Weekly Demand Forecast
HealthyReal-time production health
Known immediately — no outcomes needed.
Predictions / hour
Latency
Inputs, context & predictions
Input drift vs reference (PSI)
Prediction distribution
Routing
- Backend
- Managed by Zero Grad
- Context
- pinned to v4
- Fallback on timeout
- Managed C · 0.02% of requests
Matured outcome quality
Performance metrics are computed only for predictions with observed outcomes. Realized weekly sales take up to ~9 days to arrive, so recent predictions aren't scored yet.
Business impact
Aug 3 – Sep 20 · 100% coverageInventory effects estimated against the previous planning process over 8 backtest windows.
Model quality
Aug 3 – Sep 20WAPE — matured outcomes
By prediction week. Recent cohorts are provisional until their outcomes arrive.
Outcomes
Ground truth flows back by prediction_id. It improves evaluation and routing — never model weights.
Send outcomes
Every prediction returns a prediction_id. Send the real outcome whenever you learn it.
# Later, when the real outcome is known
zg.capability("weekly-demand").send_outcomes([
{"prediction_id": "pred_01J9X4S9VD", "outcome": {"units_sold": 418},
"observed_at": "2026-09-30T23:41:00Z"},
])Outcomes never update model weights — Zero Grad does not train a customer-specific model. They are used to measure real production quality, compare foundation models and context-selection strategies, improve routing, and decide whether a challenger beats the current production configuration.
What Zero Grad learned from your outcomes
No training doesn't mean no learning.
- Routing: intermittent series sent to the statistical baseline cut their realized WAPE from 31% to 24% (v5).
- Covariates: promo_flag improves promo-week WAPE by 3.1 pts; price added no lift and was dropped from context.
- Late sales from 2 stores are excluded from evaluation until they arrive.
What adapts: which foundation model serves which slice, how in-context examples are selected, and calibration. What doesn't: model weights.