which rung of the ladder a named step is allowed to use. ",[191,437,438],{"href":395},"Multi-agent AI"," is how many specialist roles run. You can route a single agent, and you can send a whole agent team to the flagship by mistake. ",[191,441,442],{"href":287},"MCP"," is hands, not brains: do not let a compact extract step inherit a write tool because “the helper was available.”",[174,445,446,447,451],{},"Evaluation has to live on the job, not in a model-arena screenshot. A compact model that extracts fields with a low reject rate is a success even if it would lose a public chatbot bake-off. A flagship model that drafts a concession the wiki forbids is a failure even if it is eloquent. Tie routing reviews to ",[191,448,450],{"href":449},"what-is-an-ai-workstream","workstream"," outcomes — approved writes, rejects, rework — the same way you would review any other operating policy.",[205,453,455],{"id":454},"how-this-shows-up-in-nimbus","How this shows up in Nimbus",[174,457,458],{},"Nimbus treats routing as an operating decision tied to workstream steps: task type, sensitivity, and cost — not “best everywhere.” Release gates apply regardless of which model drafted the payload.",[174,460,461],{},"NTU quotes and ceilings sit around that policy so operators see a number before they commit. Everyday extract should not consume flagship credits.",[174,463,464,465,467,468,470,471,274],{},"See ",[191,466,44],{"href":45},". For the unit of account routing sits inside, ",[191,469,273],{"href":272},". Product: ",[191,472,31],{"href":32},[205,474,476],{"id":475},"questions-people-actually-ask","Questions people actually ask",[340,478,480],{"id":479},"should-we-always-use-the-smartest-model","Should we always use the smartest model?",[174,482,483],{},"No. Use the weakest model that meets the quality bar for that step. Flagship is for judgment, not for labelling.",[340,485,487],{"id":486},"will-routing-make-answers-worse","Will routing make answers worse?",[174,489,490],{},"It can, if you under-route hard steps. Measure rejects and rework on the job. Do not route on a single anecdote.",[340,492,494],{"id":493},"is-this-the-same-as-having-several-agents","Is this the same as having several agents?",[174,496,497],{},"No. Several agents is a cast. Routing is which brain each step pays for.",[340,499,501],{"id":500},"can-we-lock-one-vendor-forever","Can we lock one vendor forever?",[174,503,504],{},"You can. You will pay for it in price, outages, and lock-in. Routing across providers is how finance keeps a second tape measure.",[340,506,508],{"id":507},"what-is-a-cascade","What is a cascade?",[174,510,511],{},"Try the cheap model first. Escalate only when a confidence or quality check says the cheap pass is not enough. It is a tactic inside a policy, not a substitute for naming steps.",[340,513,515],{"id":514},"does-a-better-model-remove-the-need-for-a-wiki","Does a better model remove the need for a wiki?",[174,517,518,519,274],{},"No. Stronger models are better at sounding like policy. Asserted playbooks still win over Drive folklore. See ",[191,520,521],{"href":426},"What is a company wiki for AI agents",[340,523,525],{"id":524},"how-do-we-set-a-quality-bar","How do we set a quality bar?",[174,527,528],{},"Start with human reject rate and rework on that step. “Compact until rejects exceed X on this workstream” is a bar. “People like the flagship” is not.",[340,530,532],{"id":531},"should-customer-facing-copy-always-use-the-flagship","Should customer-facing copy always use the flagship?",[174,534,535],{},"Not always. It should always use a human gate if it asserts a term or a price. Model size does not absorb Air Canada-style risk.",[340,537,539],{"id":538},"what-if-the-cheap-endpoint-is-in-the-wrong-region","What if the cheap endpoint is in the wrong region?",[174,541,542],{},"Then it is not cheap; it is forbidden. Put residency in the route table beside price.",[340,544,546],{"id":545},"does-routing-replace-spend-caps","Does routing replace spend caps?",[174,548,549,550,274],{},"No. Caps stop unbounded loops. Routing makes legitimate work affordable under the cap. You want both. See ",[191,551,273],{"href":272},[340,553,555],{"id":554},"can-the-model-choose-its-own-successor","Can the model choose its own successor?",[174,557,558],{},"Letting the model always escalate is how every step becomes flagship. Escalation should be a policy check, not a preference the model expresses.",[340,560,562],{"id":561},"how-does-this-show-up-in-an-rfp","How does this show up in an RFP?",[174,564,565],{},"Ask whether steps are named, whether gates apply regardless of model, and whether finance sees a normalised unit. “We use the best models” is not an answer.",[205,567,569],{"id":568},"related-reading","Related reading",[174,571,572,197,574,274],{},[191,573,273],{"href":272},[191,575,396],{"href":395},[205,577,579],{"id":578},"sources","Sources",[210,581,582,588],{},[213,583,584],{},[191,585,587],{"href":193,"rel":586},[195],"OpenAI API pricing",[213,589,590],{},[191,591,593],{"href":200,"rel":592},[195],"Anthropic pricing",{"title":155,"searchDepth":156,"depth":156,"links":595},[596,597,602,603,617,618],{"id":207,"depth":156,"text":208},{"id":291,"depth":156,"text":292,"children":598},[599,601],{"id":342,"depth":600,"text":343},3,{"id":379,"depth":600,"text":380},{"id":454,"depth":156,"text":455},{"id":475,"depth":156,"text":476,"children":604},[605,606,607,608,609,610,611,612,613,614,615,616],{"id":479,"depth":600,"text":480},{"id":486,"depth":600,"text":487},{"id":493,"depth":600,"text":494},{"id":500,"depth":600,"text":501},{"id":507,"depth":600,"text":508},{"id":514,"depth":600,"text":515},{"id":524,"depth":600,"text":525},{"id":531,"depth":600,"text":532},{"id":538,"depth":600,"text":539},{"id":545,"depth":600,"text":546},{"id":554,"depth":600,"text":555},{"id":561,"depth":600,"text":562},{"id":568,"depth":156,"text":569},{"id":578,"depth":156,"text":579},"2026-08-17","Model routing is using a cheaper, faster model for simple steps and a stronger model only when the task needs it — a policy, not a dropdown labelled “best.”","/blog/what-is-model-routing",{"title":165,"description":620},"explainer","blog/what-is-model-routing",[623,626,627,628],"model-routing","token-economics","cost","cT8uO7Fl6bffuYVI7nA4O6i99dYBnBc0RAb25dGjpU0",{"hero":631,"id":633,"title":634,"archived":149,"authors":150,"badge":150,"body":635,"date":150,"department":150,"description":639,"extension":158,"eyebrow":640,"faqHeader":150,"faqs":150,"footerBand":641,"headline":150,"image":150,"industry":150,"jobType":150,"listed":149,"location":150,"navigation":115,"openRoles":150,"pageLayout":150,"path":60,"relatedHeading":647,"seo":648,"series":150,"sitemap":115,"status":150,"stem":649,"subhead":150,"tags":150,"video":150,"whyJoin":150,"workplaceType":150,"__hash__":650},{"filename":632},"u2221455217_Flat_design_of_a_futuristic_minimalist_landscape__5d589295-cdea-4ea9-a262-be766881accf_1.png","content/blog/index.md","Exploring the future of intelligence.",{"type":152,"value":636,"toc":637},[],{"title":155,"searchDepth":156,"depth":156,"links":638},[],"Deep dives into pre-cognitive intelligence, sentient enterprises, and the evolving landscape of AI-driven business transformation.","Latest Research",{"headline":642,"description":643,"primaryLabel":644,"primaryTo":645,"secondaryLabel":646,"secondaryTo":12},"Stay at the frontier.","Subscribe for product updates and new insights.","Subscribe","/newsletter","Explore the platform","More research",{"title":634,"description":639},"blog/index","eK1RCXdDW8nfLSyKRXGB1mJm9FAmhAO6GWXwNKOMNVE",[652,1092],{"id":653,"title":654,"archived":149,"authors":655,"badge":657,"body":658,"date":619,"department":150,"description":1083,"extension":158,"eyebrow":150,"faqHeader":150,"faqs":150,"footerBand":150,"headline":150,"image":150,"industry":150,"jobType":150,"listed":149,"location":150,"navigation":115,"openRoles":150,"pageLayout":150,"path":1084,"relatedHeading":150,"seo":1085,"series":623,"sitemap":115,"status":150,"stem":1086,"subhead":150,"tags":1087,"video":150,"whyJoin":150,"workplaceType":150,"__hash__":1091},"content/blog/what-is-multi-agent-ai.md","What is Multi-Agent AI",[656],{"name":168,"to":120},{"label":170},{"type":152,"value":659,"toc":1059},[660,663,666,679,681,737,751,758,760,763,766,777,780,801,808,814,816,821,831,836,841,846,848,854,860,866,872,878,884,887,893,908,914,920,922,932,935,947,949,953,956,960,963,967,973,977,980,984,987,991,994,998,1003,1007,1010,1014,1017,1021,1026,1030,1033,1037,1040,1042,1049,1051],[174,661,662],{},"Multi-agent AI is more than one AI specialist handing work to each other, the way legal already reviews a go-to-market draft. They share a goal, pass intermediate work, and stop for a human when duties collide.",[174,664,665],{},"It is not “more than one model call.” If you can rename your agents to “saved prompts” and nothing breaks, you do not have multi-agent AI. You have prompt folders. Three chat tabs labelled Research, CRM, and Legal are still one person copying between windows.",[174,667,668,669,672,673,678],{},"Companies already hand work between departments. Multi-agent AI is useful when those hand-offs ",[244,670,671],{},"are"," the job. It is not useful as a prestige multiplier on a task one specialist should finish. ",[191,674,677],{"href":675,"rel":676},"https://www.anthropic.com/engineering/building-effective-agents",[195],"Anthropic’s note on building effective agents"," is mostly about workflows and stops, not about collecting a zoo of bots.",[205,680,208],{"id":207},[210,682,683,689,695,701,707,713,721,727],{},[213,684,685,688],{},[178,686,687],{},"Single-agent."," One policy, one tool set, one conversation. The human is the only coordinator. Right for many tasks: rewrite this email, explain this clause.",[213,690,691,694],{},[178,692,693],{},"Multi-agent."," Role specialisation, shared state scoped to the job, arbitration when agents disagree, and a stop — including “wait for approval.”",[213,696,697,700],{},[178,698,699],{},"Orchestrator."," A coordinator that assigns work to specialists. Not a licence to give every specialist the same production login.",[213,702,703,706],{},[178,704,705],{},"Agent team."," Nimbus’s product name for a department-shaped specialist group (finance, revenue, operations) that persists, rather than a zoo of user-owned bots.",[213,708,709,712],{},[178,710,711],{},"Separation of duties."," The specialist that recommends a CRM update is not the same principal that executes it without a quote.",[213,714,715,718,719,274],{},[178,716,717],{},"Shared state."," The job’s brief, wiki sections, and artefacts — not a pile of private chats. At work, this is the ",[191,720,450],{"href":449},[213,722,723,726],{},[178,724,725],{},"Arbitration."," What happens when specialists disagree. At work, legal’s “do not send” should beat go-to-market’s “looks fine,” and a named human still releases.",[213,728,729,732,733,736],{},[178,730,731],{},"Cast versus brain."," Several agents is a cast. ",[191,734,180],{"href":735},"what-is-model-routing"," is which brain each step pays for.",[174,738,739,740,743,744,747,748,750],{},"Tools are ",[178,741,742],{},"hands",". They are not ",[178,745,746],{},"roles",". A common plug so AI apps can use the same tools is useful plumbing — see ",[191,749,288],{"href":287}," — and it is also how a “multi-agent” demo quietly becomes one actor with every tool on the belt.",[174,752,753,754,757],{},"An ",[191,755,756],{"href":336},"agentic workflow"," is the sequence. Multi-agent AI is the cast. Mixing those words is how vendors sell extra model calls as organisation design.",[205,759,292],{"id":291},[174,761,762],{},"Coordination is an org-chart problem, not a model problem.",[174,764,765],{},"It affects you if:",[210,767,768,771,774],{},[213,769,770],{},"one “god agent” would need every production login",[213,772,773],{},"legal must review a draft before anyone writes CRM",[213,775,776],{},"next quarter, “why did we change this?” must still be answerable",[174,778,779],{},"Known failure modes:",[781,782,783,789,795],"ol",{},[213,784,785,788],{},[178,786,787],{},"Parallel single-agent."," Three chat tabs. No shared state. The user is the message bus.",[213,790,791,794],{},[178,792,793],{},"Agent sprawl."," Dozens of custom agents with overlapping tools and unclear write rights.",[213,796,797,800],{},[178,798,799],{},"Orchestration without memory."," A beautiful run that discards the outcome when the worker exits.",[174,802,803,804,274],{},"Multi-agent does not reduce accountability. It concentrates it on the release gate. See ",[191,805,807],{"href":806},"what-is-human-in-the-loop-ai","What is human-in-the-loop AI",[174,809,810,811,274],{},"Do not think in “number of agents.” Think in ",[178,812,813],{},"jobs that already have hand-offs",[340,815,343],{"id":342},[174,817,818,820],{},[178,819,348],{}," A finance-shaped agent team can draft a journal against the close checklist without inheriting GTM’s CRM write connector. Separation of duties is the point. Finance should still be the named signer on the ledger. Extra agents are not extra authorisation.",[174,822,823,825,826,830],{},[178,824,354],{}," Review-before-send is a real hand-off. Legal-shaped specialists should not need People Ops files “for context.” Legal also cares that internal agent debate is not treated as a signature. The ",[191,827,829],{"href":828},"what-is-a-lifecycle-graph","lifecycle graph"," should show the human at release.",[174,832,833,835],{},[178,834,363],{}," Persist teams, do not spawn a bot per user. Ops should refuse sprawl, insist on shared workstream state, and keep fail-closed writes outside the cast. Incident reviews need one chain, not three private transcripts.",[174,837,838,840],{},[178,839,369],{}," Cross-functional launches already look like this: GTM drafts, legal redlines, finance checks the discount. Encode that. Do not encode a god agent that can do all three logins. Time-to-approved-write still beats number-of-agents as a metric.",[174,842,843,845],{},[178,844,375],{}," Sprawl is an identity problem. Each specialist with overlapping write tools is another path to production. Prompt injection that turns one specialist into a tool-caller should still die at the gate. Least privilege applies per role, not “the swarm is trusted.”",[340,847,380],{"id":379},[174,849,850,853],{},[178,851,852],{},"Saved prompts as agents."," If renaming them changes nothing, they were prompts.",[174,855,856,859],{},[178,857,858],{},"Chat tabs as multi-agent."," The user is still the bus.",[174,861,862,865],{},[178,863,864],{},"More agents as more quality."," Coordination cost is real. Start from existing hand-offs.",[174,867,868,871],{},[178,869,870],{},"Agents as signers."," Internal critics reduce garbage. They are not the named human.",[174,873,874,877],{},[178,875,876],{},"Every specialist gets every tool."," That recreates the god agent with extra steps.",[174,879,880,883],{},[178,881,882],{},"Orchestration without a workstream."," No scope, no budget, no memory.",[174,885,886],{},"Good looks like: department-shaped teams that persist, inherit workstream scope (wiki, connectors, NTU budget), disagree in the open, and stop for a named signer. Failure looks like a folder of user-owned bots and a demo where five helpers share one production key.",[174,888,889,890,892],{},"A useful test: draw the analogue hand-off first. If legal already reviews a go-to-market draft before a customer sees it, you have a candidate for two specialist roles on one ",[191,891,450],{"href":449},". If one analyst extracts a table, you have a candidate for a single tool-using agent. If nobody can name the hand-off, you are inventing a cast for a play that does not exist — and you will invent overlapping tools to keep them busy.",[174,894,895,896,899,900,903,904,907],{},"Spend follows the same test. Extra specialists mean extra model calls. Without ",[191,897,898],{"href":735},"routing"," and an ",[191,901,902],{"href":272},"NTU"," ceiling, “let them debate” is an unbounded loop. Debate that never reaches a named signer is also not ",[191,905,906],{"href":806},"human-in-the-loop","; it is theatre with more speakers.",[174,909,910,911,913],{},"Memory is the other test. If the hand-off is not on the ",[191,912,829],{"href":828},", next quarter’s question — “why did we change this?” — has no answer except whoever still remembers the swarm. That is not multi-agent AI. That is parallel chat.",[174,915,916,919],{},[191,917,918],{"href":20},"Agent teams"," in Nimbus are meant to look like the departments you already have, not like a prompt gallery. If your org chart does not contain a role, do not invent an agent for it. If your org chart does contain a role that must review before release, do not skip it because a single flagship model offered to “do it all.” Number of agents is a vanity metric. Named hand-offs are not.",[205,921,455],{"id":454},[174,923,924,925,928,929,931],{},"Nimbus implements multi-agent AI as ",[178,926,927],{},"agent teams",", not as a folder of user-owned bots. Teams persist. They inherit ",[191,930,450],{"href":449}," scope — wiki sections, connectors, spend budget — and they participate in the same release process as any other actor.",[174,933,934],{},"Connectors stay read-only by default. Agent teams can draft. They cannot waive the gate. The Lifecycle Graph records the hand-offs as work, not as a swarm mystery.",[174,936,464,937,939,940,942,943,274],{},[191,938,918],{"href":20},". An ",[191,941,756],{"href":336}," is the sequence. Multi-agent AI is the cast. The OS around them is ",[191,944,946],{"href":945},"what-is-an-enterprise-ai-operating-system","What is an enterprise AI operating system",[205,948,476],{"id":475},[340,950,952],{"id":951},"isnt-this-just-several-chatgpts-talking","Isn’t this just several ChatGPTs talking?",[174,954,955],{},"Not if they share one job, one scope, and one stop. Several chats with no shared state is still you, copying.",[340,957,959],{"id":958},"do-we-need-multi-agent-ai-for-everything","Do we need multi-agent AI for everything?",[174,961,962],{},"No. A single tool-using agent is enough for many tasks. Add specialists when duties already split in the organisation.",[340,964,966],{"id":965},"does-each-agent-need-its-own-model","Does each agent need its own model?",[174,968,969,970,274],{},"Often yes for cost and quality. Classification rarely needs the flagship. Tricky policy interpretation often does. See ",[191,971,972],{"href":735},"What is model routing",[340,974,976],{"id":975},"who-is-accountable-when-several-agents-worked-on-it","Who is accountable when several agents worked on it?",[174,978,979],{},"The named human at release — not “the swarm.” Internal critics can reduce garbage that reaches the person. They are not the signer.",[340,981,983],{"id":982},"how-is-this-different-from-an-agentic-workflow","How is this different from an agentic workflow?",[174,985,986],{},"The workflow is the sequence of steps and stops. Multi-agent is whether more than one specialist role executes those steps. You can have a workflow with one agent.",[340,988,990],{"id":989},"what-is-an-agent-team-in-nimbus","What is an agent team in Nimbus?",[174,992,993],{},"A department-shaped specialist group that persists and inherits the workstream’s wiki, connectors, and budget — not a user-owned custom GPT.",[340,995,997],{"id":996},"can-agents-approve-each-others-writes","Can agents approve each other’s writes?",[174,999,1000,1001,274],{},"They can flag problems. Execution still needs a named signer and a fail-closed gate. See ",[191,1002,326],{"href":325},[340,1004,1006],{"id":1005},"why-not-one-god-agent-with-every-connector","Why not one god agent with every connector?",[174,1008,1009],{},"Because least privilege and separation of duties already exist in the company. Encoding the org chart is safer than encoding a superuser.",[340,1011,1013],{"id":1012},"how-do-we-avoid-agent-sprawl","How do we avoid agent sprawl?",[174,1015,1016],{},"One team per function that already exists, assigned onto jobs, with overlapping tools treated as an incident. Do not let every operator publish a bot.",[340,1018,1020],{"id":1019},"does-mcp-make-us-multi-agent","Does MCP make us multi-agent?",[174,1022,1023,1024,274],{},"No. MCP is how a host calls tools. Many helpers on one belt can still be one actor. See ",[191,1025,288],{"href":287},[340,1027,1029],{"id":1028},"how-do-disagreements-get-recorded","How do disagreements get recorded?",[174,1031,1032],{},"On the decision chain: what was proposed, what was objected to, what the human signed. If disagreement evaporates with the session, you have orchestration without memory.",[340,1034,1036],{"id":1035},"will-more-agents-stop-hallucinations","Will more agents stop hallucinations?",[174,1038,1039],{},"They can catch some errors the way a second reader can. They do not replace asserted wiki, citations, or a person on commitments. Air Canada-style fiction is a gate problem, not a cast-size problem.",[205,1041,569],{"id":568},[174,1043,1044,197,1047,274],{},[191,1045,1046],{"href":449},"What is an AI workstream",[191,1048,946],{"href":945},[205,1050,579],{"id":578},[210,1052,1053],{},[213,1054,1055],{},[191,1056,1058],{"href":675,"rel":1057},[195],"Anthropic, Building effective agents",{"title":155,"searchDepth":156,"depth":156,"links":1060},[1061,1062,1066,1067,1081,1082],{"id":207,"depth":156,"text":208},{"id":291,"depth":156,"text":292,"children":1063},[1064,1065],{"id":342,"depth":600,"text":343},{"id":379,"depth":600,"text":380},{"id":454,"depth":156,"text":455},{"id":475,"depth":156,"text":476,"children":1068},[1069,1070,1071,1072,1073,1074,1075,1076,1077,1078,1079,1080],{"id":951,"depth":600,"text":952},{"id":958,"depth":600,"text":959},{"id":965,"depth":600,"text":966},{"id":975,"depth":600,"text":976},{"id":982,"depth":600,"text":983},{"id":989,"depth":600,"text":990},{"id":996,"depth":600,"text":997},{"id":1005,"depth":600,"text":1006},{"id":1012,"depth":600,"text":1013},{"id":1019,"depth":600,"text":1020},{"id":1028,"depth":600,"text":1029},{"id":1035,"depth":600,"text":1036},{"id":568,"depth":156,"text":569},{"id":578,"depth":156,"text":579},"Multi-agent AI is more than one AI specialist handing work to each other — the way legal already reviews a go-to-market draft — with a shared job, a stop, and a person who must approve before a live system changes.","/blog/what-is-multi-agent-ai",{"title":654,"description":1083},"blog/what-is-multi-agent-ai",[623,1088,1089,1090],"multi-agent","agent-teams","enterprise-ai","GptGcIymv6Z_AlmM_K7GOr3c6cofGsabWQwojdUy7Xc",{"id":1093,"title":288,"archived":149,"authors":1094,"badge":1096,"body":1097,"date":619,"department":150,"description":1549,"extension":158,"eyebrow":150,"faqHeader":150,"faqs":150,"footerBand":150,"headline":150,"image":150,"industry":150,"jobType":150,"listed":149,"location":150,"navigation":115,"openRoles":150,"pageLayout":150,"path":1550,"relatedHeading":150,"seo":1551,"series":623,"sitemap":115,"status":150,"stem":1552,"subhead":150,"tags":1553,"video":150,"whyJoin":150,"workplaceType":150,"__hash__":1556},"content/blog/what-is-model-context-protocol.md",[1095],{"name":168,"to":120},{"label":170},{"type":152,"value":1098,"toc":1525},[1099,1102,1126,1133,1142,1145,1147,1198,1207,1209,1212,1219,1238,1241,1250,1253,1255,1264,1269,1274,1279,1290,1292,1298,1304,1310,1320,1330,1339,1342,1362,1368,1370,1379,1382,1388,1390,1394,1397,1401,1404,1408,1414,1418,1421,1425,1428,1432,1435,1439,1442,1446,1449,1453,1458,1462,1467,1471,1479,1483,1486,1488,1494,1496],[174,1100,1101],{},"USB did not create a data-governance programme. A common plug let keyboards, cameras, and drives talk to any computer. It did not decide who may copy the finance drive, or whether a change to the ledger needs a signer.",[174,1103,1104,1107,1108,1113,1114,1119,1120,1125],{},[178,1105,1106],{},"Model Context Protocol (MCP)"," is the same kind of open standard for AI. ",[191,1109,1112],{"href":1110,"rel":1111},"https://www.anthropic.com/news/model-context-protocol",[195],"Anthropic announced it"," as a ",[191,1115,1118],{"href":1116,"rel":1117},"https://modelcontextprotocol.io/docs/2026-07-28/getting-started/intro",[195],"common plug"," so AI apps can use the same tools and files, instead of every vendor inventing a one-off connection. The ",[191,1121,1124],{"href":1122,"rel":1123},"https://modelcontextprotocol.io/specification/2025-11-25/index",[195],"specification"," standardises how a host calls tools and reads resources.",[174,1127,1128,1129,1132],{},"In one sentence: MCP is ",[178,1130,1131],{},"plumbing, not a company strategy",". It does not decide who may update Salesforce.",[174,1134,1135,1136,1141],{},"Developers already know this pattern from the ",[191,1137,1140],{"href":1138,"rel":1139},"https://microsoft.github.io/language-server-protocol/",[195],"Language Server Protocol",": one language server, many editors, instead of rewriting autocomplete for every IDE. MCP is that idea for tools an AI can call. LSP made language servers interchangeable. It did not make every language server a safe place for customer lists.",[174,1143,1144],{},"Before this standard, every AI product invented its own way to “use a tool.” Teams spent months redoing the same wiring. That cost was real. So is the over-read: “we support MCP” is not “we have enterprise governance.”",[205,1146,208],{"id":207},[210,1148,1149,1155,1161,1167,1173,1179,1189],{},[213,1150,1151,1154],{},[178,1152,1153],{},"Protocol / standard."," Agreed wiring so products can interoperate. At work, this is the USB cable, not the access-control list on the share.",[213,1156,1157,1160],{},[178,1158,1159],{},"MCP server / helper."," A small programme that says “here are the actions I can take, and here are the files I can show you.” At work, a helper that searches Drive as a superuser is still a superuser.",[213,1162,1163,1166],{},[178,1164,1165],{},"Host / client."," The AI application that calls the helper. At work, several hosts can speak MCP and still have completely different write gates — or none.",[213,1168,1169,1172],{},[178,1170,1171],{},"Tool call."," The AI asking that helper to search a folder, look up a ticket, post a message, or query a database.",[213,1174,1175,1178],{},[178,1176,1177],{},"Resource."," A file or record the helper can expose for reading.",[213,1180,1181,1184,1185,274],{},[178,1182,1183],{},"Connector (Nimbus)."," A supported, company-controlled integration to a live system — OAuth, scoped to the job, read-only by default. That is the operator-facing story. MCP may sit at a developer edge. It is not a substitute for connectors plus ",[191,1186,1188],{"href":1187},"what-is-ai-governance","governance",[213,1190,1191,1194,1195,1197],{},[178,1192,1193],{},"Least privilege."," Which tools this ",[191,1196,450],{"href":449}," may call. Importing every available helper is how a demo becomes one actor with every production login.",[174,1199,1200,1201,1204,1205,274],{},"A tool call can still change production data. The protocol will happily pass that change along. The company still has to decide whether that is allowed. Fail-closed writes, named signers, and quoted payloads live ",[244,1202,1203],{},"above"," the plug. See ",[191,1206,326],{"href":325},[205,1208,292],{"id":291},[174,1210,1211],{},"The plug is useful. It is also easy to over-read.",[174,1213,1214,1215,1218],{},"MCP does ",[178,1216,1217],{},"not"," decide:",[210,1220,1221,1224,1227,1230,1233],{},[213,1222,1223],{},"whose login is used",[213,1225,1226],{},"whether the AI may only read, or also change a live system",[213,1228,1229],{},"who must approve a change",[213,1231,1232],{},"how the company remembers what happened",[213,1234,1235,1236],{},"which model is used for the step — see ",[191,1237,972],{"href":735},[174,1239,1240],{},"Choosing a model is choosing a brain. This standard is choosing hands. A cheap model with dangerous tools is worse than a strong model with none. Decide them separately.",[174,1242,753,1243,1245,1246,1249],{},[191,1244,756],{"href":336}," that imports every available tool is a confused workflow. Plumbing is not a stop condition. ",[191,1247,677],{"href":675,"rel":1248},[195]," is about bounding tools and stops, not about collecting helpers.",[174,1251,1252],{},"It affects you if a vendor says “we support MCP” and you hear “we have enterprise governance.” Those are different sentences.",[340,1254,343],{"id":342},[174,1256,1257,1259,1260,1263],{},[178,1258,348],{}," A helper that can post a journal is a write path, protocol or not. Finance should ask whether the host quotes the payload and requires a named signer, not whether the wiring is MCP. Spend also sits above the plug: tool loops can burn ",[191,1261,1262],{"href":272},"NTUs"," without a ceiling.",[174,1265,1266,1268],{},[178,1267,354],{}," Processing agreements, purpose, and customer data in helpers running on laptops. Legal should not treat “open standard” as “safe.” A standard plug does not create a DPIA.",[174,1270,1271,1273],{},[178,1272,363],{}," Bounded tool belts per job. Ops should refuse workflows that attach every helper “for flexibility,” and should keep human wait and budget as stops regardless of how tools are wired.",[174,1275,1276,1278],{},[178,1277,369],{}," Faster wiring to CRM and Drive can be good — if the connector is still read-only by default. GTM should not confuse a demo that updates an opportunity via MCP with a governed release.",[174,1280,1281,1283,1284,1289],{},[178,1282,375],{}," This is the sharp edge. Helpers run with some identity. Superuser search is still superuser search. Prompt injection can trick a model into requesting a tool call; the ",[191,1285,1288],{"href":1286,"rel":1287},"https://genai.owasp.org/llm-top-10/",[195],"OWASP Top 10 for LLM applications"," is the relevant list. The protocol will not save you. Least privilege, read-only defaults, and fail-closed writes will.",[340,1291,380],{"id":379},[174,1293,1294,1297],{},[178,1295,1296],{},"MCP as governance."," Wiring is not a named signer.",[174,1299,1300,1303],{},[178,1301,1302],{},"MCP as the Salesforce strategy."," You still need identity, read versus write, an approver, and a record.",[174,1305,1306,1309],{},[178,1307,1308],{},"Refusing products that do not speak MCP."," Interoperable tools are a plus. Absence of MCP is not absence of a connector. Presence of MCP is not presence of governance.",[174,1311,1312,1315,1316,1319],{},[178,1313,1314],{},"Replacing the integration platform."," MCP standardises how an AI ",[244,1317,1318],{},"talks"," to a helper. Your identity, iPaaS, and change-control stack still have to exist.",[174,1321,1322,1325,1326,274],{},[178,1323,1324],{},"Assuming retrieval will respect permissions."," Only if the helper is built that way. See ",[191,1327,1329],{"href":1328},"what-is-enterprise-rag","What is enterprise RAG",[174,1331,1332,1335,1336,1338],{},[178,1333,1334],{},"Collecting every server."," A large tool belt is a confused ",[191,1337,756],{"href":336}," and a larger attack surface.",[174,1340,1341],{},"Good looks like: MCP where it reduces duplicate wiring; operator-facing connectors that stay scoped, encrypted, and read-only by default; writes only after sign-off; no belief that the spec implemented your control framework. Failure looks like a laptop running a superuser helper pointed at production.",[174,1343,1344,1345,1348,1349,1351,1352,1355,1356,1358,1359,1361],{},"Think of the stack in layers, or you will buy the wrong layer. MCP is how a host talks to a helper. A ",[191,1346,1347],{"href":51},"connector"," is how operators attach a live system to a ",[191,1350,450],{"href":449}," with OAuth and a read-only default. ",[191,1353,1354],{"href":325},"Write-back governance"," is whether a tool call that mutates production is allowed to execute. The ",[191,1357,829],{"href":828}," is whether you can still explain the call next quarter. ",[191,1360,180],{"href":735}," is which brain issued the call. None of those jobs is in the spec, and that is fine — specs should stay thin. Trouble starts when a thin spec is sold as the thick programme.",[174,1363,753,1364,1367],{},[191,1365,1366],{"href":945},"enterprise AI operating system"," sits above plumbing the way an OS sits above USB: isolation, permissions, I/O policy, and state. USB made accessories interchangeable. It did not decide who may format the finance drive.",[205,1369,455],{"id":454},[174,1371,1372,1373,1376,1377,274],{},"Nimbus’s operator-facing integrations are ",[178,1374,1375],{},"connectors",": scoped per workstream, encrypted per tenant, read-only by default. Action connectors write only after human sign-off. See ",[191,1378,50],{"href":51},[174,1380,1381],{},"MCP can be useful at developer edges. It is not the product’s answer to “who may change CRM.” Governance, wiki, and the Lifecycle Graph still sit above any plug.",[174,1383,464,1384,197,1386,274],{},[191,1385,39],{"href":40},[191,1387,31],{"href":32},[205,1389,476],{"id":475},[340,1391,1393],{"id":1392},"is-mcp-how-we-should-connect-salesforce","Is MCP how we should connect Salesforce?",[174,1395,1396],{},"Not by itself. You still need identity, read vs write rights, an approver, and a record. A standard plug does not provide those.",[340,1398,1400],{"id":1399},"should-we-refuse-products-that-dont-speak-mcp","Should we refuse products that don’t speak MCP?",[174,1402,1403],{},"No. Interoperable tools are a plus. Absence of MCP is not absence of a connector. Presence of MCP is not presence of governance.",[340,1405,1407],{"id":1406},"does-mcp-replace-our-integration-platform","Does MCP replace our integration platform?",[174,1409,1410,1411,1413],{},"No. It standardises how an AI ",[244,1412,1318],{}," to a helper. Your integration, identity, and change-control stack still has to exist.",[340,1415,1417],{"id":1416},"can-mcp-make-retrieval-respect-permissions","Can MCP make retrieval respect permissions?",[174,1419,1420],{},"Only if the helper is built that way. A tool that searches Drive as a superuser is still a superuser.",[340,1422,1424],{"id":1423},"is-mcp-the-same-as-a-nimbus-connector","Is MCP the same as a Nimbus connector?",[174,1426,1427],{},"No. A connector is the operator-facing, company-controlled integration: OAuth, workstream scope, read-only default. MCP is a developer wiring standard that might sit at an edge.",[340,1429,1431],{"id":1430},"does-the-spec-require-fail-closed-writes","Does the spec require fail-closed writes?",[174,1433,1434],{},"No. The spec does not require a quoted Salesforce payload, a named approver, or a fail-closed write. Those are product and policy choices.",[340,1436,1438],{"id":1437},"how-does-this-relate-to-usb-and-lsp","How does this relate to USB and LSP?",[174,1440,1441],{},"USB and LSP are the right analogies: interoperability of accessories and language servers. Neither is an access-control programme. Do not buy MCP as if it were.",[340,1443,1445],{"id":1444},"can-we-let-every-agent-team-install-their-own-mcp-servers","Can we let every agent team install their own MCP servers?",[174,1447,1448],{},"That is how you get overlapping write rights and no inventory. Treat helpers like production integrations: owners, scope, and a default of read-only.",[340,1450,1452],{"id":1451},"does-mcp-choose-the-model","Does MCP choose the model?",[174,1454,1455,1456,274],{},"No. Routing is which brain you pay for. MCP is which hands that brain can use. Decide them separately. See ",[191,1457,972],{"href":735},[340,1459,1461],{"id":1460},"is-we-support-mcp-a-good-rfp-answer-for-governance","Is “we support MCP” a good RFP answer for governance?",[174,1463,1464,1465,274],{},"It is a good answer for tool interoperability. For governance, ask about quotes, named signers, workstream scope, and the ",[191,1466,829],{"href":828},[340,1468,1470],{"id":1469},"what-is-the-security-failure-mode","What is the security failure mode?",[174,1472,1473,1474,1478],{},"A helper with broad credentials, a host with no gate, and a model tricked into calling ",[1475,1476,1477],"code",{},"update_record",". The protocol did its job. Your company did not.",[340,1480,1482],{"id":1481},"should-customer-facing-bots-get-mcp-tools-to-internal-crm","Should customer-facing bots get MCP tools to internal CRM?",[174,1484,1485],{},"That is how a public conversation inherits production hands. Scope tools as tightly as you would scope a workstream — usually, do not.",[205,1487,569],{"id":568},[174,1489,1490,197,1492,274],{},[191,1491,337],{"href":336},[191,1493,326],{"href":325},[205,1495,579],{"id":578},[210,1497,1498,1504,1510,1515,1520],{},[213,1499,1500],{},[191,1501,1503],{"href":1110,"rel":1502},[195],"Anthropic, Introducing the Model Context Protocol",[213,1505,1506],{},[191,1507,1509],{"href":1122,"rel":1508},[195],"Model Context Protocol specification",[213,1511,1512],{},[191,1513,1140],{"href":1138,"rel":1514},[195],[213,1516,1517],{},[191,1518,1288],{"href":1286,"rel":1519},[195],[213,1521,1522],{},[191,1523,1058],{"href":675,"rel":1524},[195],{"title":155,"searchDepth":156,"depth":156,"links":1526},[1527,1528,1532,1533,1547,1548],{"id":207,"depth":156,"text":208},{"id":291,"depth":156,"text":292,"children":1529},[1530,1531],{"id":342,"depth":600,"text":343},{"id":379,"depth":600,"text":380},{"id":454,"depth":156,"text":455},{"id":475,"depth":156,"text":476,"children":1534},[1535,1536,1537,1538,1539,1540,1541,1542,1543,1544,1545,1546],{"id":1392,"depth":600,"text":1393},{"id":1399,"depth":600,"text":1400},{"id":1406,"depth":600,"text":1407},{"id":1416,"depth":600,"text":1417},{"id":1423,"depth":600,"text":1424},{"id":1430,"depth":600,"text":1431},{"id":1437,"depth":600,"text":1438},{"id":1444,"depth":600,"text":1445},{"id":1451,"depth":600,"text":1452},{"id":1460,"depth":600,"text":1461},{"id":1469,"depth":600,"text":1470},{"id":1481,"depth":600,"text":1482},{"id":568,"depth":156,"text":569},{"id":578,"depth":156,"text":579},"Model Context Protocol is a common plug so AI apps can use the same tools, like USB for accessories. It is plumbing, not a company strategy, and it does not decide who may update Salesforce.","/blog/what-is-model-context-protocol",{"title":288,"description":1549},"blog/what-is-model-context-protocol",[623,1554,1375,1555],"mcp","tools","pyCsyLW6Bee4HSZZGalnpk2Ef8Ouo6MpiGQykRPKqaI",{"enabled":149,"message":1558,"linkLabel":78,"linkHref":79,"id":1559,"title":1560,"archived":149,"authors":150,"badge":150,"body":1561,"date":150,"department":150,"description":155,"extension":158,"eyebrow":150,"faqHeader":150,"faqs":150,"footerBand":150,"headline":150,"image":150,"industry":150,"jobType":150,"listed":149,"location":150,"navigation":115,"openRoles":150,"pageLayout":150,"path":1565,"relatedHeading":150,"seo":1566,"series":150,"sitemap":115,"status":150,"stem":1567,"subhead":150,"tags":150,"video":150,"whyJoin":150,"workplaceType":150,"__hash__":1568},"We're hiring! Join the team building the Sentient Enterprise.","content/shared/hiring.md","Hiring banner",{"type":152,"value":1562,"toc":1563},[],{"title":155,"searchDepth":156,"depth":156,"links":1564},[],"/shared/hiring",{"title":1560,"description":155},"shared/hiring","-6bioYD7lKYokGUVU3ff4hHTvB-sDyOMuCptKHnojfk",{"fold":1570,"id":1574,"title":1575,"archived":149,"authors":150,"badge":150,"body":1576,"date":150,"department":150,"description":155,"extension":158,"eyebrow":150,"faqHeader":150,"faqs":150,"footerBand":1580,"headline":150,"image":150,"industry":150,"jobType":150,"listed":149,"location":150,"navigation":115,"openRoles":150,"pageLayout":150,"path":1584,"relatedHeading":150,"seo":1585,"series":150,"sitemap":115,"status":150,"stem":1586,"subhead":150,"tags":150,"video":150,"whyJoin":150,"workplaceType":150,"__hash__":1587},{"headline":1571,"description":1572,"primaryLabel":8,"primaryTo":1573,"secondaryLabel":646,"secondaryTo":12},"Run frontier AI your business actually owns.","Governed agent swarms, 2,000+ integrations, and a knowledge graph that stays inside your walls. Free 7-day trial.","/checkout","content/shared/cta.md","Site CTAs",{"type":152,"value":1577,"toc":1578},[],{"title":155,"searchDepth":156,"depth":156,"links":1579},[],{"headline":1581,"description":1582,"primaryLabel":8,"primaryTo":1573,"secondaryLabel":1583,"secondaryTo":84},"See what governed AI looks like on your stack.","Connect your tools, run a workstream, and keep every decision on your ledger - free for 7 days.","Talk to our team","/shared/cta",{"title":1575,"description":155},"shared/cta","YHK6Fb8AvCPR1zZq7R_xiXUG0hwhP5UxHA8Ix52JQp4",1787194074603]