Five domains. Sixteen rules.
Every rule is a file. Every file is versioned, testable, and overridable.
Acts when sure. Advises when not.
Every decision carries a confidence score. High-impact actions (jockey swap, picker reassign) require 0.90. Advisory actions publish at any confidence. You set the thresholds; MOTION never executes unsafely.
- Per-kind thresholds: jockey_swap 0.90, picker_reassign 0.90, scan_reroute 0.85
- Advisory kinds (pre_shift_alert) publish at any confidence
- Full audit trail — decision, outcome, operator override
- Kill-switch per rule, per pack, or global — one click
Test two policies. Keep the winner.
Activate policy B on 20% of tenants or yards. Seven days later, MOTION compares detention reduction, pick accuracy, and yard throughput. Auto-roll to the winner, or hold and review. No more "I think this rule works better" — the numbers decide.
- Per-tenant or per-yard activation splits
- 7-day comparison window with statistical significance
- Cross-domain correlation via the ONE signal bus
- Async rule engine — decisions never block the request path
Rules that talk to each other.
A late ASN signal from dispatch isn't a dispatch problem. It's a yard problem (jockey idle), a pick problem (wave stalls), and a detention problem (free time ticking). ONE's signal bus correlates signals across domains so rules can trigger each other.
- Pub-sub across 9 domains (dock, yard, driver, gate, inventory, one, sred, twin, wms)
- Cross-domain correlation windows (default 120s)
- Signal provenance — every rule firing is traceable to its triggers
- Event deduplication + replay protection built-in
Let MOTION run it.
Fifteen minute demo. We'll run sixteen rules against your live warehouse and show you the decision ledger.