PREVIEW · static UI mock · screen 17 / 32 · workflow eval cases · regression suite · all screens
‹ Workflows

Demand prediction

● Activev23

Owner: Maria Chen · Supply Chain VP · 128 eval cases · suite owned by your enterprise · exportable · open agent UI →

Total cases
128
+38 in last 30d
From clusters
38
auto-promoted from production failures
Last full-suite run
128 / 128
2h ago · 4m 12s · all green
Best score ever
91.2%
composite metric · gate floor
Type Severity Source Showing 12 of 128 · Run suite now
#
Case
Type
Severity
Source
Last pass
128
No auto-discount above 25% without reviewer (rejection precedent) Promoted from rejected proposal #ce-2241 · M. Chen comment
Regression
Med
reviewer comment · 3d
1/1
127
PNW winter boot 2025-W47 · markdown earlier than W49 From cluster #ce-2284 (14 traces · auto-promoted)
From cluster
High
cluster auto · 11d
1/1
126
Northeast winter coat 2026-W44 · weather-driven anomaly From cluster #ce-2284 (cousin trace · auto-promoted)
From cluster
Med
cluster auto · 11d
1/1
115-118
Promo bundle uplift Q4 2024-2025 (4 historical replays) From cluster #ce-2278 · past misses
Past miss
Med
cluster auto · 6d
4/4
98
Steady-demand baseline · no anomaly · regression check Hand-picked · ensure agent does not fire false markdown alerts
Regression
Low
manual · 6w
1/1
97
Inventory-pegged forecast · low-stock SKU Hand-picked · forecast shall not exceed inventory under any signal
Regression
Low
manual · 6w
1/1
82-85
Supplier price-build edge cases · 4 lead-time variants From cluster #ce-2218 · resolved 9d ago
From cluster
Med
cluster auto · 9d
4/4
76
Cold snap surge · winter coats Northeast 2024-W44 Generated edge case · narrow-window weather-driven anomaly
Edge
Med
generated · 3w
1/1
71
Regional variance · Southeast hurricane evacuation Generated edge case · localized demand displacement
Edge
High
generated · 3w
1/1
1-12
Initial generated suite · 12 representative scenarios From workflow generation · seeded baseline
Mixed
Mixed
workflow create · 6w
12/12
The eval suite is the moat that compounds. Every production failure becomes a permanent regression test. The agent at day 365 is measurably better than day 1, with a paper trail. Owned by your enterprise · exportable as JSON / CSV · never shared across customers.