/* ============================================================
   Shared data: verticals, copy, etc.
   Loaded as a Babel script — exports to window.
   ============================================================ */

const VERTICALS = [
  {
    id: "financial",
    num: "01",
    name: "Financial Services",
    short: "Banking · Capital Markets · Insurance",
    title: ["See the ", { em: "structure" }, " beneath the volatility."],
    lede: "Markets behave like ecosystems, not equations. Traditional models capture trends; the Axiom captures the regime shifts beneath them — the pressure changes that produce the trends in the first place.",
    situation: [
      "Risk desks rely on models trained for yesterday's market. By the time correlations break, the damage is already booked.",
      "Capital allocators ask: which variables actually moved this outcome? Standard attribution returns a forest of features and no answer."
    ],
    applications: [
      { num: "A", title: "Regime-aware risk", text: "Detect the macro pressure shift before the correlation matrix does. Recalibrate hedges in hours, not quarters." },
      { num: "B", title: "Causal attribution", text: "Move from ‘which features correlated’ to ‘which axes of pressure produced this'. Defensible to a regulator." },
      { num: "C", title: "Counter-fraud reasoning", text: "Surface coordination patterns invisible to graph-only methods. Network + temporal + intent in one lens." },
      { num: "D", title: "Underwriting depth", text: "Price tails that historical data alone never witnessed by reasoning from first principles of exposure." }
    ],
    metrics: [
      { value: ["3.4", { em: "×" }], label: "Earlier regime detection", note: "Versus rolling-window VaR on 2008, 2020, 2023 events" },
      { value: ["−42", { em: "%" }], label: "False-positive AML alerts", note: "Pilot with a top-20 North American bank, 18 months" },
      { value: ["100%", null], label: "Auditable derivations", note: "Every decision traceable to source pressure variables" }
    ]
  },
  {
    id: "healthcare",
    num: "02",
    name: "Healthcare & Pharma",
    short: "Providers · Payers · Biotech",
    title: ["Reason about the ", { em: "whole patient" }, ", not the latest scan."],
    lede: "Clinical decisions are made in a fog of partial signal — labs from Tuesday, imaging from last month, a complaint from yesterday. The Axiom integrates these as a single evolving picture rather than a list of inputs.",
    situation: [
      "Modern pharma drowns in candidate molecules and reads them through a sieve of trial cost. The early-signal noise floor is too high.",
      "Health systems hold richer longitudinal data than any single clinician can hold in mind, yet the loop back to bedside decisions remains thin."
    ],
    applications: [
      { num: "A", title: "Diagnostic synthesis", text: "Pull labs, imaging, history, and notes into one reasoning surface that names what the picture is, and what it isn't yet." },
      { num: "B", title: "Trial design", text: "Stratify cohorts on the variables that actually move outcomes, not the demographic boxes that proxy them." },
      { num: "C", title: "Real-world evidence", text: "Read post-market signal as a coherent system, not a stream of disconnected reports." },
      { num: "D", title: "Operational forecasting", text: "Census, staffing, and acuity as one model — anticipate ten days out, not ten hours." }
    ],
    metrics: [
      { value: ["−38", { em: "%" }], label: "Diagnostic time-to-decision", note: "Across three academic medical centers, complex cases" },
      { value: ["2.1", { em: "×" }], label: "Trial enrollment efficiency", note: "Phase II oncology, eligibility-aware matching" },
      { value: ["18", { em: "mo" }], label: "Faster signal to action", note: "Real-world evidence pipeline, pharmacovigilance" }
    ]
  },
  {
    id: "government",
    num: "03",
    name: "Government & Defense",
    short: "Intelligence · Logistics · Policy",
    title: ["Decisions that ", { em: "hold up" }, " in the room and on the record."],
    lede: "Sovereign decisions demand explanations, not outputs. The Axiom is the rare system where the path to a conclusion is the product — every inference traceable to the pressure variables that generated it.",
    situation: [
      "Coalition operations move on shared situational awareness; today, that awareness is reconstructed from incompatible models trained against incompatible doctrines.",
      "Policy decisions are made under second- and third-order effects no single domain expert can hold. Models that can't show their work are inadmissible."
    ],
    applications: [
      { num: "A", title: "All-source fusion", text: "OSINT, HUMINT, SIGINT, GEOINT integrated as a single field, not stacked dashboards." },
      { num: "B", title: "Wargame second-order", text: "Run policies forward through their consequences. Surface what the room hasn't thought of yet." },
      { num: "C", title: "Logistics under pressure", text: "Contested-environment routing that reasons about adversary intent, not just historical flow." },
      { num: "D", title: "Provenance & audit", text: "Every recommendation accompanied by its derivation. Cleared for FOIA, congressional inquiry, allied review." }
    ],
    metrics: [
      { value: ["IL-5", null], label: "Authorization in progress", note: "Targeting Q3 2026 for high-impact workloads" },
      { value: ["Zero", { em: "" }], label: "Opaque inferences", note: "Every output ships with its derivation tree" },
      { value: ["100%", null], label: "On-premise capable", note: "Air-gapped or sovereign-cloud deployments supported" }
    ]
  },
  {
    id: "manufacturing",
    num: "04",
    name: "Manufacturing & Industrial",
    short: "Process · Discrete · Heavy Industry",
    title: ["The plant as a ", { em: "living system" }, ", not a dashboard."],
    lede: "A line is a network of pressures: thermal, mechanical, chemical, human. Most analytics see only the symptoms — temperature spikes, yield drops. The Axiom reads the pressure shifts that produce them.",
    situation: [
      "Predictive maintenance models catch failures already in progress. The window between signal and failure keeps narrowing.",
      "Yield engineering walks a tightrope between throughput and quality. Lever-pulling depends on tribal knowledge that retires."
    ],
    applications: [
      { num: "A", title: "Causal yield", text: "Find the upstream variable that's actually driving downstream defect — not the one that correlates loudest." },
      { num: "B", title: "Pre-failure reasoning", text: "Detect the regime shift before the symptom. Buy days, not minutes." },
      { num: "C", title: "Supply-chain resilience", text: "Read the network's stress signature. Re-route before the disruption propagates." },
      { num: "D", title: "Knowledge continuity", text: "Encode the tribal intuition of senior operators into a reasoning system the next cohort can interrogate." }
    ],
    metrics: [
      { value: ["+11", { em: "%" }], label: "Yield in pilot lines", note: "Semiconductor fab, two-quarter pilot" },
      { value: ["−27", { em: "%" }], label: "Unplanned downtime", note: "Combined-cycle gas turbine fleet, 12 months" },
      { value: ["72h", { em: "" }], label: "Mean lead-time advantage", note: "From signal to actionable intervention" }
    ]
  },
  {
    id: "retail",
    num: "05",
    name: "Retail & Consumer Goods",
    short: "Brand · Merch · Supply",
    title: ["Understand why the ", { em: "trend" }, " happened, before the next one does."],
    lede: "Demand is not a curve to fit; it is a chorus of impulses, elements, and pressures. The Axiom reads the chorus, not just the result, so merchants can act in the window when action still changes the outcome.",
    situation: [
      "Forecasts arrive accurate and useless — confirming what already happened. Re-buys, allocations, and markdowns made too late to compound.",
      "Customer ‘insight' is a thousand dashboards and no synthesis. Product, marketing, and supply each act on a different signal."
    ],
    applications: [
      { num: "A", title: "Cohort-true forecasting", text: "Demand modeled at the granularity it actually behaves — neither single-store nor single-SKU, but their interaction." },
      { num: "B", title: "Allocation that learns", text: "Re-stock decisions that react to regime shifts, not just last week's velocity." },
      { num: "C", title: "Trend genesis", text: "Surface the underlying social, cultural, and economic pressures producing demand — not just demand itself." },
      { num: "D", title: "Margin-aware promotion", text: "Price and promote knowing the lift's full cost across cohorts, channels, and time." }
    ],
    metrics: [
      { value: ["−23", { em: "%" }], label: "Stockout days", note: "Apparel retailer, full-line rollout, FY25" },
      { value: ["+8.4", { em: "pp" }], label: "Full-price sell-through", note: "vs. prior-year matched assortment" },
      { value: ["6w", { em: "" }], label: "Earlier trend identification", note: "Versus internal analytics baseline" }
    ]
  },
  {
    id: "tech",
    num: "06",
    name: "Tech & SaaS",
    short: "Platforms · Infrastructure · Products",
    title: ["From metrics to the ", { em: "system" }, " producing them."],
    lede: "Every product team is buried in dashboards that report what happened. The Axiom is a reasoning layer on top of your data that explains why — and tells you the smallest move that would change the outcome.",
    situation: [
      "Growth, retention, and reliability are tracked in three different rooms with three different mental models. The org argues about which metric to optimize because no one can see the system whole.",
      "AI products built on conventional LLMs hallucinate confidently. Enterprise buyers ask: where did this come from? Standard stacks can't answer."
    ],
    applications: [
      { num: "A", title: "Causal product analytics", text: "Move from ‘these two cohorts diverged' to ‘this is the variable that produced the divergence'." },
      { num: "B", title: "Reasoning layer", text: "Embed the Axiom as the inference core of your product — auditable, deterministic where it matters." },
      { num: "C", title: "SRE intelligence", text: "Read incident telemetry as a coherent field. Find the root cause class, not just the symptom." },
      { num: "D", title: "Strategy synthesis", text: "Run roadmap candidates through second-order effects before the quarter, not after." }
    ],
    metrics: [
      { value: ["6.2", { em: "×" }], label: "Hypotheses tested per quarter", note: "Engineering org, 400 ppl, baseline vs. Axiom-integrated" },
      { value: ["−61", { em: "%" }], label: "Time to root-cause incidents", note: "Cross-service distributed systems" },
      { value: ["100%", null], label: "Traceable inferences", note: "Every model output carries its derivation" }
    ]
  },
  {
    id: "consulting",
    num: "07",
    name: "Consulting Firms",
    short: "Strategy · Transformation · Advisory",
    title: ["A reasoning ", { em: "spine" }, " for the work that justifies the fee."],
    lede: "The deliverable is not a deck — it is the synthesis behind it. Axiom-equipped teams compress weeks of triangulation into days, while raising the floor on the rigor any associate can produce.",
    situation: [
      "Engagement velocity is bounded by the slowest human in the loop. Most of that human's time is spent reconciling sources, not generating insight.",
      "Clients increasingly ask: show me the reasoning, not the conclusion. PowerPoint logic doesn't survive contact with a sharp CFO."
    ],
    applications: [
      { num: "A", title: "Synthesis at speed", text: "Reconcile a hundred interviews, a thousand documents, and three years of market data into one defensible narrative." },
      { num: "B", title: "Hypothesis stress-testing", text: "Run every recommendation through the second- and third-order effects partners used to model intuitively." },
      { num: "C", title: "IP that compounds", text: "Each engagement deepens the firm's reasoning asset, not just its case-study library." },
      { num: "D", title: "Defensible deliverables", text: "Every chart, every claim, every recommendation traceable to its source pressure variables." }
    ],
    metrics: [
      { value: ["2.8", { em: "×" }], label: "Engagement throughput", note: "Mid-size strategy firm, partnered pilot" },
      { value: ["−54", { em: "%" }], label: "Time per synthesis cycle", note: "From source materials to draft narrative" },
      { value: ["+38", { em: "%" }], label: "Net revenue per partner", note: "Trailing-twelve-month post-deployment" }
    ]
  },
  {
    id: "energy",
    num: "08",
    name: "Energy & Utilities",
    short: "Generation · Grid · Transition",
    title: ["Coordinate a ", { em: "grid" }, " that no one designed for what it's becoming."],
    lede: "The grid is becoming a different system every year — distributed generation, intermittent inputs, electrified loads. Traditional control reasons about a steady-state that no longer exists.",
    situation: [
      "Operators are asked to integrate renewables and EVs into systems whose physics models predate either. The control loop runs faster than the planning horizon.",
      "Capital plans for the energy transition span decades but rest on assumptions revisited yearly. Defending those assumptions to a regulator gets harder, not easier."
    ],
    applications: [
      { num: "A", title: "Grid-edge reasoning", text: "Coordinate millions of distributed assets as one system, not a flock of optimizers." },
      { num: "B", title: "Transition planning", text: "Stress-test decadal capital plans against the full space of policy, weather, and demand regimes." },
      { num: "C", title: "Wildfire & extreme-event posture", text: "Read the precursor state of the network, not just the current weather." },
      { num: "D", title: "Customer reasoning", text: "Programs that meet customers in the actual structure of their consumption, not a demographic proxy." }
    ],
    metrics: [
      { value: ["+14", { em: "%" }], label: "Renewables integration headroom", note: "Without additional firm capacity" },
      { value: ["−31", { em: "%" }], label: "Forced outages, peak season", note: "Distribution utility, three-summer comparison" },
      { value: ["3", { em: "decades" }], label: "Plan horizon, defensible", note: "Stress-tested against 1,200 regime paths" }
    ]
  },
  {
    id: "education",
    num: "09",
    name: "Education & Research",
    short: "R1 Universities · National Labs · Think Tanks",
    title: ["A research ", { em: "instrument" }, ", not another tool."],
    lede: "Universities and labs aren't asking for productivity gains. They're asking for an instrument — something that lets a sharp researcher see a class of questions that were previously unreachable.",
    situation: [
      "Cross-disciplinary work breaks on translation costs. The molecular biologist and the systems engineer can't share a model of the problem in any time-bounded way.",
      "The replication crisis is a reasoning crisis. Methods sections record what was done, not the chain of inferences that produced the claim."
    ],
    applications: [
      { num: "A", title: "Cross-domain synthesis", text: "Reason coherently across physics, biology, economics — without forcing the work into a single domain's vocabulary." },
      { num: "B", title: "Hypothesis generation", text: "Not search; generation. Surface the questions a domain hasn't asked yet but is positioned to answer." },
      { num: "C", title: "Methods-as-derivation", text: "Publish the reasoning chain alongside the paper. Replication becomes inspection." },
      { num: "D", title: "Teaching by structure", text: "Show students how a domain reasons, not just what it knows." }
    ],
    metrics: [
      { value: ["12", { em: "" }], label: "Cross-disciplinary collaborations", note: "Active institutional partners, 2025–26" },
      { value: ["Open", { em: "" }], label: "Core algorithm", note: "Inspectable derivation for every output" },
      { value: ["NAICS", { em: "" }], label: "541710 / R&D licensed", note: "Physical, engineering, and life sciences" }
    ]
  }
];

window.VERTICALS = VERTICALS;