diff --git a/src/App.firstWorkflowNavigation.test.tsx b/src/App.firstWorkflowNavigation.test.tsx
index e73c04a..284e00b 100644
--- a/src/App.firstWorkflowNavigation.test.tsx
+++ b/src/App.firstWorkflowNavigation.test.tsx
@@ -25,7 +25,7 @@ vi.mock("./screens/Welcome", () => ({
   Welcome: ({ onShow, onSample }: { onShow: () => void; onSample: () => void }) => (
     <div>
       <button type="button" onClick={onShow}>
-        Show OpenAdapt your task
+        Show OpenAdapt your workflow
       </button>
       <button type="button" onClick={onSample}>
         See it work on a sample app
@@ -121,7 +121,7 @@ it("stays on the welcome screen when its durable stage cannot be saved", async (
 
   render(<App />);
   const start = await screen.findByRole("button", {
-    name: "Show OpenAdapt your task",
+    name: "Show OpenAdapt your workflow",
   });
   fireEvent.click(start);
 
@@ -129,10 +129,10 @@ it("stays on the welcome screen when its durable stage cannot be saved", async (
   expect(alert.textContent).toContain(
     "Desktop couldn't save your place in setup. Check the local engine, then try again.",
   );
-  expect(screen.getByRole("button", { name: "Show OpenAdapt your task" })).toBeTruthy();
+  expect(screen.getByRole("button", { name: "Show OpenAdapt your workflow" })).toBeTruthy();
   expect(screen.queryByText("First workflow recording")).toBeNull();
 
-  fireEvent.click(screen.getByRole("button", { name: "Show OpenAdapt your task" }));
+  fireEvent.click(screen.getByRole("button", { name: "Show OpenAdapt your workflow" }));
   await waitFor(() =>
     expect(screen.getByText("First workflow recording")).toBeTruthy(),
   );
@@ -259,7 +259,7 @@ it("keeps navigation locked while the supervised run is running", async () => {
 
   await waitFor(() =>
     expect(
-      (screen.getByRole("button", { name: "Tasks" }) as HTMLButtonElement)
+      (screen.getByRole("button", { name: "Workflows" }) as HTMLButtonElement)
         .disabled,
     ).toBe(true),
   );
@@ -303,7 +303,7 @@ it("keeps navigation locked from recording stop until Check it opens", async ()
       capture_id: "capture-1",
     });
   });
-  const tasks = screen.getByRole("button", { name: "Tasks" });
+  const tasks = screen.getByRole("button", { name: "Workflows" });
   expect((tasks as HTMLButtonElement).disabled).toBe(true);
 
   firstState = { ...firstState, stage: "review", workflow_id: "workflow-1" };
@@ -333,9 +333,9 @@ it("puts Settings and every expert tool behind the gear", async () => {
   });
 
   render(<App />);
-  expect(await screen.findByRole("heading", { name: "Tasks" })).toBeTruthy();
+  expect(await screen.findByRole("heading", { name: "Workflows" })).toBeTruthy();
   const nav = screen.getByRole("navigation", { name: "Desktop navigation" });
-  expect(nav.textContent).toBe("TasksNeeds youResults");
+  expect(nav.textContent).toBe("WorkflowsNeeds youResults");
 
   fireEvent.click(screen.getByRole("button", { name: "Settings and advanced tools" }));
   expect(screen.getByRole("menuitem", { name: /^Settings/ })).toBeTruthy();
@@ -371,12 +371,12 @@ it("leaves loading when authoring status is missing", async () => {
 
   render(<App />);
   expect(
-    await screen.findByRole("button", { name: "Show OpenAdapt your task" }),
+    await screen.findByRole("button", { name: "Show OpenAdapt your workflow" }),
   ).toBeTruthy();
   expect(screen.queryByText("Loading…")).toBeNull();
 });
 
-it("won't start another task while an uncertain first run is unchecked", async () => {
+it("won't start another workflow while an uncertain first run is unchecked", async () => {
   vi.mocked(engineTry).mockImplementation(async (command, _params, fallback) => {
     const boot = bootDefaults(command);
     if (boot !== undefined) return boot;
@@ -394,13 +394,13 @@ it("won't start another task while an uncertain first run is unchecked", async (
 
   render(<App />);
   expect(await screen.findByText("First supervised replay")).toBeTruthy();
-  fireEvent.click(screen.getByRole("button", { name: "Tasks" }));
+  fireEvent.click(screen.getByRole("button", { name: "Workflows" }));
   fireEvent.click(
-    await screen.findByRole("button", { name: "Show OpenAdapt a new task" }),
+    await screen.findByRole("button", { name: "Record a workflow" }),
   );
 
   expect((await screen.findByRole("alert")).textContent).toContain(
-    "Check the target of your last run before you record another task.",
+    "Check the target of your last run before you record another workflow.",
   );
   expect(await screen.findByText("First supervised replay")).toBeTruthy();
   expect(screen.queryByText("First workflow recording")).toBeNull();
@@ -427,7 +427,7 @@ it("says Local only, not synced, when nothing is backed up to Cloud", async () =
   bootTasks({ authenticated: true }, () => ({ state: "local", queued: 0 }));
 
   render(<App />);
-  expect(await screen.findByRole("heading", { name: "Tasks" })).toBeTruthy();
+  expect(await screen.findByRole("heading", { name: "Workflows" })).toBeTruthy();
   const status = screen.getByLabelText("Desktop status");
   await waitFor(() => expect(status.textContent).toContain("Local only"));
   expect(status.textContent).not.toMatch(/synced/i);
@@ -439,7 +439,7 @@ it("never shows a sync state the engine didn't report", async () => {
   bootTasks({ authenticated: true }, (fallback) => fallback);
 
   render(<App />);
-  expect(await screen.findByRole("heading", { name: "Tasks" })).toBeTruthy();
+  expect(await screen.findByRole("heading", { name: "Workflows" })).toBeTruthy();
   const status = screen.getByLabelText("Desktop status");
   expect(status.textContent).not.toMatch(/synced|local/i);
 });
diff --git a/src/App.tsx b/src/App.tsx
index 063d728..995e90d 100644
--- a/src/App.tsx
+++ b/src/App.tsx
@@ -94,7 +94,7 @@ type PairingState = {
 };
 
 const NAV: { route: "tasks" | "needs" | "results"; label: string }[] = [
-  { route: "tasks", label: "Tasks" },
+  { route: "tasks", label: "Workflows" },
   { route: "needs", label: "Needs you" },
   { route: "results", label: "Results" },
 ];
@@ -175,7 +175,7 @@ function DesktopBrand({ onOpen }: { onOpen?: () => void }) {
   }
   return (
     <button
-      aria-label="Open OpenAdapt Desktop tasks"
+      aria-label="Open OpenAdapt Desktop workflows"
       className="product-brand"
       onClick={onOpen}
       type="button"
@@ -193,7 +193,7 @@ function EngineOffline({ onRetry, busy }: { onRetry: () => void; busy: boolean }
     <div className="content simple">
       <PageHead
         title="Waiting for the local engine"
-        sub="Desktop can't reach the OpenAdapt engine on this computer, so it can't show your tasks yet. It tries again when the engine starts. If this doesn't clear, quit and reopen OpenAdapt."
+        sub="Desktop can't reach the OpenAdapt engine on this computer, so it can't show your workflows yet. It tries again when the engine starts. If this doesn't clear, quit and reopen OpenAdapt."
       />
       <div className="row">
         <Button variant="primary" disabled={busy} onClick={onRetry}>
@@ -695,7 +695,7 @@ export default function App() {
       const resumed = routeForFirstWorkflow(current);
       if (resumed) setRoute(resumed);
       setFirstWorkflowStageError(
-        "Check the target of your last run before you record another task.",
+        "Check the target of your last run before you record another workflow.",
       );
       return;
     }
@@ -797,7 +797,7 @@ export default function App() {
   const navigationLockTitle = recording
     ? "Stop and save this recording before you leave."
     : workflowCompiling
-      ? "Keep this screen open while OpenAdapt builds the task from your recording."
+      ? "Keep this screen open while OpenAdapt builds the workflow from your recording."
       : firstWorkflowRunning
         ? "Keep the run open until it stops."
         : firstRunPersistencePending
@@ -863,7 +863,7 @@ export default function App() {
                 </span>
               )}
               {recording ? (
-                <span className="recording" title="OpenAdapt is recording your task">
+                <span className="recording" title="OpenAdapt is recording your workflow">
                   <StatusDot tone="warn" />
                   <strong>Recording</strong>
                 </span>
diff --git a/src/lib/attention.ts b/src/lib/attention.ts
index 6c52c6e..722b427 100644
--- a/src/lib/attention.ts
+++ b/src/lib/attention.ts
@@ -226,7 +226,7 @@ export function attentionCopy(item: AttentionItem): AttentionCopy {
           ? `OpenAdapt couldn't find “${targetName(item)}” on the screen`
           : "OpenAdapt couldn't find the next thing to click",
         sentence:
-          "The app may look different from when you showed the task. It stopped instead of guessing where to click.",
+          "The app may look different from when you showed the workflow. It stopped instead of guessing where to click.",
         prevented: "It stopped before saving. Nothing was written.",
         preventedTone: "ok",
         choices: [...choices].sort((left, right) =>
diff --git a/src/lib/checkPlan.ts b/src/lib/checkPlan.ts
index a7bc793..62710ff 100644
--- a/src/lib/checkPlan.ts
+++ b/src/lib/checkPlan.ts
@@ -139,7 +139,7 @@ const CASE_COPY: Record<string, { label: string; explanation: string }> = {
       "A second copy of the save button appears on the page. OpenAdapt must stop instead of guessing.",
   },
   stale_identity: {
-    label: "Record changed mid-task",
+    label: "Record changed during the run",
     explanation:
       "The record on screen changes after OpenAdapt checked it. OpenAdapt must stop before saving.",
   },
@@ -189,7 +189,7 @@ function caseView(
  * program, a kind this version doesn't know) may run steps Check it can't see.
  */
 const SHOWN_NODE_KINDS = new Set(["action", "terminal", "branch", "loop", "business_decision"]);
-const HIDDEN_STEPS_CAPTION = "Runs a part of the task that isn't shown step by step here";
+const HIDDEN_STEPS_CAPTION = "Runs a part of the workflow that isn't shown step by step here";
 
 /** The control a step clicks, by its structure, before the OCR-based title. */
 function stepCaption(
@@ -255,7 +255,7 @@ export function checkPlan(response: QualificationProject): CheckPlan {
       canArm: Boolean(control?.identity.can_arm),
       identityText: params.length
         ? `Before saving, it checks that the record on screen matches the ${params.join(" and ")} you give it. If they differ, it stops and asks you.`
-        : "Before saving, it checks the record on screen against the details it saw when you showed the task. If they differ, it stops and asks you.",
+        : "Before saving, it checks the record on screen against the details it saw when you showed the workflow. If they differ, it stops and asks you.",
       checked: effects.length > 0 && tiers.length === effects.length,
       hasRecordChecks: effects.length > 0 || node.effects.length > 0,
       checkText: effects.length
diff --git a/src/lib/outcomes.ts b/src/lib/outcomes.ts
index 6ccdbbb..b94eefb 100644
--- a/src/lib/outcomes.ts
+++ b/src/lib/outcomes.ts
@@ -188,7 +188,7 @@ export function resultCopy(plain: Pick<PlainResult, "key" | "nothingWritten">):
         : {
             label: "Didn't finish",
             sentence:
-              "It couldn't complete the task, and it can't tell whether anything was saved.",
+              "It couldn't complete the workflow, and it can't tell whether anything was saved.",
             next: "Check the record before you try again.",
             tone: "crit",
           };
diff --git a/src/lib/parseItems.ts b/src/lib/parseItems.ts
index e163328..c960179 100644
--- a/src/lib/parseItems.ts
+++ b/src/lib/parseItems.ts
@@ -114,7 +114,7 @@ export function parseItems(
   if (inputs.length === 0) {
     return {
       ...empty,
-      error: "This task doesn't take any values, so there's nothing to paste.",
+      error: "This workflow doesn't take any values, so there's nothing to paste.",
     };
   }
 
diff --git a/src/lib/tasks.ts b/src/lib/tasks.ts
index f1eec69..c2a805e 100644
--- a/src/lib/tasks.ts
+++ b/src/lib/tasks.ts
@@ -134,7 +134,7 @@ export function taskStatus(
       icon: "half",
       next: "check",
       nextLabel: "Continue checking",
-      detail: "Finish checking this task before it runs on real items.",
+      detail: "Finish checking this workflow before it runs on real items.",
     };
   }
   return {
diff --git a/src/screens/Advanced.tsx b/src/screens/Advanced.tsx
index 67b460c..aae6fe7 100644
--- a/src/screens/Advanced.tsx
+++ b/src/screens/Advanced.tsx
@@ -35,8 +35,8 @@ export function Advanced({
 
   const tools = [
     {
-      title: "All task versions",
-      body: "Every version of every task with replay, governed run, run detail, and teach controls.",
+      title: "All workflow versions",
+      body: "Every version of every workflow with replay, governed run, run detail, and teach controls.",
       action: onLibrary,
       label: "Open all versions",
     },
@@ -70,20 +70,20 @@ export function Advanced({
     <div className="content simple">
       <PageHead
         title="Advanced"
-        sub="For automation engineers and IT. These tools work on the same tasks as the main screens."
+        sub="For automation engineers and IT. These tools work on the same workflows as the main screens."
       />
       <section className="advanced-tool">
         <div>
           <h2>Exact checks</h2>
           <p>
-            The full readiness test for one task: the program map, each step's
+            The full readiness test for one workflow: the program map, each step's
             risk, right-record and record-check settings, test cases, approval,
             locking, export, and deploy.
           </p>
         </div>
         {tasks.length > 0 ? (
           <div className="advanced-picker">
-            <Field label="Task" htmlFor={`${id}-task`}>
+            <Field label="Workflow" htmlFor={`${id}-task`}>
               <select
                 id={`${id}-task`}
                 className="input"
@@ -102,7 +102,7 @@ export function Advanced({
             </Button>
           </div>
         ) : (
-          <p className="simple-sub">Show OpenAdapt a task first.</p>
+          <p className="simple-sub">Record a workflow first.</p>
         )}
       </section>
       {tools.map((tool) => (
diff --git a/src/screens/CheckIt.test.tsx b/src/screens/CheckIt.test.tsx
index 63da06e..3ce0ced 100644
--- a/src/screens/CheckIt.test.tsx
+++ b/src/screens/CheckIt.test.tsx
@@ -119,7 +119,7 @@ function engine(initial: ReturnType<typeof project>) {
   });
 }
 
-it("says what the task changes and how it checks, in plain words", async () => {
+it("says what the workflow changes and how it checks, in plain words", async () => {
   engine(project());
   render(<CheckIt workflowId="wf-1" onExactChecks={() => {}} onRun={() => {}} />);
 
@@ -180,12 +180,12 @@ function unprovenProgram() {
   return base;
 }
 
-it("never calls a task read-only while a step's effect is unknown", async () => {
+it("never calls a workflow read-only while a step's effect is unknown", async () => {
   engine(unprovenProgram());
   render(<CheckIt workflowId="wf-1" onExactChecks={() => {}} onRun={() => {}} />);
 
-  expect(await screen.findByText(/can't tell yet whether this task saves anything/)).toBeTruthy();
-  expect(screen.queryByText(/This task only reads/)).toBeNull();
+  expect(await screen.findByText(/can't tell yet whether this workflow saves anything/)).toBeTruthy();
+  expect(screen.queryByText(/This workflow only reads/)).toBeNull();
   expect(screen.getByText(/2 steps might change something/)).toBeTruthy();
   expect(screen.getByText("Type the insurance no")).toBeTruthy();
   expect(screen.getByText("Click “日”")).toBeTruthy();
@@ -198,13 +198,13 @@ it("never calls a task read-only while a step's effect is unknown", async () =>
   expect((run as HTMLButtonElement).disabled).toBe(true);
 });
 
-it("says a task only reads when every step is proven read-only", async () => {
+it("says a workflow only reads when every step is proven read-only", async () => {
   const readOnly = unprovenProgram();
   readOnly.graph.nodes = [readOnly.graph.nodes[1]];
   engine(readOnly);
   render(<CheckIt workflowId="wf-1" onExactChecks={() => {}} onRun={() => {}} />);
 
-  expect(await screen.findByText(/This task only reads/)).toBeTruthy();
+  expect(await screen.findByText(/This workflow only reads/)).toBeTruthy();
 });
 
 it("lists a save the program declares without an effect and asks how to check it", async () => {
@@ -226,7 +226,7 @@ it("lists a save the program declares without an effect and asks how to check it
   expect(await screen.findByText("Click the spot you clicked when you showed it")).toBeTruthy();
   expect(screen.getByText("OpenAdapt doesn't know yet what this step saves.")).toBeTruthy();
   expect(screen.getByText("OpenAdapt doesn't know yet how to confirm this save.")).toBeTruthy();
-  expect(screen.queryByText(/This task only reads/)).toBeNull();
+  expect(screen.queryByText(/This workflow only reads/)).toBeNull();
   expect(screen.getByText(/can't tell yet whether 1 other step changes anything/)).toBeTruthy();
   fireEvent.click(screen.getByRole("button", { name: "These look right" }));
   await waitFor(() =>
@@ -400,7 +400,7 @@ it("runs one drill at a time in the signed-in browser", async () => {
   render(<CheckIt workflowId="wf-1" onExactChecks={() => {}} onRun={() => {}} />);
 
   expect(await screen.findByText(/in your signed-in browser/)).toBeTruthy();
-  expect(screen.getByText(/Open the screen where the task starts before you run it/)).toBeTruthy();
+  expect(screen.getByText(/Open the screen where the workflow starts before you run it/)).toBeTruthy();
   fireEvent.click(screen.getByRole("button", { name: "Run the next drill" }));
 
   expect(await screen.findByText(/then run the next drill/)).toBeTruthy();
@@ -466,7 +466,7 @@ it("describes each drill as the change Flow makes before the save", async () =>
   for (const label of [
     "Wrong record",
     "Two possible matches",
-    "Record changed mid-task",
+    "Record changed during the run",
     "No record check",
     "Weak record check",
   ]) {
diff --git a/src/screens/CheckIt.tsx b/src/screens/CheckIt.tsx
index d1f0a05..9c42c14 100644
--- a/src/screens/CheckIt.tsx
+++ b/src/screens/CheckIt.tsx
@@ -653,7 +653,7 @@ export function CheckIt({
         setProject(latest);
         setDrillNote(
           batch.length < pending.length
-            ? "That drill ended the way its description says. Open the screen where the task starts, then run the next drill."
+            ? "That drill ended the way its description says. Open the screen where the workflow starts, then run the next drill."
             : "Every drill ended the way its description says.",
         );
       }
@@ -675,7 +675,7 @@ export function CheckIt({
         if (!certified.ok || !(certified as QualificationProject).certification_current) {
           if (certified.ok) setProject(certified as QualificationProject);
           fail(
-            "OpenAdapt can't approve this task yet. Open the exact checks to see what's still open.",
+            "OpenAdapt can't approve this workflow yet. Open the exact checks to see what's still open.",
             String(certified.error || ""),
           );
           return;
@@ -689,7 +689,7 @@ export function CheckIt({
           { workflow_id: workflowId },
         );
         if (!sealed.ok || !sealed.workflow_id) {
-          fail("OpenAdapt approved the task but couldn't lock this version.", String(sealed.error || ""));
+          fail("OpenAdapt approved the workflow but couldn't lock this version.", String(sealed.error || ""));
           return;
         }
         setApprovedId(sealed.workflow_id);
@@ -698,7 +698,7 @@ export function CheckIt({
       }
       setApprovedId(workflowId);
     } catch (error) {
-      fail("OpenAdapt couldn't approve this task.", String(error));
+      fail("OpenAdapt couldn't approve this workflow.", String(error));
     } finally {
       setBusy(null);
     }
@@ -750,7 +750,7 @@ export function CheckIt({
       {attachedBrowser(target) && (
         <p className="simple-sub">
           It starts from the screen your signed-in browser shows. Open the
-          screen where the task starts before you run it.
+          screen where the workflow starts before you run it.
         </p>
       )}
       <details className="technical">
@@ -822,7 +822,7 @@ export function CheckIt({
       />
 
       {loadError && (
-        <Callout tone="warn" title="Desktop couldn't read this task's checks">
+        <Callout tone="warn" title="Desktop couldn't read this workflow's checks">
           {loadError}
           <div className="callout-actions">
             <Button size="sm" onClick={() => void load()}>
@@ -852,13 +852,13 @@ export function CheckIt({
           <CheckSection index={1} title="What it changes" state={foundState(plan.changes)}>
             {plan.writes.length === 0 && plan.unproven.length === 0 ? (
               <p>
-                This task only reads. None of its steps saves anything, so there's
+                This workflow only reads. None of its steps saves anything, so there's
                 no record to check.
               </p>
             ) : plan.writes.length === 0 ? (
               <p>
-                OpenAdapt can't tell yet whether this task saves anything, so it
-                won't treat it as a task that only reads.
+                OpenAdapt can't tell yet whether this workflow saves anything, so it
+                won't treat it as a workflow that only reads.
               </p>
             ) : (
               <>
@@ -886,7 +886,7 @@ export function CheckIt({
                   {plan.writes.length === 0
                     ? `${plan.unproven.length} step${plan.unproven.length === 1 ? "" : "s"} might change something.`
                     : `OpenAdapt can't tell yet whether ${plan.unproven.length} other step${plan.unproven.length === 1 ? "" : "s"} change${plan.unproven.length === 1 ? "s" : ""} anything.`}{" "}
-                  Before you approve the task, say what each one does in Advanced:
+                  Before you approve the workflow, say what each one does in Advanced:
                   exact checks.
                 </p>
                 <details className="technical">
@@ -925,7 +925,7 @@ export function CheckIt({
                 {!step.identityArmed && !step.canArm && (
                   <p className="question">
                     OpenAdapt didn't see the record's name or ID on the screen
-                    where it saves. Show the task again with the record's name
+                    where it saves. Show the workflow again with the record's name
                     or ID visible before you save, or set this up in the exact
                     checks.
                   </p>
@@ -1048,7 +1048,7 @@ export function CheckIt({
             state={plan.writes.length === 0 && !plan.practice ? "todo" : plan.practiceState}
           >
             <p>
-              OpenAdapt does the task once, start to finish, on a test record you
+              OpenAdapt does the workflow once, start to finish, on a test record you
               choose, with every check on. Before it saves, it shows you the
               filled form and waits for you to approve. Then it saves for real
               in the app, so use a test record. Keep the app where you can see
@@ -1069,12 +1069,12 @@ export function CheckIt({
               <p>
                 {plan.drillsState === "not_needed"
                   ? "No drills are needed, because no step saves anything."
-                  : "OpenAdapt sets up five drills from your practice item when you run it. Each one makes the task go wrong on purpose right before the save, and must stop before saving."}
+                  : "OpenAdapt sets up five drills from your practice item when you run it. Each one makes the workflow go wrong on purpose right before the save, and must stop before saving."}
               </p>
             ) : (
               <>
                 <p>
-                  Each drill runs the task with your test values and changes one
+                  Each drill runs the workflow with your test values and changes one
                   thing on purpose right before the save. A drill passes only
                   when OpenAdapt's own check catches the change and stops before
                   saving. Nothing is saved, and no drill asks you to approve.
diff --git a/src/screens/Login.tsx b/src/screens/Login.tsx
index 4f396d9..109f38f 100644
--- a/src/screens/Login.tsx
+++ b/src/screens/Login.tsx
@@ -61,7 +61,7 @@ export function Login({
       <Card className="auth-card">
         <CardHead
           title="Connect to OpenAdapt Cloud"
-          sub="Sign in to share tasks and run history with your team. You can keep using OpenAdapt on this computer without an account."
+          sub="Sign in to share workflows and run history with your team. You can keep using OpenAdapt on this computer without an account."
         />
 
         <Button
diff --git a/src/screens/Onboarding.test.tsx b/src/screens/Onboarding.test.tsx
index 87c5897..0a6a445 100644
--- a/src/screens/Onboarding.test.tsx
+++ b/src/screens/Onboarding.test.tsx
@@ -43,7 +43,7 @@ it("doesn't call permissions missing when the engine can't check them", async ()
   expect(screen.queryByText("needed")).toBeNull();
   expect(screen.queryByText(/Grant the permissions/)).toBeNull();
   expect(screen.getByText(/can't reach the local engine/)).toBeTruthy();
-  const start = screen.getByRole("button", { name: "Choose an app and task" });
+  const start = screen.getByRole("button", { name: "Choose an app and workflow" });
   expect((start as HTMLButtonElement).disabled).toBe(true);
 });
 
diff --git a/src/screens/Onboarding.tsx b/src/screens/Onboarding.tsx
index 7873596..6bfd0c9 100644
--- a/src/screens/Onboarding.tsx
+++ b/src/screens/Onboarding.tsx
@@ -105,13 +105,13 @@ export function Onboarding({ onStart }: { onStart?: () => void }) {
         <h1>Recording setup</h1>
         <p>
           Recording a desktop app, a remote desktop, or Citrix needs these
-          permissions and the local video engine. Browser tasks and the sample
+          permissions and the local video engine. Browser workflows and the sample
           app don't need either.
         </p>
         <div className="row" style={{ justifyContent: "center" }}>
           {onStart && (
             <Button variant="primary" disabled={!ready} onClick={onStart}>
-              Choose an app and task
+              Choose an app and workflow
             </Button>
           )}
           <Button variant="ghost" onClick={refresh}>
diff --git a/src/screens/RecordReview.test.tsx b/src/screens/RecordReview.test.tsx
index c8833d6..8a8330c 100644
--- a/src/screens/RecordReview.test.tsx
+++ b/src/screens/RecordReview.test.tsx
@@ -79,7 +79,7 @@ it("shows automatic compile progress and retries a retained recording", async ()
   act(() => {
     eventMocks.handlers.get(EVT.RECORDING_STOPPED)?.({ capture_id: "cap-1" });
   });
-  expect(await screen.findByText("Building the task from your recording")).toBeTruthy();
+  expect(await screen.findByText("Building the workflow from your recording")).toBeTruthy();
   expect(screen.getByText(/recording stays as it was/i)).toBeTruthy();
 
   act(() => {
@@ -90,7 +90,7 @@ it("shows automatic compile progress and retries a retained recording", async ()
       recording_retained: true,
     });
   });
-  expect(await screen.findByText("OpenAdapt couldn't build the task")).toBeTruthy();
+  expect(await screen.findByText("OpenAdapt couldn't build the workflow")).toBeTruthy();
   expect(screen.getByText(/raw recording was retained/i)).toBeTruthy();
 
   vi.mocked(engineInvoke).mockResolvedValue({
@@ -164,7 +164,7 @@ it("opens one workflow when the event and command return report the same compile
   });
 });
 
-it("requires a real target and a named task for the first workflow", async () => {
+it("requires a real target and a named workflow for the first workflow", async () => {
   vi.mocked(engineTry).mockImplementation(async (command) => {
     if (command === CMD.GET_CAPABILITIES) return capabilityReport();
     return {
@@ -183,12 +183,12 @@ it("requires a real target and a named task for the first workflow", async () =>
   render(<RecordReview firstWorkflow onCompiled={() => {}} />);
 
   const recordButton = screen.getByRole("button", {
-    name: "Record this task",
+    name: "Record this workflow",
   }) as HTMLButtonElement;
   expect(recordButton.disabled).toBe(true);
-  expect(screen.getByText("Describe the task you want to record.")).toBeTruthy();
+  expect(screen.getByText("Describe the workflow you want to record.")).toBeTruthy();
 
-  fireEvent.change(screen.getByLabelText("What does this task do?"), {
+  fireEvent.change(screen.getByLabelText("What does this workflow do?"), {
     target: { value: "Save one test value" },
   });
   expect(
@@ -230,7 +230,7 @@ it("keeps an invalid or unavailable target out of recording and rechecks it", as
   });
 
   render(<RecordReview firstWorkflow onCompiled={() => {}} />);
-  fireEvent.change(screen.getByLabelText("What does this task do?"), {
+  fireEvent.change(screen.getByLabelText("What does this workflow do?"), {
     target: { value: "Read one test record" },
   });
   fireEvent.change(screen.getByLabelText("Page URL"), {
@@ -245,14 +245,14 @@ it("keeps an invalid or unavailable target out of recording and rechecks it", as
     (await screen.findAllByText(/Browser capture permission is required/)).length,
   ).toBeGreaterThan(0);
   expect(
-    (screen.getByRole("button", { name: "Record this task" }) as HTMLButtonElement)
+    (screen.getByRole("button", { name: "Record this workflow" }) as HTMLButtonElement)
       .disabled,
   ).toBe(true);
   fireEvent.click(screen.getByRole("button", { name: "Check again" }));
   await waitFor(() =>
     expect(
       (screen.getByRole("button", {
-        name: "Record this task",
+        name: "Record this workflow",
       }) as HTMLButtonElement).disabled,
     ).toBe(false),
   );
@@ -277,7 +277,7 @@ it("shows a recording failure and lets the user retry", async () => {
   });
 
   render(<RecordReview firstWorkflow onCompiled={() => {}} />);
-  fireEvent.change(screen.getByLabelText("What does this task do?"), {
+  fireEvent.change(screen.getByLabelText("What does this workflow do?"), {
     target: { value: "Read one test record" },
   });
   fireEvent.change(screen.getByLabelText("Page URL"), {
@@ -286,7 +286,7 @@ it("shows a recording failure and lets the user retry", async () => {
   await waitFor(() =>
     expect(
       (screen.getByRole("button", {
-        name: "Record this task",
+        name: "Record this workflow",
       }) as HTMLButtonElement).disabled,
     ).toBe(false),
   );
@@ -323,7 +323,7 @@ it("never tells a first-time user to leave the page URL blank", async () => {
   });
 
   render(<RecordReview firstWorkflow onCompiled={() => {}} />);
-  fireEvent.change(screen.getByLabelText("What does this task do?"), {
+  fireEvent.change(screen.getByLabelText("What does this workflow do?"), {
     target: { value: "Save one test value" },
   });
 
@@ -334,7 +334,7 @@ it("never tells a first-time user to leave the page URL blank", async () => {
     screen.getByText("Enter the page URL for the app you want to record."),
   ).toBeTruthy();
   const recordButton = screen.getByRole("button", {
-    name: "Record this task",
+    name: "Record this workflow",
   }) as HTMLButtonElement;
   expect(recordButton.disabled).toBe(true);
 });
diff --git a/src/screens/RecordReview.tsx b/src/screens/RecordReview.tsx
index ebfc0fd..8d53c5d 100644
--- a/src/screens/RecordReview.tsx
+++ b/src/screens/RecordReview.tsx
@@ -357,16 +357,16 @@ export function RecordReview({
         <div className="titles">
           <h1>
             {firstWorkflow
-              ? "Show OpenAdapt a task"
+              ? "Show OpenAdapt a workflow"
               : "Record with all options"}
           </h1>
         </div>
       </div>
 
       {firstWorkflow && (
-        <Callout title="Pick a task you can check yourself">
-          Do the task on a test record, the way you normally would. A
-          good task takes a minute or two and ends with a save you can see.
+        <Callout title="Pick a workflow you can check yourself">
+          Do the workflow on a test record, the way you normally would. A
+          good workflow takes a minute or two and ends with a save you can see.
           OpenAdapt doesn't send this recording anywhere unless you choose to
           share or upload it.
         </Callout>
@@ -376,13 +376,13 @@ export function RecordReview({
         <CardHead
           title={
             firstWorkflow
-              ? "Which app, and what's the task?"
+              ? "Which app, and what's the workflow?"
               : "What are you recording?"
           }
           sub={
             firstWorkflow
-              ? "Open the app first. Choose its type here, then say in a few words what the task does."
-              : "OpenAdapt will run this task in the same place you record it."
+              ? "Open the app first. Choose its type here, then say in a few words what the workflow does."
+              : "OpenAdapt will run this workflow in the same place you record it."
           }
         />
         <ExecutionTargetForm
@@ -403,10 +403,10 @@ export function RecordReview({
           </Button>
         )}
         <Field
-          label={firstWorkflow ? "What does this task do?" : "Task description"}
+          label={firstWorkflow ? "What does this workflow do?" : "Workflow description"}
           hint={
             firstWorkflow
-              ? "For example: Add a claim status note. This name shows on the task's card."
+              ? "For example: Add a claim status note. This name shows on the workflow's card."
               : "Optional. Saved with the recording. OpenAdapt doesn't send it anywhere unless you choose to upload the recording."
           }
           htmlFor={`${fieldPrefix}-record-task`}
@@ -427,9 +427,9 @@ export function RecordReview({
           />
         </Field>
         {firstWorkflow && (taskMissing || targetIssue) && (
-          <Callout tone="info" title="Finish the task setup">
+          <Callout tone="info" title="Finish the workflow setup">
             {taskMissing
-              ? "Describe the task you want to record."
+              ? "Describe the workflow you want to record."
               : targetIssue}
           </Callout>
         )}
@@ -449,7 +449,7 @@ export function RecordReview({
         )}
         <p className="page-sub">
           OpenAdapt keeps what you enter here out of its logs, and doesn't
-          send it anywhere unless you choose to deploy the task.
+          send it anywhere unless you choose to deploy the workflow.
         </p>
         {target.backend === "web" && runtime && runtime.state !== "ready" && (
           <Callout
@@ -492,7 +492,7 @@ export function RecordReview({
               disabled={busy || (firstWorkflow && !firstWorkflowReady)}
               onClick={start}
             >
-              {firstWorkflow ? "Record this task" : "Start recording"}
+              {firstWorkflow ? "Record this workflow" : "Start recording"}
             </Button>
           ) : (
             <>
@@ -539,16 +539,16 @@ export function RecordReview({
           <CardHead
             title={
               phase === "compiling"
-                ? "Building the task from your recording"
+                ? "Building the workflow from your recording"
                 : phase === "failed"
-                  ? "OpenAdapt couldn't build the task"
+                  ? "OpenAdapt couldn't build the workflow"
                   : "Your recording is saved"
             }
             sub={`capture ${lastCapture}`}
           />
           <div role="status" aria-live="polite">
             {phase === "compiling" && (
-              <Callout tone="info" title="Building the task">
+              <Callout tone="info" title="Building the workflow">
                 OpenAdapt is turning your recording into steps it can repeat
                 exactly. Your recording stays as it was.
               </Callout>
@@ -576,7 +576,7 @@ export function RecordReview({
                 ? "Building…"
                 : phase === "failed"
                   ? "Try building again"
-                  : "Build the task"}
+                  : "Build the workflow"}
             </Button>
             <Button
               variant="ghost"
diff --git a/src/screens/Results.tsx b/src/screens/Results.tsx
index 60994e1..2cb630b 100644
--- a/src/screens/Results.tsx
+++ b/src/screens/Results.tsx
@@ -129,7 +129,7 @@ export function Results({ onNeedsYou }: { onNeedsYou: () => void }) {
             No results yet
           </Status>
           <p>
-            Run a checked task on a list of items. Each item shows up here, and
+            Run a checked workflow on a list of items. Each item shows up here, and
             only items OpenAdapt read back from the record count as done.
           </p>
         </div>
@@ -197,12 +197,12 @@ export function Results({ onNeedsYou }: { onNeedsYou: () => void }) {
       {switches.length > 0 && <SwitchSection switches={switches} />}
 
       <section className="results-section" aria-labelledby="by-task">
-        <h2 id="by-task">By task</h2>
+        <h2 id="by-task">By workflow</h2>
         <div className="table-scroll">
           <table className="results-table">
             <thead>
               <tr>
-                <th>Task</th>
+                <th>Workflow</th>
                 <th className="num">Done and checked</th>
                 <th className="num">Needed a person</th>
                 <th className="num">Prevented</th>
@@ -264,7 +264,7 @@ export function Results({ onNeedsYou }: { onNeedsYou: () => void }) {
         <p className="fine-print">
           The time saved is an estimate, not a measurement. It multiplies the
           items OpenAdapt finished and checked by the minutes each one takes by
-          hand. The default is how long your recording of the task took; type
+          hand. The default is how long your recording of the workflow took; type
           your team's own number if it's different. Time spent answering Needs
           you isn't subtracted.
         </p>
@@ -345,7 +345,7 @@ function SwitchSection({ switches }: { switches: SwitchSummary[] }) {
           const status = switchStatus(entry.state);
           return (
             <li key={entry.workflow_id}>
-              <span className="switch-summary-task">{entry.task_name || "A task"}</span>
+              <span className="switch-summary-task">{entry.task_name || "A workflow"}</span>
               <Status tone={status.tone} icon={status.icon}>
                 {status.word}
               </Status>
diff --git a/src/screens/RunIt.test.tsx b/src/screens/RunIt.test.tsx
index da793b1..425db69 100644
--- a/src/screens/RunIt.test.tsx
+++ b/src/screens/RunIt.test.tsx
@@ -108,13 +108,13 @@ it("stops the list when a save may have gone through", async () => {
   expect(screen.getAllByText("Check the record").length).toBeGreaterThan(0);
 });
 
-it("sends an unchecked task to Check it first", async () => {
+it("sends an unchecked workflow to Check it first", async () => {
   engine("checking");
   const onCheck = vi.fn();
   render(<RunIt workflowId="wf-1" onCheck={onCheck} onNeedsYou={() => {}} onResults={() => {}} />);
 
   await paste("member_id,note\nM-1,Reviewed");
-  expect(await screen.findByText("Check this task first")).toBeTruthy();
+  expect(await screen.findByText("Check this workflow first")).toBeTruthy();
   expect((screen.getByRole("button", { name: "Run 1 item" }) as HTMLButtonElement).disabled).toBe(true);
   fireEvent.click(screen.getByRole("button", { name: "Check it" }));
   await waitFor(() => expect(onCheck).toHaveBeenCalledWith("wf-1"));
diff --git a/src/screens/RunIt.tsx b/src/screens/RunIt.tsx
index 7f50b27..b0d34e7 100644
--- a/src/screens/RunIt.tsx
+++ b/src/screens/RunIt.tsx
@@ -114,7 +114,7 @@ export function RunIt({
       setMode(approval && approval.ok ? approval : null);
       setWorkflow((Array.isArray(list) ? list : []).find((item) => item.id === workflowId) ?? null);
       if (!details.ok) {
-        setLoadError(details.error || "Desktop couldn't read this task's inputs.");
+        setLoadError(details.error || "Desktop couldn't read this workflow's inputs.");
         setInputs([]);
         return;
       }
@@ -322,13 +322,13 @@ export function RunIt({
       />
 
       {loadError && (
-        <Callout tone="warn" title="Desktop couldn't read this task">
+        <Callout tone="warn" title="Desktop couldn't read this workflow">
           {loadError}
         </Callout>
       )}
       {notReady && (
-        <Callout tone="info" title="Check this task first">
-          OpenAdapt runs a task on real items only after it's checked and
+        <Callout tone="info" title="Check this workflow first">
+          OpenAdapt runs a workflow on real items only after it's checked and
           approved.
           <div className="callout-actions">
             <Button size="sm" variant="primary" onClick={() => onCheck(workflowId)}>
@@ -369,7 +369,7 @@ export function RunIt({
         {source === "paste" && (
           <Field
             label="Paste rows from a spreadsheet or a list"
-            hint="Put the column names in the first row. For a task with one value, one item per line is enough."
+            hint="Put the column names in the first row. For a workflow with one value, one item per line is enough."
             htmlFor={`${idPrefix}-paste`}
           >
             <textarea
diff --git a/src/screens/Runner.tsx b/src/screens/Runner.tsx
index 457dbc1..728023c 100644
--- a/src/screens/Runner.tsx
+++ b/src/screens/Runner.tsx
@@ -101,9 +101,9 @@ export function Runner() {
       </div>
 
       <Callout tone="info" title="How this works">
-        When you turn this on, OpenAdapt Cloud can send approved tasks here.
+        When you turn this on, OpenAdapt Cloud can send approved workflows here.
         This computer reaches out to Cloud, and Cloud can't reach in. Each
-        task's approval is checked here before it runs, and only results with
+        workflow's approval is checked here before it runs, and only results with
         no PHI go back.
       </Callout>
 
diff --git a/src/screens/Sample.tsx b/src/screens/Sample.tsx
index cecda08..f8c0f50 100644
--- a/src/screens/Sample.tsx
+++ b/src/screens/Sample.tsx
@@ -107,7 +107,7 @@ export function Sample({
         if (payload?.state === "browser") {
           setDetail(payload.detail || "Preparing the browser…");
         } else if (payload?.state === "running") {
-          setDetail("Running the sample task twice…");
+          setDetail("Running the sample workflow twice…");
         }
       },
     );
@@ -186,16 +186,16 @@ export function Sample({
           </div>
           <p className="provenance-line">{PROVENANCE}</p>
           <section className="sample-next">
-            <h2>Now try your own task</h2>
+            <h2>Now try your own workflow</h2>
             <p>
-              Show OpenAdapt a task in the app where you already do it.
+              Show OpenAdapt a workflow in the app where you already do it.
               Use a test record.
             </p>
             <div className="row">
               <Button variant="primary" onClick={onShow}>
-                Show OpenAdapt your task
+                Show OpenAdapt your workflow
               </Button>
-              <Button onClick={onTasks}>Back to Tasks</Button>
+              <Button onClick={onTasks}>Back to Workflows</Button>
             </div>
           </section>
         </>
diff --git a/src/screens/Settings.tsx b/src/screens/Settings.tsx
index 1924de5..adc8484 100644
--- a/src/screens/Settings.tsx
+++ b/src/screens/Settings.tsx
@@ -202,7 +202,7 @@ export function Settings({
       <Card>
         <CardHead
           title="Where your data stays"
-          sub="Choose where recordings and tasks live, and whether they can leave this computer."
+          sub="Choose where recordings and workflows live, and whether they can leave this computer."
         />
         <Field label="Data location">
           <SegControl<DeploymentLane>
@@ -216,7 +216,7 @@ export function Settings({
         </Field>
         {cfg.deployment_lane === "byoc" && (
           <Callout tone="info" title="Patient data stays in your environment">
-            OpenAdapt doesn't upload recordings, tasks, or fixes. When a stopped
+            OpenAdapt doesn't upload recordings, workflows, or fixes. When a stopped
             run is reported, only a short note about it, with no PHI, goes to
             your dashboard. Backup, sync, and security software on this
             computer can still copy its files.
diff --git a/src/screens/Tasks.tsx b/src/screens/Tasks.tsx
index 9b41896..8447d3f 100644
--- a/src/screens/Tasks.tsx
+++ b/src/screens/Tasks.tsx
@@ -203,12 +203,12 @@ export function Tasks({
   return (
     <div className="content simple">
       <PageHead
-        title="Tasks"
-        sub="What OpenAdapt does for you, and what each task needs next."
+        title="Workflows"
+        sub="What OpenAdapt does for you, and what each workflow needs next."
         action={
           tasks.length > 0 ? (
             <Button variant="primary" onClick={actions.onShow}>
-              Show OpenAdapt a new task
+              Record a workflow
             </Button>
           ) : undefined
         }
@@ -218,16 +218,16 @@ export function Tasks({
         <p className="simple-sub">Loading…</p>
       ) : tasks.length === 0 ? (
         <section className="first-task" aria-label="Get started">
-          <h2>Hand off your first task</h2>
+          <h2>Hand off your first workflow</h2>
           <p>
-            Show OpenAdapt a task in the app where you already do it.
+            Show OpenAdapt a workflow in the app where you already do it.
             OpenAdapt repeats it, checks each saved record, and stops to ask you
             when something doesn't match.
           </p>
           <div className="choice-row">
             <ChoiceButton
               tone="primary"
-              title="Show OpenAdapt your task"
+              title="Show OpenAdapt your workflow"
               sub="Do it once on test data while OpenAdapt watches."
               onClick={actions.onShow}
             />
diff --git a/src/screens/WatchRun.test.tsx b/src/screens/WatchRun.test.tsx
index 6ebfdba..5114ac6 100644
--- a/src/screens/WatchRun.test.tsx
+++ b/src/screens/WatchRun.test.tsx
@@ -304,7 +304,7 @@ it("requires the first user to review the compiled workflow before a supervised
 
   fireEvent.click(
     screen.getByRole("checkbox", {
-      name: "I reviewed these steps. The task uses test data, and I will keep the target app in view.",
+      name: "I reviewed these steps. The workflow uses test data, and I will keep the target app in view.",
     }),
   );
   expect(runButton.disabled).toBe(false);
diff --git a/src/screens/WatchRun.tsx b/src/screens/WatchRun.tsx
index 79c5de8..00926aa 100644
--- a/src/screens/WatchRun.tsx
+++ b/src/screens/WatchRun.tsx
@@ -478,7 +478,7 @@ export function WatchRun({
                     }
                   />
                   <span>
-                    I reviewed these steps. The task uses test data, and I will
+                    I reviewed these steps. The workflow uses test data, and I will
                     keep the target app in view.
                   </span>
                 </label>
diff --git a/src/screens/Welcome.tsx b/src/screens/Welcome.tsx
index 1b598ac..61d5e4d 100644
--- a/src/screens/Welcome.tsx
+++ b/src/screens/Welcome.tsx
@@ -6,7 +6,7 @@ import { ChoiceButton } from "../ui/simple";
 const STEPS = [
   {
     title: "Show it",
-    body: "Do the task on test data, in the app where you already do it.",
+    body: "Do the workflow on test data, in the app where you already do it.",
   },
   {
     title: "Check it",
@@ -30,7 +30,7 @@ export function Welcome({
   return (
     <div className="content simple welcome">
       <section className="welcome-hero">
-        <h1>Hand off the last step of a task</h1>
+        <h1>Hand off the last step of a workflow</h1>
         <p className="welcome-lead">
           OpenAdapt enters approved information into the apps you already use,
           like EMRs, payer portals, and desktop or Citrix apps, and checks that
@@ -61,14 +61,14 @@ export function Welcome({
           onClick={onSample}
         />
         <ChoiceButton
-          title="Show OpenAdapt your task"
-          sub="Pick the app, then do the task while OpenAdapt watches."
+          title="Show OpenAdapt your workflow"
+          sub="Pick the app, then do the workflow while OpenAdapt watches."
           onClick={onShow}
         />
       </div>
 
       <p className="welcome-foot">
-        OpenAdapt doesn't send your recordings or tasks anywhere unless you
+        OpenAdapt doesn't send your recordings or workflows anywhere unless you
         choose to share or upload them. Backup, sync, and security software on
         this computer can still copy its files.{" "}
         <button type="button" className="link-button" onClick={onConnectCloud}>
diff --git a/src/styles/app.css b/src/styles/app.css
index c288324..0169d16 100644
--- a/src/styles/app.css
+++ b/src/styles/app.css
@@ -46,13 +46,14 @@ table {
   margin: 0;
 }
 
-/* Statements in the serif; everything a person operates in the sans. */
+/* Titles use --font-heading (the sans, one face across Desktop, Cloud and
+   the portal). */
 h1,
 h2 {
   font-family: var(--font-heading);
   letter-spacing: -0.012em;
   color: var(--text-strong);
-  font-weight: 430;
+  font-weight: var(--fw-heading);
   margin: 0;
   line-height: 1.15;
   text-wrap: balance;
@@ -848,7 +849,6 @@ a:hover {
   border: 1px solid var(--border);
   border-radius: var(--radius-2);
   padding: var(--space-5);
-  box-shadow: var(--shadow-sm);
 }
 .card + .card {
   margin-top: var(--space-4);
@@ -1650,7 +1650,7 @@ select.input:focus {
   margin-bottom: var(--space-6);
 }
 .simple-head h1 {
-  font-size: 32px;
+  font-size: 28px;
 }
 .simple-sub {
   margin: var(--space-2) 0 0;
@@ -1918,7 +1918,6 @@ select.input:focus {
   border: 1px solid var(--border);
   border-radius: var(--radius-2);
   background: var(--panel);
-  box-shadow: var(--shadow-sm);
 }
 .task-title-row {
   display: flex;
@@ -2442,13 +2441,13 @@ select.input:focus {
 .stat-warn .stat-label {
   color: var(--pill-warn-fg);
 }
-/* A measurement: the serif, with tabular figures. */
+/* A measurement, with tabular figures. */
 .stat-value {
   color: var(--text-strong);
   font-family: var(--font-heading);
   font-feature-settings: var(--num-tabular);
   font-size: 32px;
-  font-weight: 430;
+  font-weight: var(--fw-heading);
   letter-spacing: -0.015em;
   line-height: 1.05;
 }
diff --git a/src/styles/tokens.css b/src/styles/tokens.css
index 4d991c0..3de06e1 100644
--- a/src/styles/tokens.css
+++ b/src/styles/tokens.css
@@ -340,14 +340,16 @@
 
 /* -----------------------------------------------------------------------------
    4. TYPOGRAPHY
-      Two faces (Clinical Calm). The serif carries what a person reads as a
-      statement or a measurement: page titles, the question on a stop, big
-      numbers, compared values. The sans carries everything a person operates.
+      One face. The sans carries titles, numbers and everything a person
+      operates, the same as OpenAdapt Cloud and the phone decision portal.
+      --font-heading is the single switch: set it back to var(--font-serif)
+      (and --fw-heading to 430) to restore serif titles.
       Monospace is only for run IDs and code inside Technical details.
    -------------------------------------------------------------------------- */
 :root {
   --font-ui: var(--font-sans);
-  --font-heading: var(--font-serif);
+  --font-heading: var(--font-sans);
+  --fw-heading: 600;
 
   /* Type scale (px), tuned for dense product UI. Canonical carries a rem scale
      sized for editorial layout; this one is a product scale, which is a scale
diff --git a/src/ui/ApprovalSwitch.tsx b/src/ui/ApprovalSwitch.tsx
index 9cca8c6..9d83740 100644
--- a/src/ui/ApprovalSwitch.tsx
+++ b/src/ui/ApprovalSwitch.tsx
@@ -176,7 +176,7 @@ export function ApprovalSwitch({
       {confirming ? (
         <div className="switch-confirm" role="group" aria-label="Turn off the switch">
           <p>
-            Turn it off for this task? It applies before the next save, even
+            Turn it off for this workflow? It applies before the next save, even
             offline. Turning it on again needs an admin and the privacy contact
             in OpenAdapt Cloud, and new items approved one by one.
           </p>
diff --git a/src/ui/ItemApproval.tsx b/src/ui/ItemApproval.tsx
index 4bc5eb3..a5d3590 100644
--- a/src/ui/ItemApproval.tsx
+++ b/src/ui/ItemApproval.tsx
@@ -160,7 +160,7 @@ export function ItemApprovalPanel({
                 ))}
               </dl>
             ) : (
-              <p className="simple-sub">This task takes no values.</p>
+              <p className="simple-sub">This workflow takes no values.</p>
             )}
             <p className="simple-sub approval-step">
               Next step: {request.step_intent || "the save"}
diff --git a/src/ui/JudgmentCaseCapture.tsx b/src/ui/JudgmentCaseCapture.tsx
index 3c3439e..4f740f8 100644
--- a/src/ui/JudgmentCaseCapture.tsx
+++ b/src/ui/JudgmentCaseCapture.tsx
@@ -463,7 +463,7 @@ export function JudgmentCaseCapture({
         <Button variant="primary" disabled={busy} onClick={() => void capture()} data-testid="capture-judgment-case">
           {busy ? "Saving…" : "Save reviewed case locally"}
         </Button>
-        <span className="page-sub">Flow seals the case into the next qualification revision. This does not create a runtime task.</span>
+        <span className="page-sub">Flow seals the case into the next qualification revision. This does not start a run.</span>
       </div>
     </Card>
   );
diff --git a/src/ui/ShadowMode.tsx b/src/ui/ShadowMode.tsx
index e820e18..fee46e2 100644
--- a/src/ui/ShadowMode.tsx
+++ b/src/ui/ShadowMode.tsx
@@ -142,7 +142,7 @@ export function ShadowPromptBanner({
   }
 
   return (
-    <div className="shadow-prompts" aria-label="Repeated tasks">
+    <div className="shadow-prompts" aria-label="Repeated workflows">
       {status.pending.map((prompt) => (
         <Callout key={prompt.motif_id} title={prompt.title}>
           <p>{prompt.body}</p>
@@ -293,7 +293,7 @@ export function ShadowModeCard() {
     <Card>
       <CardHead
         title="Shadow mode"
-        sub="Off unless you turn it on. When on, OpenAdapt notices a task you repeat and offers to make a draft. It never runs anything. Its encrypted buffer on this computer is cleared at least every 24 hours, and when you turn it off or quit."
+        sub="Off unless you turn it on. When on, OpenAdapt notices a workflow you repeat and offers to make a draft. It never runs anything. Its encrypted buffer on this computer is cleared at least every 24 hours, and when you turn it off or quit."
       />
       {!status.known ? (
         <p className="page-sub">Checking shadow mode…</p>