--- a/code/econ_scenarios/production.py
+++ b/code/econ_scenarios/production.py
@@ -114,14 +114,27 @@
     return Potential(dlnr, lnSL, lN, lnW, lnYL, lnK, math.nan)
 
 
+# Structural-sensitivity pilot (The Unjournal, 29 Oct 2026): partial adjustment of the capital
+# stock toward the paper's schedule. Each month the capital supplied is a weighted average of last
+# month's stock and the schedule's level, (1 - lam) K_prev + lam e^{eps dlnr}, both as levels
+# relative to the no-AI path. lam = 1 is the paper exactly; lam = 0 freezes the stock at its no-AI
+# path. simulate() sets CAPITAL_ADJUSTMENT from its half-life argument and updates K_PREV.
+CAPITAL_ADJUSTMENT = 1.0
+K_PREV = [1.0]
+
+
 def capital_supply(dlnr: float, assets: float, cal: Calibration) -> float:
     """Log gap of the capital supplied at rental gap ``dlnr``: the paper's schedule e^{eps dlnr}
     (Eq. 18) plus ``assets``, workers' extra assets over the no-AI capital stock, which is zero in
     the paper. At eps = inf the rental rate is pegged and the supply is whatever is demanded. Where
     negative extra assets exceed the schedule, at a rental rate far below the no-AI one, nothing
     is supplied and the log gap is minus infinity; the market clears at a higher rate.
+
+    Pilot variant: the schedule's level enters with weight CAPITAL_ADJUSTMENT and last month's
+    stock with the rest (see the comment above); at weight 1 this is the paper's line unchanged.
     """
-    level = math.exp(cal.eps * dlnr) + assets
+    lam = CAPITAL_ADJUSTMENT
+    level = (1.0 - lam) * K_PREV[0] + lam * math.exp(cal.eps * dlnr) + assets
     return math.log(level) if level > 0.0 else -math.inf
 
 
--- a/code/econ_scenarios/simulate.py
+++ b/code/econ_scenarios/simulate.py
@@ -19,6 +19,7 @@
 from .calibration import Calibration, Scenario, rate_from_fraction
 from .labor import SteadyState, hires, steady_state
 from .paths import AIState, ScenarioPaths
+from . import production
 from .production import actual_economy, demand_given_wage, potential, shift_N
 
 if TYPE_CHECKING:
@@ -298,6 +299,17 @@
     return ell_next, U_next
 
 
+def capital_adjustment(half_life_months: float) -> float:
+    """Monthly weight on the capital-supply schedule for a given half-life of the gap between the
+    stock and the schedule: 0 months is the paper (weight 1), infinity freezes the stock (weight 0).
+    """
+    if half_life_months <= 0.0:
+        return 1.0
+    if math.isinf(half_life_months):
+        return 0.0
+    return 1.0 - 0.5 ** (1.0 / half_life_months)
+
+
 def simulate(
     scenario: Scenario,
     cal: Calibration | None = None,
@@ -307,7 +319,33 @@
     | Callable[[int, dict[str, list[float]], Callable[[float], float]], float]
     | None = None,
     freeze_after: float | None = None,
+    capital_half_life: float = 0.0,
 ) -> Simulation:
+    """The paper's monthly system, with the pilot's capital-adjustment switch.
+
+    ``capital_half_life`` (months) sets how fast the capital stock closes its gap to the paper's
+    supply schedule; 0, the default, is the paper's instantaneous schedule. See ``_simulate``.
+    """
+    saved = production.CAPITAL_ADJUSTMENT
+    production.CAPITAL_ADJUSTMENT = capital_adjustment(capital_half_life)
+    production.K_PREV[0] = 1.0
+    try:
+        return _simulate(scenario, cal, horizon, level_form, worker_assets, freeze_after)
+    finally:
+        production.CAPITAL_ADJUSTMENT = saved
+        production.K_PREV[0] = 1.0
+
+
+def _simulate(
+    scenario: Scenario,
+    cal: Calibration | None = None,
+    horizon: float = 2030.0,
+    level_form: str = "exact",
+    worker_assets: np.ndarray
+    | Callable[[int, dict[str, list[float]], Callable[[float], float]], float]
+    | None = None,
+    freeze_after: float | None = None,
+) -> Simulation:
     """Run the monthly system of Table A.1 from t0 through ``horizon``.
 
     ``level_form="first_order"`` uses the first-order rows of panel B for the reported potential
@@ -384,6 +422,7 @@
         S_C, S_N, H_C, H_N, f = _matching(ss, cal, mu, U, v_C, v_N)
         # Step 8: the actual economy at realized employment.
         act = actual_economy(x, dlnA, (ell[0], ell[1]), ss.ell0, cal, assets)
+        production.K_PREV[0] = math.exp(act.lnK)  # pilot: next month adjusts from this stock
         lnW_avg = math.log(
             (math.exp(lnW_C) * ell[0] + math.exp(act.lnW_N) * ell[1])
             / (ell[0] + ell[1]),