Skip to content
MOTION ENGINE

OMNIS doesn't watch.
It runs.

Most warehouse software shows you dashboards. MOTION makes decisions. Sixteen rules across five domains — detention prediction, yard throughput, pick quality, battery-aware routing, and vision-correlated safety. Every dock-scoped action runs through a pre-execution safety gate. Confidence-gated. Policy A/B tested. Every action carries a receipt.

MOTION · LIVE 13 RULES ACTIVE
detention.risk_predictionEXEC 0.92
TRL-4821 projected 85m — escalate carrier
yard.priority_inversionEXEC 0.88
P7 move running while P1 waits — reorder
pick_quality.slow_pickerADVISE 0.76
Session 42 median 78s — fatigue likely
battery.route_to_chargerEXEC 0.91
FK-04 battery 14% — route to C2 after job
0
Active Rules
Four rule packs
0
Confidence Gate
Per-kind threshold
0
Policy A/B
Auto-compare window
ONE
Signal Bus
Cross-domain correlate

Five domains. Sixteen rules.

Every rule is a file. Every file is versioned, testable, and overridable.

DETENTION4 rules
Predict detention accrual before free time expires. Default free time 120m, warning lead 30m. Uses scan velocity + pallet-count expectations (26 per 53' trailer).
detention.risk_prediction
Forecast billable detention 30m before accrual
detention.stalled_unload
Scan velocity under 0.3/min with pallets left
YARD THROUGHPUT3 rules
Detect queue backlog, priority inversion, stale moves. Triggers when queue grows past 12, or P7 moves execute while P1-P3 wait more than 15 minutes.
yard.priority_inversion
Low-priority executing while high-priority waits
yard.queue_backlog
Backlog over 12 moves triggers re-sequence
PICK QUALITY3 rules
Catch errors before customer escalation. Flags slow pickers (median 60s+), high short-rate sessions (10%+), and AI-fallback dominance (means marker is missing or obscured).
pick_quality.slow_picker
30%+ slow picks in session flags fatigue
pick_quality.ai_fallback_dominance
40%+ AI-fallback means marker issue
BATTERY ROUTING3 rules
Keep assets charged without blocking jobs. Predicts battery runout from current SOC and job length. Routes to opportunity-charge stations during idle windows.
battery.route_to_charger
SOC under 20% + no job → charger route
battery.opportunity_charge
Idle 10m+ → top-up at nearest station
CONFIDENCE GATING

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
DECISION LEDGER · 09:41 – 12:00
Jockey swap Y3→D7EXEC 0.92
Picker reassign A12EXEC 0.88
Late ASN alertADVISE 0.76
Detention risk D6HOLD 0.71
Battery route FK-04EXEC 0.91
Override: swap revertedOPERATOR
POLICY A/B · week 26
POLICY A · 80%
4.2%
detention rate
97.1%
pick accuracy
POLICY B · 20%
2.9%
detention rate
98.4%
pick accuracy
B wins both KPIs over 7 days. Auto-promote recommended — awaiting operator confirm.
POLICY A/B

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
ONE SIGNAL BUS

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
CROSS-DOMAIN CORRELATION
dispatch.asn_late correlates with
yard.jockey_idle + wms.wave_stalled
detention.risk_prediction fires at 0.94

Let MOTION run it.

Fifteen minute demo. We'll run sixteen rules against your live warehouse and show you the decision ledger.