https://wyecliff.ai/assets/deploy-B_o1FEYP.js

1import{g as s,j as e,ay as h,n,l as y}from"./index-e6vuDWmB.js";import{C as m}from"./CapabilityPage-Cul5foGZ.js";import{C as g}from"./CapabilityGridHero-B2tgomfG.js";import{C as w}from"./cpu-BetmQgXV.js";import{A as i}from"./activity-B7ZyP-aY.js";import"./arrow-left-uVPLveLE.js";const f=[["path",{d:"m12 14 4-4",key:"9kzdfg"}],["path",{d:"M3.34 19a10 10 0 1 1 17.32 0",key:"19p75a"}]],r=s("gauge",f);const v=[["path",{d:"M15 6a9 9 0 0 0-9 9V3",key:"1cii5b"}],["circle",{cx:"18",cy:"6",r:"3",key:"1h7g24"}],["circle",{cx:"6",cy:"18",r:"3",key:"fqmcym"}]],a=s("git-branch",v),b="Deploy is where AI moves from a working system into daily operations. We put the model, data flows, permissions, monitoring, and human review in place so the system can be trusted after launch.";function k(l=16,u=13){const d=[[6,0],[5,1],[6,1],[7,1],[4,2],[6,2],[8,2],[3,3],[6,3],[9,3],[6,4],[6,5],[6,6],[6,7],[6,8],[6,9],[6,10]],c=l-6,p=u-5;return d.map(([o,t])=>[o+c,t+p]).filter(([o,t])=>o>=0&&t>=0)}function P(){return e.jsx(m,{slug:"deploy",name:"Deploy",summary:"Deploy is where AI moves from a working system into daily operations. We put the model, data flows, permissions, evals, monitoring, and human review paths in place so the system can be trusted after launch.",hero:e.jsx(g,{squares:k(16,13),headlineTop:"Deploy AI",headlineAccent:"People Trust",lede:b,pillLabel:"See Live Deployments"}),whatWeDo:{label:"What we deploy",headline:e.jsxs(e.Fragment,{children:["Copilot rollout and AI your ",e.jsx("span",{className:"text-purple-600",children:"team trusts."})]}),intro:"We take the working system and prepare it for real work: production data, live users, security rules, review queues, eval coverage, and monitoring your leaders can trust.",points:[{title:"Production Environment",body:"We deploy through your cloud, identity, network, and security patterns so the system fits how your IT team operates."},{title:"Evaluation Coverage",body:"We test prompts, model behavior, tool calls, and edge cases against examples from your real work."},{title:"Operational Monitoring",body:"We track cost, speed, accuracy, usage, and drift so the system does not quietly degrade."},{title:"Human Review",body:"We define which decisions can be automated, which need approval, and what audit trail should exist."}]},activities:{label:"What happens in the rollout",items:[{Icon:r,label:"Production Pilot"},{Icon:n,label:"Tenant Integration"},{Icon:a,label:"Eval Suite"},{Icon:y,label:"Controlled Cutover"},{Icon:i,label:"Live Monitoring"}],cards:[{Icon:r,title:"Deployment Planning",sub:"Pilot, integration, evaluation, cutover, and monitoring planned across IT, security, leaders, and operators."},{Icon:a,title:"Eval Suite",sub:"An eval suite written against real examples from your workflows, with regression testing on every prompt or model change."},{Icon:n,title:"Tenant Integration",sub:"Cloud deployment aligned with your IAM, secrets, network, and security requirements, using Anthropic and Microsoft partner patterns."},{Icon:i,title:"Launch And Monitoring",sub:"Human review paths for high risk work, plus launch reporting across usage, accuracy, latency, cost, and drift."}]},processSection:{headline:e.jsxs(e.Fragment,{children:[e.jsx("span",{className:"text-foreground",children:"How Deploy "}),e.jsx("span",{className:"text-purple-300",children:"Works."})]})},deliverables:[{Icon:w,roi:"Runs in your VPC",title:"Claude On Your Tenant",sub:"Deployed through Anthropic's API or Azure AI Foundry, inside your own network."},{Icon:a,roi:"No silent drift",title:"Eval Suite For Every Decision",sub:"Versioned eval cases, regression tested on every prompt or model change."},{Icon:r,roi:"Live from day one",title:"Observability And SLAs",sub:"Latency, cost, and accuracy dashboards, with error budgets sized to your risk."},{Icon:h,roi:"Audit ready",title:"Human In The Loop Guardrails",sub:"Risk scored escalation, review queues, and audit trails for regulated workflows."}],testimonial:{quote:"The contract intelligence platform went live in 30 days and now flags risk on every single contract we touch. The eval suite they built means we trust the AI to surface clauses our partners would have missed on a Frid

1ay afternoon.",author:"VP Operations",role:"National Insurance Brokerage"},engagement:{name:"Production Deployment",intro:"This is where the system becomes part of daily operations. We align with IT, security, leaders, and operators so launch is measured, monitored, and ready for real business volume.",timeline:"Structured around your cloud, risk profile, and cutover requirements",bullets:["Deployment planning across pilot, integration, evaluation, cutover, and monitoring","Anthropic Official Partner and Microsoft Cloud Partner deployment patterns","An eval suite written against real examples from your workflows","Cloud deployment aligned with your IAM, secrets, network, and security requirements","Human review paths for high risk or regulated work","Launch reporting across usage, accuracy, latency, cost, and drift"]},faqs:[{q:"How do you stop the AI from hallucinating in production?",a:"Three layers. First, every agent action runs against an eval suite of real examples from your operators, and anything that regresses gets blocked before it ships. Second, risk tiered actions go through a human review queue. A paralegal reviews every contract clause flagged as high risk, and a senior CSR confirms every order routing exception. Third, observability surfaces drift in production. If accuracy slides 1.5% week over week, your team gets paged."},{q:"Why Anthropic Claude specifically?",a:"For the workflows we ship, like contract review, claims triage, document intelligence, and agentic operations, Claude consistently outperforms in our internal evals on instruction following, citation discipline, and refusal accuracy. We are an Anthropic Official Partner because the model fits the regulated, judgment heavy work our clients run. We will happily deploy other models when they are a better fit, and we have Azure OpenAI deployments in production too. Claude is just the default for the work we do most."},{q:"Does this live in your environment or ours?",a:"Yours. Every production deployment lands inside your Azure or AWS tenant, behind your auth, on your network. The Anthropic API call is the only outbound dependency, and that goes through your egress, not ours. We do not run a hosted Wyecliff AI platform, so every system is operable by your team or another vendor without us."},{q:"What happens when Anthropic ships a new Claude model?",a:"Your eval suite is the answer. When a new model lands, we re run the suite against it, surface accuracy, latency, and cost deltas, and you decide whether to upgrade. Most clients on retainer upgrade within two weeks because the eval makes the decision boring. If you are not on retainer, we run the eval as a one off engagement, usually a half day."},{q:"How is this different from a Microsoft Copilot rollout?",a:"Copilot rollouts are about adoption of a general purpose tool. Production deployments are about shipping a specific, custom AI system to a specific workflow with named outputs and SLAs. We do both. Often you want both, shipping custom AI to the workflows where it matters and rolling Copilot out everywhere else."},{q:"How long until it is live in production?",a:"Most first deployments go live in four to eight weeks. We start with one workflow, ship it to a small group, and prove the eval numbers before we widen the rollout. Larger platform builds take longer, but you see a working system in production early instead of waiting months for a big reveal."}],prev:{slug:"build",label:"Build ← Apps & Agents"},next:{slug:"train",label:"Train → Enablement"}})}export{P as component};