diff --git a/messages/de.json b/messages/de.json
index c018aab94..b8f1a6c9f 100644
--- a/messages/de.json
+++ b/messages/de.json
@@ -1501,6 +1501,8 @@
     "heroRegenerating": "Wird generiert",
     "regenerateAnalysis": "Analyse neu starten",
     "regenerateSuccess": "Analyse wurde neu erstellt",
+    "warmAssessments": "Auswertungen vorbereiten",
+    "warmStarted": "Auswertungen werden im Hintergrund erstellt",
     "heroFallbackSubtitle": "Täglicher Blick auf deine Trends — direkt aus den Zahlen, die du protokollierst.",
     "heroGreetingMorning": "Guten Morgen",
     "heroGreetingAfternoon": "Guten Tag",
diff --git a/messages/en.json b/messages/en.json
index 9019923b6..ca23e5853 100644
--- a/messages/en.json
+++ b/messages/en.json
@@ -1501,6 +1501,8 @@
     "heroRegenerating": "Regenerating",
     "regenerateAnalysis": "Re-run analysis",
     "regenerateSuccess": "Analysis refreshed",
+    "warmAssessments": "Prepare assessments",
+    "warmStarted": "Assessments are being prepared in the background",
     "heroFallbackSubtitle": "A daily read of your trends, drawn straight from the numbers you've logged.",
     "heroGreetingMorning": "Good morning",
     "heroGreetingAfternoon": "Good afternoon",
diff --git a/messages/es.json b/messages/es.json
index 101db930e..b34fa799b 100644
--- a/messages/es.json
+++ b/messages/es.json
@@ -1501,6 +1501,8 @@
     "heroRegenerating": "Generando",
     "regenerateAnalysis": "Volver a iniciar el análisis",
     "regenerateSuccess": "Análisis regenerado",
+    "warmAssessments": "Preparar evaluaciones",
+    "warmStarted": "Las evaluaciones se están preparando en segundo plano",
     "heroFallbackSubtitle": "Una mirada diaria a tus tendencias — directamente de los datos que registras.",
     "heroGreetingMorning": "Buenos días",
     "heroGreetingAfternoon": "Buenas tardes",
diff --git a/messages/fr.json b/messages/fr.json
index d4b64818c..cf4ceb691 100644
--- a/messages/fr.json
+++ b/messages/fr.json
@@ -1501,6 +1501,8 @@
     "heroRegenerating": "Génération en cours",
     "regenerateAnalysis": "Relancer l’analyse",
     "regenerateSuccess": "Analyse regénérée",
+    "warmAssessments": "Préparer les évaluations",
+    "warmStarted": "Les évaluations sont en cours de préparation en arrière-plan",
     "heroFallbackSubtitle": "Un coup d’œil quotidien sur tes tendances — directement issu des chiffres que tu saisis.",
     "heroGreetingMorning": "Bonjour",
     "heroGreetingAfternoon": "Bonjour",
diff --git a/messages/it.json b/messages/it.json
index 72861adbc..175434875 100644
--- a/messages/it.json
+++ b/messages/it.json
@@ -1501,6 +1501,8 @@
     "heroRegenerating": "Generazione in corso",
     "regenerateAnalysis": "Riavvia l’analisi",
     "regenerateSuccess": "Analisi rigenerata",
+    "warmAssessments": "Prepara le valutazioni",
+    "warmStarted": "Le valutazioni vengono preparate in background",
     "heroFallbackSubtitle": "Uno sguardo quotidiano ai tuoi trend — direttamente dai numeri che registri.",
     "heroGreetingMorning": "Buongiorno",
     "heroGreetingAfternoon": "Buon pomeriggio",
diff --git a/messages/pl.json b/messages/pl.json
index 308ed202d..22b4b7bc5 100644
--- a/messages/pl.json
+++ b/messages/pl.json
@@ -1501,6 +1501,8 @@
     "heroRegenerating": "Generowanie",
     "regenerateAnalysis": "Uruchom analizę ponownie",
     "regenerateSuccess": "Analiza wygenerowana ponownie",
+    "warmAssessments": "Przygotuj analizy",
+    "warmStarted": "Analizy są przygotowywane w tle",
     "heroFallbackSubtitle": "Codzienne spojrzenie na twoje trendy — prosto z liczb, które rejestrujesz.",
     "heroGreetingMorning": "Dzień dobry",
     "heroGreetingAfternoon": "Dzień dobry",
diff --git a/src/app/insights/page.tsx b/src/app/insights/page.tsx
index 0153fc9d9..e3fd38cf6 100644
--- a/src/app/insights/page.tsx
+++ b/src/app/insights/page.tsx
@@ -1,9 +1,10 @@
 "use client";
 
+import { useEffect } from "react";
 import { useQuery } from "@tanstack/react-query";
 import dynamic from "next/dynamic";
 import Link from "next/link";
-import { TrendingUp } from "lucide-react";
+import { TrendingUp, Sparkles } from "lucide-react";
 
 import { useAuth } from "@/hooks/use-auth";
 import { queryKeys } from "@/lib/query-keys";
@@ -14,6 +15,7 @@ import { Button } from "@/components/ui/button";
 import { EmptyState } from "@/components/ui/empty-state";
 import { HeroStrip } from "@/components/insights/hero-strip";
 import { useInsightsAdvisorQuery } from "@/components/insights/use-insights-advisor";
+import { useInsightsWarm } from "@/components/insights/use-insights-warm";
 import { useCoachLaunch } from "@/lib/insights/coach-launch-context";
 import { useAnalyticsQuery } from "@/lib/queries/use-analytics-query";
 // v1.4.41 W-ORG — shared shape lives in `src/types/analytics.ts` as
@@ -150,6 +152,17 @@ export default function InsightsPage() {
   const analyticsQuery = useAnalyticsQuery();
   const analytics = analyticsQuery.data as AnalyticsData | undefined;
 
+  // v1.8.7.1 — on-demand full assessment warm. The auto-trigger fires once
+  // per session as soon as the overview confirms the user has data, so the
+  // status cards (and the iOS client, which reads the same routes) land on
+  // warm caches without the user waiting. The manual button below re-warms
+  // everything on demand.
+  const { warm, isWarming, autoWarmOnce } = useInsightsWarm();
+  const hasData = !!data && data.totalMeasurements > 0;
+  useEffect(() => {
+    autoWarmOnce(isAuthenticated && hasData);
+  }, [autoWarmOnce, isAuthenticated, hasData]);
+
   // Empty-state shortcut — only paint once the comprehensive query has
   // resolved AND reported zero measurements. While it's in-flight we
   // fall through to the streamed shell so the user gets the hero +
@@ -210,6 +223,18 @@ export default function InsightsPage() {
         healthScore={analytics?.healthScore ?? undefined}
       />
 
+      <div className="flex justify-end">
+        <Button
+          size="sm"
+          variant="outline"
+          onClick={warm}
+          disabled={isWarming}
+        >
+          <Sparkles className="size-4" />
+          {t("insights.warmAssessments")}
+        </Button>
+      </div>
+
       {flags.briefing && (
         <DailyBriefing
           briefing={briefingPayload}
diff --git a/src/components/insights/use-insights-warm.ts b/src/components/insights/use-insights-warm.ts
new file mode 100644
index 000000000..53934b35f
--- /dev/null
+++ b/src/components/insights/use-insights-warm.ts
@@ -0,0 +1,101 @@
+"use client";
+
+import { useCallback, useRef } from "react";
+import { useMutation } from "@tanstack/react-query";
+import { toast } from "sonner";
+
+import { queryKeys } from "@/lib/query-keys";
+import { useTranslations } from "@/lib/i18n/context";
+
+/**
+ * v1.8.7.1 — on-demand full assessment warm.
+ *
+ * POSTs `/api/insights/pregenerate`, which enqueues a forced warm of every
+ * AI assessment (comprehensive insight + the seven specialised status cards
+ * + every data-bearing generic metric assessment) for the calling user on
+ * the worker and returns immediately. The cards then fill via their existing
+ * stale-while-revalidate GETs — there is nothing to invalidate here.
+ *
+ * Two entry points share the one mutation:
+ *   - `warm()` — the manual "prepare assessments" button. Shows a toast.
+ *   - `autoWarmOnce(enabled)` — a fire-and-forget background warm fired ONCE
+ *     per browser session (sessionStorage-gated) when the Insights overview
+ *     mounts with data, so a returning user lands on warm caches without
+ *     tapping anything. The auto path is silent (no toast) and tolerates a
+ *     blocked/failed request — the server-side anti-spam bucket and the
+ *     nightly cron are the catch-net.
+ */
+
+const AUTO_WARM_SESSION_KEY = "healthlog:insights:auto-warm";
+
+async function postWarm(): Promise<void> {
+  // Same-origin relative-path fetch — exempt from the safe-fetch rule by
+  // construction. A 429 (warm already in progress) is an expected,
+  // harmless outcome, not an error.
+  const res = await fetch("/api/insights/pregenerate", {
+    method: "POST",
+    headers: { "Content-Type": "application/json" },
+    body: "{}",
+  });
+  if (!res.ok && res.status !== 429) {
+    throw new Error(`HTTP ${res.status}`);
+  }
+}
+
+export interface UseInsightsWarmResult {
+  /** Manual trigger — POSTs the warm and shows a toast. */
+  warm: () => void;
+  /** True while the warm POST is in flight. */
+  isWarming: boolean;
+  /**
+   * Fire the background warm once per session. Pass `enabled` (e.g.
+   * authenticated AND the user has data) — the warm only fires on the
+   * first call where `enabled` is true, and never again this session.
+   */
+  autoWarmOnce: (enabled: boolean) => void;
+}
+
+export function useInsightsWarm(): UseInsightsWarmResult {
+  const { t } = useTranslations();
+
+  const mutation = useMutation({
+    mutationKey: queryKeys.insightsPregenerate(),
+    mutationFn: postWarm,
+  });
+
+  const { mutate } = mutation;
+
+  const warm = useCallback(() => {
+    mutate(undefined, {
+      onSuccess: () => toast.success(t("insights.warmStarted")),
+    });
+  }, [mutate, t]);
+
+  // The auto-trigger must fire at most once per session even across the
+  // remounts a route navigation causes. A ref guards within the mount;
+  // sessionStorage guards across mounts / navigations.
+  const firedThisMount = useRef(false);
+  const autoWarmOnce = useCallback(
+    (enabled: boolean) => {
+      if (!enabled || firedThisMount.current) return;
+      firedThisMount.current = true;
+      try {
+        if (window.sessionStorage.getItem(AUTO_WARM_SESSION_KEY)) return;
+        window.sessionStorage.setItem(AUTO_WARM_SESSION_KEY, String(Date.now()));
+      } catch {
+        // sessionStorage can throw under strict-privacy modes — fall
+        // through and fire the warm anyway; the server bucket de-dupes.
+      }
+      // Silent background warm — no toast, errors swallowed by the
+      // mutation's default (no onError handler surfaces it).
+      mutate();
+    },
+    [mutate],
+  );
+
+  return {
+    warm,
+    isWarming: mutation.isPending,
+    autoWarmOnce,
+  };
+}
diff --git a/src/lib/query-keys.ts b/src/lib/query-keys.ts
index 0aeae6832..d9dd2c23b 100644
--- a/src/lib/query-keys.ts
+++ b/src/lib/query-keys.ts
@@ -113,6 +113,15 @@ export const queryKeys = {
    */
   insightsMetricStatus: (metric: string, locale: string) =>
     ["insights", "metric-status", metric, locale] as const,
+  /**
+   * v1.8.7.1 — mutationKey for the on-demand full-warm POST
+   * (`/api/insights/pregenerate`). Lives in the factory so the bare-array
+   * ESLint rule stays satisfied; the warm enqueues every assessment
+   * generation on the worker, so there is no read cache to invalidate
+   * here — the existing status queries pick up the warmed text via their
+   * own stale-while-revalidate.
+   */
+  insightsPregenerate: () => ["insights", "pregenerate"] as const,
 
   medications: () => ["medications"] as const,
   medicationDetail: (id: string) => ["medications", id] as const,